What's New

We rewrote our Terms of Service, Privacy Policy, and Refunds & Billing policy in plain language, and brought every other policy page in line with them, from the Cookie Notice to the Sub-Processors list. They now name the company you are dealing with, spell out one clear refund rule, confirm that your credits roll over for as long as you are subscribed, and describe exactly how reports and appeals are handled.

The next time you open Pixel Dojo you will be asked to read and accept the updated Terms once. It takes one click, and nothing about your account, your credits, or your saved work changes.

Face Swap has returned to Characters & Faces. Pick a target photo and the face you want in it, choose face-only or full head, and the swap keeps the original pose, lighting, and background. One credit per swap.

Both photos in a face swap now pass the same safety check as every other uploaded image. Before your first swap, the tool asks you to confirm once that everyone in your photos is an adult who agreed to be there.

Sending a photo to a tool with Use in from My Media no longer refills the other image slot from an old draft. Only the photo you picked lands in the composer.

The Creator Agreement replaces the old content toggle in Settings. It links to the Terms of Service and says plainly what you confirm: you are 18 or older, you hold the rights to everything you upload and have consent from anyone shown, and automated abuse checks can lead to suspension or termination. Everyone accepts it once, even if the old toggle was already on.

Happy Horse video clips start at 3 seconds now. The 2 second option always failed after charging and refunding your credits, so it is gone.

WAN 3.0 now checks your reference image size up front. Images smaller than 240x240 pixels get a clear message naming your image's actual size, before any credits move.

WAN 2.7 and WAN 2.6 Image Edit now accept small reference images. Anything under 240 pixels on a side is enlarged for you before the edit runs, so a tiny character portrait no longer fails every time and hands your credit back.

Z-Image Turbo shows a plain message when a generation takes too long, instead of a raw error code.

The MiniMax H3 prompting guide now covers H3 Max: the same prompt run on both models side by side, what a start frame does to the bill on each, and a plain rule for when to pick which.

MiniMax H3 Max is here. It makes 5 to 15 second clips with native synced audio at a flat 2 credits per second, whether you pick 768p or 480p. A 5 second clip is just 10 credits.

H3 Max animates a start frame or an end frame and keeps your resolution choice either way. Find it in the MiniMax H3 tool, in Canvas, and in the API.

Flux 2 Flex no longer lets a run fail when the aspect ratio is set to match your input image but no image is attached. The option now only appears once an image is added, and the form catches the mismatch before any credits move.

Flux 3 Video now checks your source clip's length before starting a video extension. Clips over the 15 second limit get a clear message up front instead of a failed run.

When an image model returns nothing at all, you now see an explanation that its safety system likely declined the request, with your credits refunded, instead of a confusing technical error.

WAN 3.0 and the QWEN image tools were rejecting ordinary prompts after a filtering change earlier this week made them far too strict. That over-blocking is gone. Everyday scenes, people, and stylized content generate normally again.

Seedance reference images of real people briefly failed to upload. Uploads work again.

Seedance 2.5 no longer fails when your prompt asks it to extend or edit a video. It used to reject the request and hand your credits back. Now it just runs.

WAN 3.0 counts a reference video against the same 30 second budget as the clip it makes. Auto length is no longer offered once you attach one, and if your clip would not fit, it now tells you the longest length you can pick before you spend anything.

When a model blocks a generation with its own safety filter, you now get a plain explanation of what happened. Before, some tools showed a generic "provider reported the job failed" message that made a content block look like an outage. Credits are refunded the same way.

WAN 2.2 Video has two image to video modes again. Pick Image to Video LoRA for quality tiers, LoRA models, an optional end frame, 1080p and 10 second clips.

A new Trust & Safety page explains how we keep the platform safe and gathers every policy in one place. Find it in the footer or at /trust.

Report a problem directly from the site. The new report form on the Trust & Safety page goes straight to our safety team. No email needed.

The new Answers library at /answers collects hundreds of short guides with real measured results for picking models and settings.

Ideogram 4 now has three clear tiers: Turbo (1 credit), Balanced (2 credits), and Quality (3 credits). The old Fast and Instant tiers are retired — anything still requesting them runs as Turbo automatically.

WAN Image Edit no longer fails on very large images. Feeding a 4K generation back in for editing used to error out after a long wait — big images are now optimized automatically before editing starts.

WAN 2.7 Video now checks your driving audio's length the moment you upload it, so you find out right away if a clip is too long instead of after a failed generation.

The Help & Support chat now works properly on phones. The message box used to hide behind the browser toolbar and keyboard so you couldn't type — it now stays visible right above the keyboard.

Reve 2.1 has been retired after its upstream model became unavailable. Everything you made with it is safe in your library, and every generation that failed was refunded automatically. For multi-image editing and remixing, try Nano Banana or Qwen 3 Pro.

AI assistant and API integrations now respect your model pick for video jobs. If you name a video tool and hand it a clip, it runs on that exact tool instead of quietly switching to a different, pricier model.

The assistant upscale skill now handles video. Send a clip with its length in seconds and it comes back sharpened, up to 4K.

Video editing in Canvas now takes reference images. Kling Video Edit, WAN Video Edit, Seedance Video Edit 2.0 and 2.5, and HappyHorse Video Edit all accept them right in the prompt bar, and Seedance 2.5 takes reference audio too.

LoRA training is back. The Z-Image, WAN 2.2 Image, and Flux 2 trainers have returned, now running on our own dedicated training hardware. They are back in the trainer menus and in Character Studio.

Fast is now the default speed on WAN 3.0. New videos finish in about 2 minutes instead of 15 to 20. Standard is still in the speed menu if you prefer the lower price.

The WAN 3.0 prompt box is tidier. The reference limits now sit beside the add buttons instead of taking up their own lines.

LTX-2 video has been retired. In-progress jobs still finish and settle normally. WAN 3.0 is the recommended replacement for new videos.

The music video export panel is much simpler. Your song and the export button, and that is it. Sync controls only appear if your shots drift out of step with the song.

Shots with singing no longer get silenced by mistake. The planner only marks a shot as instrumental when your lyrics prove it, like an intro or a tagged break.

Swapping in a new song now recuts it to your shots automatically.

Big renders and upscales can be used as inputs again. A 4K image or a 4x upscale used to be rejected as too large when you fed it back in as a reference. It is now prepared for you automatically, at full resolution.

When a model is briefly unavailable, you get a plain message saying so and that your credits were refunded, instead of a technical error.

On phones, the film settings panel stays on screen and the add-to-cast window scrolls, so every button is reachable again.

Bigger touch targets across the storyboard, and delete buttons now show on tap instead of hiding behind a hover.

Music videos mark a shot as instrumental only when your lyrics actually show a break, so performers stop going silent over sung sections.

Nano Banana, Gemini Image, Grok Image Edit, MiniMax H3 and Omni Flash sometimes failed with a confusing technical message during a brief provider hiccup. Those now retry on their own and usually go through.

If one of those does fail, you get a plain message that the model is briefly unavailable and your credits were refunded.

When a generation fails, you now see the real reason more often — validation tips, model blocks, and provider messages — instead of a generic failure toast.

Canvas only says your credits were refunded after the refund is confirmed, including if you refresh while a job is settling.

Seedance 2 checks start and end frames up front so out-of-range images fail at pick time, not after you spend credits.

The Help & Support panel no longer steals clicks when it is closed, and reporting a bug from Help no longer fights the drawer for Escape or sits under the dimmed overlay.

Fast mode on WAN 3.0. Same model, same quality, finished in about 2 minutes instead of 15 to 20.

The one video at a time limit is gone. Run as many WAN 3.0 generations as you like, up to 30 seconds each with spoken dialogue in the same file.

Film Studio can now render clips with WAN 3.0, and WAN 3.0 is live on the public API for developers.

Film Studio has a Clip quality setting. Pick Draft to render every clip at a lower resolution for up to 55% fewer credits, then switch back to Standard for the final takes. It lives in the film's Settings panel and only shows up on engines that have a cheaper tier.

MiniMax H3 now offers a 768p output at 2 credits per second, down from 3 at 2K. Pick it from the Resolution chip on the tool page, in Canvas, or with the new resolution option in the API. Start-frame videos stay at 2K, since the model sets the size from your frame.

The Recent generations panel now loads fast even after a heavy session with hundreds of images. Thumbnails load as you scroll instead of all at once, so your recents show up right away.

Tool Generate buttons no longer flash Subscribe for paying members while the page loads — and unpaid Subscribe is clickable again.

Prompt Enhance now uses tool-specific rewriting for more models (including Flux Studio and Ideogram), pulls fuller saved-prompt examples, and stays honest when Enhance is busy — your original prompt is left unchanged.

Enhanced prompts no longer arrive wrapped in leftover markdown code fences.

Signed-in subscribers see Dashboard in the marketing nav instead of Subscribe. Signed-out visitors still get Sign Up.

Plan copy now matches reality: every paid plan includes the same tools and the same generation speed — tiers differ by credit volume only (no priority queue).

Marketing counts now use the live tool total from the sidebar, so badges and headlines stay accurate as new tools ship.

The Seedream 5 Pro/Lite picker no longer lists credit prices. Cost still shows on Generate so you always see the rate for your current settings.

FLUX Video Upscale is in Make it sharper. Sharpen clips up to 20 seconds to 1080p, 2K, or 4K with Precise mode for faces and products, or Creative mode for richer detail — original audio stays intact.

WAN video pricing is now one simple rate across the family. WAN 2.7 at 1080p is 4.5 credits per second, the same as WAN 3.0. 720p stays 2.5 credits per second.

Seedream 5 Pro pricing now follows output size cleanly: 1.25 credits for 1K images and 2.5 credits for 2K images. Seedream 5 Lite stays 1 credit.

Creating a new account works again everywhere. If you or a friend hit an error page while trying to sign up over the last couple of days, that door is fixed and wide open.

Signing out of one account when you use several no longer lands on a broken page — you go straight to sign-in so you can switch accounts smoothly.

The Pixel Dojo agent now ranks models the same way the rest of the site does — by what creators actually use — so its first pick is more likely to be the tool you wanted.

Specialized skills like campaigns, avatars, and effects only show up when your conversation calls for them. That keeps the agent focused instead of offering every tool at once.

Before a generation runs, the agent checks you have enough credits and tells you upfront. No more starting a job only to hit an insufficient-credits error mid-flow.

Oversized reference images are resized automatically before send, and the agent stops retrying the same image when a safety check blocks it — fewer mysterious repeat failures.

Long films and music videos now show a proper thumbnail in My Library and the community gallery. Anything past a few minutes came back as a blank black tile, so a finished film was the hardest thing to find in your own library. Every film already exported has been fixed too.

The same blank black tile showed up on your dashboard Recent Creations, leaderboards, contests, and a few other spots that preview saved videos. Those now use the stored thumbnail too, so long exports look right everywhere you browse your work.

WAN 3.0 now actually uses the reference images you attach. Unless your prompt named each image, they were being ignored, so the video came back as if you had never added them. Your references are named for you now, and the clip follows them.

Tap a reference thumbnail and it drops the right name into your prompt. Each image is labelled Image 1, Image 2 and so on, so you can point at one directly. Try "the woman in Image 1 walks into the forest in Image 2".

Generate all frames no longer skips shots. Some frames quietly never started, so you had to click again and again to work through a film. They now retry on their own, and one click takes the whole storyboard.

Frames render in a steady stream instead of batches. A slow frame no longer holds up the ones behind it, so a full storyboard finishes noticeably faster.

Planning a longer film or music video no longer fails partway through. A two or three minute song used to spend five minutes planning and then come back as an error. Planning now runs in the background and finishes, however long it needs.

Every planner thinks harder before it writes. Scripts, shot plans, and revisions now take more time to work out pacing, wardrobe, and continuity. Longer to plan, better to render, and planning is still free.

You can see what planning is doing. A live panel shows how long it has been working, whether it is still thinking or writing your shots out, and how many shots have landed so far.

Planning keeps going if you leave. Close the tab, reload, or come back later and your film is still being planned, with the storyboard filled in when you return.

Tighten cuts at export. One toggle trims a beat off the start and end of every clip, so your film cuts faster and feels more alive. Your clips are untouched, and exporting again with it off restores the original cut.

Camera moves now come with something to film. The script writer pairs every push-in, pan, and dolly with an action happening during the move, so moving shots read as cinema instead of a drifting photo.

Some members could not sign in after our August 10 update. Google sign-in failed with an error, and email codes never arrived. The cause is fixed. If you are still stuck, sign in once from a private browser window and everything resets.

Seedance 2.5 costs less at every size. A 5 second clip at 720p is 25 credits instead of 30, and 30 seconds at 1080p is 375 credits instead of 450. Video Edit and Extend came down too. Same model, same quality, lower price.

Seedance 2.5 renders at 1080p. Pick it under resolution on the tool page or in Canvas, for both new videos and edits.

Reference to video moved into the main video tools. Grok Video, Happy Horse, and Kling Video each have a Reference mode built in, so one page covers every way to start. Your prompt and settings carry over when you switch, and old links still work.

WAN 3.0 Video is open in beta. The newest WAN generation is live on its tool page, one generation at a time while we scale it up. Canvas and API access come with the full launch.

No more blank error screens. In the rare case something breaks, you now get a real error page with a Try again button and details you can send to support. A rare bug that could blank the whole site until you cleared browser data is gone for good.

On WAN 3.0 you can tap a reference image to drop it into your prompt, so you can say which reference you mean without typing it out.

Qwen video tools now tell you plainly when a clip is blocked because the photo shows a real person, so you know what to change instead of guessing.

The Jump back in section on the dashboard sits properly on phones again.

The dashboard stops jumping while it loads. The top strip and the cards below it now land in place on the first paint, so nothing slides out from under your cursor.

Qwen 3 Pro edits stopped failing partway through. A handful of edits were dropping out with a download error even though nothing was wrong with your image. They now retry on their own and go through.

No more being told you hit a rate limit when you did not. Busy periods used to show a quota message and send you off to wait for nothing. You now get the real reason, and a slow failure gives up quickly instead of leaving you watching a spinner.

Character voice samples stay put. A voice clip you added to a character could go quiet a day later and leave a player that would not play. New voices are stored to keep, and the handful of characters whose clips were already lost now simply show no voice, so you can add a fresh one.

A long voice sample on a character no longer breaks every shot that character appears in. Voice clips need to be 2 to 30 seconds, and we check that when you add one instead of letting the shot fail later. If a character already has a longer clip saved, the message now tells you to trim it.

Characters stop teleporting between cuts. Every scene now carries its own map of who stands where and which side the camera shoots from, and every shot in the scene follows it.

Every character gets a written voice: age, tone, pace, and accent, kept word for word across the whole film so they sound like the same person in every scene.

Period lock. Set an era like 1985 London and every frame and clip stays in it. No stray smartphones in your western.

Scripts got livelier. Characters actually talk now, with quick, characterful exchanges, scenes cut faster, and every genre from comedy to horror writes with its own storytelling craft built in.

Music videos are now directed by genre. A rap video, a K-pop video, and a dream-pop video each get their own visual grammar, from beat-matched cuts to formation choreography.

Film controls moved to where they belong. Click any cast member on the storyboard to edit their outfit and voice, set the era right from the film header, and Settings keeps just the machinery.

LTX-2 Video now runs LTX-2.5, a much stronger version of the model. Sharper faces and fine detail, better motion, and synced audio on every clip.

LTX-2 Video costs a lot less. A six second 1080p clip is now 2 credits, down from 12. Ten seconds is 4 credits, down from 20. Twenty seconds is 10 credits, down from 40. 1440p and 4K dropped just as hard.

It is also a lot quicker. Longer clips saw the biggest jump. A ten second clip that took around twelve minutes now renders in about one.

The Fast and Pro choice is gone. LTX-2.5 has one quality setting and it is the good one, so there is nothing left to pick between.

Extend Video has been retired from LTX. The new version does not support it yet. Merge Videos and Film Studio are the way to build longer pieces for now.

Film Studio exports were failing with a gateway error page instead of starting. Merge Videos and the Video Analyzer were hitting the same wall. All three are back. If an export ever cannot start now, you get a plain message saying so, and nothing is charged.

Starting an export is fast again. It no longer re-checks clips against your song before it begins, because Film Studio already checks each clip as it finishes rendering.

Music videos now plan tight, close shots of whoever is on camera. Faces and hands look far better at these clip lengths than wide or full body framing does. Wide shots are still used for scenery and b-roll.

Grok Image has a new Imagine 2 tier. Pick it from the quality menu to generate with the newest Grok model at 1K or 2K, for 2 credits an image.

Imagine 2 does edits too. Attach an image, describe the change, and the edit comes back at the same size as your original.

Seedream edits sometimes stopped with a message about not being able to load your reference image, even when the image was perfectly fine. Those hiccups are now retried behind the scenes, so the edit just runs. Credits were always returned when this happened, and still are if an edit truly cannot finish.

Film Studio frames and clips now have a Save step. Render as many versions of a shot as you like, then Save the one you want — it is kept for good. There is a Save on every card, plus Save all frames and Save all clips in the toolbar when you want to keep a whole pass at once.

Anything you do not save is cleared after 24 hours. Each card tells you how long an unsaved frame or clip has left, so nothing disappears without warning — and if one does expire, the shot says so and offers to render it again.

Saved clips land in your film's Library folder the moment you save them, so the folder holds the takes you chose instead of every attempt. Exporting still needs its clips saved first, so a finished film always plays from files that are here to stay.

Rodin 3D models now preview and download correctly. Finished models were being saved in the wrong format, so the 3D preview did not load and the download gave you an image file instead of your model. You now get a spinning preview of every model and a real GLB file to download.

Each Rodin job returns more than one model, so every download is now labeled by type. No more guessing which of the identical buttons to press.

Music Video Mode in Film Studio. Upload a song, describe your artist and the vibe, paste lyrics if you have them, and get a full storyboard planned to your track. Every shot is generated performing its exact part of the song, and the final export plays your original audio start to finish with no gaps.

Your song is cut to your shots automatically. Music videos live on fast cuts, so each shot gets the precise slice of the track it covers. Re-time or add shots and one click re-syncs everything.

Nine looks made for music videos: K-pop, Rap video, Neon club, Stage performance, Retro camcorder, Y2K pop, Moody R&B, Rock grit, and Dream pop.

Lip-sync you can trust. MiniMax H3 leads music videos by default because it performs your track with frame-accurate timing, and every generated shot is checked against the song. If one lands off the beat, we tell you before you export.

Film Studio exports now retry automatically when a temporary hiccup interrupts the render, and new films start in 16:9 Cinematic.

All Tools and My Stuff are now one menu. Open it and you land on For You: pick up where you left off, jump to the tools you use most, and see what is popular across PixelDojo this week.

Every recommendation shows its receipt. You used this 82 times this month. 56,000 uses this week. New this week. If we suggest a tool, you can see exactly why.

Categories are named for what you want to do, like Make a video or Make it sharper. Models inside each category are ordered by real usage from the last two weeks and stay fresh automatically.

Brand new to PixelDojo? The menu starts by asking what you want to make and points you to a great first tool for it.

The dashboard now greets you with Jump back in: one-click returns to the tools you just used and the ones you use most. New creators get the same what-do-you-want-to-make cards right on the dashboard.

Videos that finished after a delay sometimes landed in your library labeled as images. They now arrive as videos every time.

A few Seedance 2.5 video shots showed as failed and asked you to retry even though the video had finished rendering. Finished shots now come back every time.

Clicking a creator from the leaderboards or a shared profile link sometimes showed an empty gallery even though they had public work. Creator pages now find the right person every time, even if the link capitalizes their name differently.

The Top Creators leaderboard loads noticeably faster.

Community gallery pages now show the date and time each image or video was created, right alongside the model and prompt.

Your Library media details now include a creation timestamp too, so you can tell exactly when you generated something.

Saving your profile settings could fail with an error even when everything looked fine. Profile saves now go through reliably.

The safety check that keeps photos of children out of generation tools now runs on a much more accurate age model, and it answers in a fraction of a second. It also covers every upload surface, including Canvas, inpainting, and outpaint.

Film Studio soundtracks no longer disappear from the final export. If you attached music to your film, it now reliably makes it into the finished video.

Profile photos taken on iPhone now upload correctly. Saving your profile with an iPhone photo used to fail with a generic error.

Your website and social links now show up on your public profile and stay filled in when you reopen settings. They were saving but never displaying.

If a profile save fails, you now see the actual reason, like a username being taken, instead of a generic error message.

Some users could not save their profile at all because another account had a similar username. Saving your profile now works again for everyone.

Flux 3 Video is now a Film Studio engine. Your approved storyboard frame becomes the exact first frame of the clip, and shots can run up to 20 seconds, the longest of any engine.

MiniMax H3 References joins the engine picker. It sends your storyboard frame and up to 8 character images along as identity guides, for shots where keeping faces consistent matters more than an exact starting frame.

Film Studio now uses Seedance 2.5 by default. Shots can run up to 30 seconds in one take instead of 15, with the same storyboard frame, character references, voice references, and spoken dialogue. You can still pick any other engine from the shot settings.

Videos in your history now recover from network hiccups on their own. If a clip fails to load, we quietly retry it a couple of times before saying anything. Most of the time you will never notice.

Video error messages now tell the truth. A clip that hits a loading problem used to be declared gone even when it was perfectly safe on our servers. Now you get an honest message, a Retry button, and Download keeps working the whole time.

When a history clip really has passed its 24 hour window, the card now says exactly that, so you know to save or download the clips you want to keep.

When a generation fails, the card now tells you your credits were refunded. No more wondering whether you were charged for something that never finished. This covers every image and video tool on the site.

Long video histories are much lighter on phones. Only the clips near your screen load now, and clips quietly unload as you scroll past them, so a full day of videos no longer overwhelms the player on mobile.

If a finished video ever fails to reach our storage on the first try, we now automatically try again the next time you open your history, so results stay retrievable for their full 24 hours.

Videos in your history play again. For part of today, a video could show its opening frame with a play button that did nothing, which looked like the clip had turned back into the picture you started from. Nothing was ever lost. The videos were sitting on our servers the whole time, and they play normally now. This covers every video tool on the site.

If something goes wrong the moment credits are deducted, we can now tell for certain whether you were actually charged. Before, that was a guess. Every run now leaves a receipt, so support can answer the question instead of estimating.

For API users: webhook deliveries now check the destination address at the moment they connect, not just when you set them up.

Refunds for batch image runs now match exactly what you paid. If you bought during a sale, refund amounts used to be figured from the regular price instead of your discounted price.

Model choices in Canvas settings now show readable names everywhere. A few tools listed internal version codes in their dropdowns instead of plain names.

A single problem email can no longer hold up everyone else's. The email queue now sets a bad message aside and keeps going.

For API users: asking about a job ID that never existed now says Job not found instead of claiming your results expired, and the model catalog responds faster.

Saving an image now tells you the truth. If a save failed behind the scenes, the button used to flip to Saved anyway, and the image could quietly expire a day later. Now a failed save keeps the Save button active so you can try again, and Saved means saved. This covers every image tool on the site.

If you ask for several images at once and some of them never start, the credits for the ones that never started now come back to you automatically.

Slow model trainings no longer get canceled while they are still working. A training that finished late used to be marked as failed even though your model was ready. Finished trainings are now always delivered, and we wait much longer before calling one stuck.

Canvas no longer removes a finished step when its preview briefly fails to load. A momentary network hiccup used to permanently delete work you had paid for. Now we double-check the file is really gone before removing anything.

Deleted videos stay deleted. On a couple of video tools, a result you removed from your history could reappear a day later with a note saying it had been recovered.

Scrolling the community gallery with the Models filter on no longer shows the same card twice or skips cards, and like counts stay accurate even on a fast double-click.

Webhooks on the public API now deliver. Set webhook_url on a job and your server gets notified the moment it finishes, instead of polling for the result.

The My Stuff menu now scrolls to the end on a phone. The last few items used to sit underneath the browser's address bar with no way to reach them, so Usage Report and the items above it were effectively invisible. The menu now sizes itself to the part of the screen you can actually see.

Your video history on a phone no longer shows black tiles with a play button. Every video in your history used to start loading the moment the page opened, and a phone will only load a handful at a time, so everything after those stayed blank. Videos now load as you scroll down to them, and anything you have already saved shows a still frame straight away. This covers every tool on the site that makes video.

Fullscreen video is easier to get out of on a phone. Swiping back now closes the player instead of taking you off the tool you were working in, tapping the dark area around the video closes it too, and the video fills the screen properly when you turn your phone sideways.

Photos straight off an iPhone now upload anywhere on the site. HEIC files used to be turned away with a note asking you to convert them first. We convert them for you now, so you can use the photo you have instead of hunting for another copy of it. AVIF files work the same way.

In Film Studio, a shot you attached your own video to could get stuck. Every attempt to render it again failed with "That reference is a file type we can't use", and retrying never helped. The still we saved for the shot was being mistaken for a video file. Stuck shots work again with no changes on your end.

Uploading a GIF now says plainly that we can't use GIFs and asks for a JPG, PNG, or WebP instead.

MiniMax H3 now takes video references. Add up to 3 clips and the camera move, pacing and scene stay with you across shots. Clips need to be 2 to 15 seconds.

It takes audio references too, up to 3 clips. Give it a voice, a music bed or room tone and the soundtrack it generates follows what you gave it.

Mix all three kinds of reference in one go: up to 9 images, 3 videos and 3 audio clips together. References and start frames are still an either-or choice.

Hover any reference thumbnail in the prompt bar to see it full size. The little tiles stay small, so nothing moves, but you can now check the image or clip you added without opening it.

The same hover preview works everywhere you add references: every tool page, Canvas, Prompt Studio, and Character, Film and Marketing Studio. Extra references hidden behind the count tile preview too.

Character sheets can now be built with Seedream 5 Pro, Seedream 4.5, or Qwen 3 Pro. Pick one from the model list in the Character Studio Sheet tab or on the Character Sheet Generator page.

Seedream 4.5 makes a full turnaround sheet for 1 credit, the cheapest option alongside WAN 2.7. Seedream 5 Pro and Qwen 3 Pro are 2 credits a sheet.

You can now edit a prompt by hand in Prompt Studio. Hit Edit on any version, change the words yourself, then Save. It updates that version in place, so a small fix no longer costs you a turn with the agent.

Enhance now writes in each model's own format, across Seedream, WAN, Seedance, Gemini, VEO, Qwen, FLUX, Z-Image and Ideogram. Every one of these models publishes its own rules, and they disagree with each other, so one house style was never going to suit all of them.

Text you want printed inside an image is now quoted the way each model expects, so signs, posters and labels come out spelled right far more often.

Ask a video model for several beats and you get a proper shot list. Ask for dialogue, music, or sound effects and each one is written where the model expects it, so it lands on the right moment.

Refine and the prompt tips now follow the model you are actually using, instead of giving the same generic advice everywhere.

Featured Tools on your dashboard now scrolls sideways like the On Sale row, so every featured tool sits on one tidy line instead of spilling onto a second row.

Seedance 2.5 is live. Ask for up to 30 seconds and you get one continuous shot, not clips stitched together. Write your prompt as timed beats and it follows them in order.

Spoken audio is generated with the picture in 11 languages, including Chinese, Japanese, Korean, Spanish, Portuguese, Arabic, Thai, and Vietnamese. Lip movement matches the words.

Guide a shot with up to 50 reference materials in one go: 30 images, 10 video clips, and 10 audio tracks. Lock a character's face, borrow a camera move from a clip you like, or hand it a voice to match. Tag them in your prompt and say what each one is for.

Start from a still, or set both a start and an end frame and let it fill the motion between them. Your image sets the shape of the video, so nothing gets cropped.

A Thinking switch lets the model reason about a long prompt before it starts. Off by default, and it costs no extra credits either way.

Seedance 2.5 is also selectable in Film Studio as a shot engine, and it carries your character and voice references through.

Seedance 2.5 Video Edit is live under Edit Videos. Point it at a clip you already have and change what happens inside it: swap a subject, add something to the scene, or take something out. The result keeps the original shape and length.

The same tool can extend a clip instead of editing it. Choose whether the new footage continues after your video or leads in before it, and pick how long it runs.

Refine now works in Canvas. Enhance your prompt, then open the side panel and tell the agent what to change in plain words. It knows which model you are pointed at, and you can generate straight from the panel.

The Canvas prompt bar is now the same one you use on every tool page. References sit in a row along the top, the prompt gets its own full width underneath, and your settings and the generate button share one line.

The prompt box grows as you type instead of holding you to a couple of visible lines. Long prompts are readable again without scrolling inside a small box.

One place to add things. Images, videos and audio used to have separate buttons scattered around the prompt bar, some of them doing the same job twice. Now a single plus opens one menu with everything on it.

Seedream 5 Pro images now land in about 19 seconds instead of nearly two minutes. Seedream 5 Lite went from 28 seconds to 17. Same price, same settings, same quality controls. Just a much shorter wait.

The speed-up applies everywhere Seedream 5 runs: the tool pages, Canvas, Character Studio, Reality Polisher, Consistent Characters, and the API.

Prompt Studio is here. Describe what you want to make, then tell the AI prompt engineer how to change it — every instruction becomes a new numbered version you can compare, delete, save, or open straight in the tool it was written for. Included with every subscription, no credits spent.

Refine on every tool. After you Enhance a prompt, a Refine button now opens the same back-and-forth with the prompt engineer, seeded with your enhanced prompt, on every generation tool on the site.

The public prompt gallery at pixeldojo.ai/prompts got a real purpose: every shared prompt now opens directly in the tool it was made for, prompt pre-filled and ready to run.

Every tool's composer got a cleaner layout: reference images on top, your prompt and settings together in one panel, and one clearly highlighted action. Buttons you can't press yet now say so instead of looking broken, in light mode too.

Character Studio, Marketing Studio and Film Studio now share that same composer. Tap a reference tile in Character Studio to drop its name into your prompt; remove Film Studio cast members right on their tile.

The settings that matter for whichever model you've picked — aspect ratio, quality, duration, resolution — now sit right in the Canvas prompt bar. Changing one used to mean opening Settings, then opening a second More Settings panel inside it, and reading through everything the model accepts.

Canvas has proper reference slots. On models that take multiple references — like Seedance 2 Reference — you can add images, video clips, and audio tracks right in the composer, with a running count so you always know how many more will fit.

You can upload a video straight into Canvas. Editing or extending a clip previously meant generating one there first, so footage from your own camera or library had nowhere to go.

Controls that a model ignores no longer appear in Canvas. A few tools were showing options — like a model-tier picker on a tier you had already chosen — that quietly had no effect on the result.

Save and Download now sit in the top right corner of your Canvas result. They used to sit along the bottom, where the row of previous steps covered them up.

Kling Video Reference is now its own choice in Canvas, under Video Edit. Point it at a clip you already have and it builds a new video from that motion. It sits next to Kling Reference to Video, which still works from a still image, so you pick the one that matches what you're starting with and see the right price before you generate.

You can now rename any image or video in your library. Open it from the library, click Edit next to the title, and type whatever you want — titles used to be auto-written for you with no way to change them.

Your title is what the community gallery shows too, so anything you publish carries the name you picked instead of an auto-generated one. Titles also appear on your library cards now.

Prompt Enhance now understands Seedance 2 and MiniMax H3. It keeps your Image 1 and Video 1 references intact, structures shots the way each model expects, and uses the right phrasing for edits and extensions.

The Seedance 2.0, Seedance 2.5, and MiniMax H3 prompting guides now cover advanced reference prompting: how to give each uploaded image, video, or audio clip one clear job, structure multi-shot scenes, and keep characters consistent from shot to shot. Worked examples included.

Film Studio's shot planner now writes more physical motion direction and carries each shot's ending pose into the next one, so continuous scenes cut together more smoothly.

Qwen 3 Pro is now a frame model in Film Studio. Pick it per shot or for the whole storyboard: sharp 2K stills with the closest instruction following of the frame lineup, 2 credits per frame, up to 3 character references.

The credit counter in the top bar now updates the moment Film Studio charges for a frame, clip, batch, soundtrack, or portrait, instead of waiting for the next page load.

If a network hiccup left your generation stuck on the spinner and you clicked Generate again, each click used to start and charge a separate job. Retrying the same generation now reconnects you to the one already running instead of charging you again.

If your connection dropped mid-generation on WAN 2.6, Video Transform, or Magnific Upscaler, the finished result was rescued behind the scenes but never appeared in your history. Rescued generations now show up on the page like any other result.

If saving a video to My Media fails, the video card now keeps its Save button instead of showing Saved. You can retry right away or download the clip before the temporary link expires, so a failed save can no longer cost you a video.

Use your own image as any shot's frame, right from the shot editor. Upload a photo and the clip animates from it, free.

Create new characters, locations, and props without leaving the shot you're editing. Upload your own reference image or generate one, and the new asset drops straight into the shot.

Your manual edits to a shot description now always stick. The automatic frame check no longer rewrites your text while retrying a frame.

Shot cards keep every button visible while a frame is waiting for approval, instead of pushing Edit off the card.

Flux 3 Video is here. The Flux family's first video model generates clips up to 20 seconds with synchronized sound, in 720p or 1080p, from 5 credits per second.

Start from a text prompt, or animate a still image. Add an optional end frame and the model generates the journey between your two images.

Already have a clip? Feed it in and Flux 3 Video will extend or transform it, up to 15 seconds.

Flux 3 Video is also available in Canvas, so you can generate a clip and chain it straight into your next step.

Typing or pasting a very large seed number no longer fails your generation. Seed inputs across all tools and Canvas now keep the value in the supported range automatically.

When a reference image is blocked by content moderation, you now see a clear message explaining what happened instead of a generic error. Your credits are refunded automatically.

Qwen 3 Pro is here. Write a prompt and it generates an image. Add up to 3 reference images and it edits instead. One tool, both jobs, with sharper detail and closer instruction following than Qwen Image 2.

A full Qwen 3 Pro prompting guide is live, with a prompt framework, six copy-ready sample prompts with real results, and editing recipes. The tool's Community tab starts stocked with browsable examples you can pull prompts from.

Qwen 3 Pro is now a selectable model in Character Studio's Shots tab. Up to 3 references, 2 credits per image.

Simple pricing: 1 credit per image at 1K resolution, 2 credits at 2K. It works in Canvas too, so you can chain it with any other tool.

Removing an image or video from Recent generations or a tool's history now sticks. Deleted items no longer come back after a refresh, and the removal follows you across devices.

Seedream 4.5 now tells you right away when a prompt or image gets blocked by content moderation. Before, it quietly retried the same blocked request several times, so you waited up to a minute longer for the same answer.

Blocked generations now show a clear plain-language message instead of raw technical error text.

MiniMax H3 is here. Start from a prompt, from a still you want to animate, or from reference images of a character you want to keep consistent. All three live in one tool, so you pick the starting point and the tool follows.

Add up to 9 reference images to lock a character, a look, or a scene across a shot. Or set a start frame and an end frame and let the clip travel between them.

Clips come out at 2K with sound generated alongside the picture, not dubbed on afterwards. Length runs from 5 to 15 seconds at 3 credits per second.

There is a MiniMax H3 prompting guide to go with it, built by running two dozen test clips and comparing them side by side. Timed beats, on-screen text that comes back correctly spelled, reference images, first and last frames, and how to direct the sound.

When a model refuses to return a generation, you now get a plain explanation instead of a raw error code. Blocked generations refund automatically, as always.

A WAN 3.0 prompting guide built by testing the model rather than reading a spec sheet. Prompts were run in matched pairs, changing one technique at a time, and every clip on the page is one of those runs.

Side by side comparisons show what each technique actually buys you. Splitting a long clip into stages with a clear end state for each is the biggest win, and writing out what a camera move should look like beats just naming it.

The example clips are now in the community gallery, with the prompt that made each one.

Character Studio has prompt enhance. Type a rough idea, tap the wand, and it writes the full scene for you. It looks at your character's reference photo first, so the scene fits the person you built instead of a stranger.

Save and load scene prompts. Keep the ones that work, then pull them back into any character. Your saved prompts sit in the same library the other tools use.

A full prompting guide for Seedance 2.5 is up, ahead of the model landing on August 7. It covers the core prompt formula, how to give each reference image, clip, or audio file its own job, and how to build a 30 second video as a set of stages.

Templates for the new modes: editing an existing video, extending a clip forwards or backwards, first and last frames, keyframe sequences, storyboard grids, and rendering over a rough blockout. Every template has a copy button.

Reference tables for camera language, popular shot techniques, transitions, and the audio and subtitle syntax, plus a checklist to run through before you hit generate.

Use prompt now brings back your full prompt. Long prompts were getting cut off with a trailing ..., so redoing an older clip quietly ran a shorter prompt than the one you wrote. This affected every tool.

Recalled prompts no longer carry an edited with tag on the end. You had to delete it by hand before generating again.

Film Studio shot descriptions take much longer text now. The box used to stop accepting input partway through a detailed shot.

Merge Videos can now add music. Describe the mood you want and we write an instrumental track sized to your sequence, or paste a track you already have. It mixes in when you export, quietly under any clip audio you kept.

Typing a music link into Film Studio now works. The box used to cut the address off partway through, so the export came back asking for a valid link.

We got early access to WAN 3.0 and made a page of real clips so you can see it before it lands. It generates up to 30 seconds in one go, with sound in the same file, and it can pick the shape of the frame and even the length for you. Have a look at the WAN 3.0 Preview link in the footer.

Character images saved one at a time now show up properly in the gallery instead of opening to a blank space.

You can now pick a voice clip for a character on iPhone and iPad. The file picker was greying out MP3, WAV and voice memo files so there was nothing to choose.

Add Voice now lives in the window where you create or edit a character, right below the reference images, instead of sitting outside it.

Voice clips are kept for good. They used to disappear a day or so after you added them, so a character would quietly lose its voice.

Your 100-credit referral bonus now lands the moment a friend you invited subscribes. If you are owed credits from an earlier referral, reach out and we will sort it out.

The assistant no longer loses track of an image from your library partway through a long chat. Jobs that used to fail with a download error now start the first time.

Connect PixelDojo to Claude or any coding agent and it now handles video and audio references, more character models with multiple reference photos, and model-specific settings. Same smarts as the in-app assistant.

Ask the assistant for any model-specific setting and it can apply it. If a value is off, it gets the exact allowed options back, fixes the call, and you are never charged for the miss.

The assistant now picks the most popular model that can actually handle your request. Name a model that cannot take your video and it switches to one that can, and tells you why.

Aspect ratios and durations from the assistant now apply on far more models, snapping to what each one supports instead of being quietly ignored.

Creating a character, location, or prop now locks it into every shot that mentions it, automatically. A note tells you how many shots use it and which ones were already rendered without it, so you know exactly what to re-render. Your rendered clips are never touched.

Lock in a look you already like: when you create an asset, pick any frame your film has already rendered as its reference image. It costs nothing, and later shots will match it.

Re-plan all shots only appears after you have actually edited the script, and it says up front when rendered clips are at stake. Casting and asset changes apply on their own, so nothing on the script page can wipe your storyboard by accident.

The shot editor got a full redesign. The frame and clip previews are now tabs, each showing only the settings that re-render that piece, and everything autosaves as you type. No more Save button to forget.

Match previous shot now explains itself: a matched frame is copied from the previous clip's ending, so the editor says so, shows the real cost, and makes clear your motion text still drives the clip.

Credit estimates for films using Match previous shot now show what those shots actually cost instead of the standard frame price.

Redoing the same matched shot from two windows at once could charge for it twice. Now only the first request goes through.

Typing in the shot editor while shots are rendering no longer pauses the storyboard's progress updates.

A shot with Match previous shot turned on no longer asks for a description before it will run. That frame is copied straight from the previous clip's last moment, so there is nothing for a description to change. You still write one for the clip, which is what gives the shot its motion.

Adding a shot to a scene now carries the shot before it: the same cast, props, outfits, and models. New shots used to start empty, so characters could turn up in different clothes halfway through a scene.

Sharpening a matched frame now takes about a second and costs a tenth of what it used to. It runs on a faster backbone and comes back at a slightly higher resolution than before, so a long chain of matched shots holds its quality for far less.

A matched shot no longer gets stuck when the sharpening step has trouble. That frame is copied from the previous clip and is already correct, so you now keep it and carry on instead of watching a spinner or losing the shot.

Making a character without a photo works now. Describe who you want in Character Studio, generate four portraits, and pick one as the reference. It used to take the credits and then tell you nothing came back.

Audio references work in the assistant. Attach a track alongside an image or clip and Seedance 2 uses it to drive the shot.

Voiceover emotion choices are more forgiving. The assistant knows the full list now, and anything outside it falls back to a natural read instead of failing.

The Download button in the Recent drawer now saves the file. Before, it opened the image in a new tab and left you to save it yourself.

Downloading a video from Auto-Caption, Video Reframe, or Video Upscaler on your phone now opens the share sheet, so you can save straight to your camera roll. Canvas downloads work the same way now.

Downloads keep their real file type. SVGs from Recraft and WebP or JPEG images no longer get saved with a .png name.

Results now fill the card on your phone. The buttons moved off the picture and under it, and images run one per row instead of two. A 16:9 generation is finally visible on a phone instead of being half covered by controls.

Tap a result to see it full size. No more long pressing and hoping you miss a button.

The close button on a full size image was hidden behind the top bar on mobile. It is back where you can reach it.

P-Image Ideogram is live. It is built for images that have to carry words. Posters, packaging, signage, book covers, labels. Put the exact wording in quotes in your prompt and it renders letters that hold together instead of turning to mush.

You pick the price. A thinking budget setting runs from Very Low at 0.1 credits to High at 1 credit per image. Use Very Low to explore a lot of ideas cheaply, then switch to High for the one you want to keep.

Seven aspect ratios, 1K or 2K output, and JPG, PNG or WebP. There is a prompting guide with copyable examples, and the tool is available in Canvas and on the API.

Adding Pixel Dojo to Claude could fail with a message about a name pattern, and once it did, every tool in that chat stopped working. Our skills are renamed to a format Claude accepts, so the connection works on the first try. Nothing to change on your end. If you connected before, reconnect the server and your tools come back.

Older setups keep working. Saved workflows and any agent that learned the previous skill names still run, so nothing you built needs rewriting.

The Video Upscaler now runs on FlashVSR, a faster new engine. Upscaling costs 1 credit per second of video, half the previous price.

A 2K output option joins 720p, 1080p, and 4K. A nice middle ground when 4K is more than you need.

Your exact prompt is always the one that runs. A hidden setting could let an AI rewriter quietly rephrase your prompt on Qwen image edits and, for API users, on some video tools. It is now off everywhere unless you turn it on yourself.

Results now appear in the order you submitted them. A slow job that finished late could jump to the top of your history and look like your newest generation, which made it seem like an old prompt came back on its own.

Videos we recover for you after a connection drop now show their real tool and submit time in Recent generations, instead of popping up as a brand new mystery clip.

Reusing a seed from an earlier result? The composer now shows a Seed badge on the Options button across 28 tools, so a locked seed is never a silent surprise. Clear the field and the badge goes away.

Job status responses from the API now say processing in some cases that previously said pending. The terminal statuses, completed and failed, are unchanged, so polling loops that wait for a finished job need no changes. If your integration matches on the pending string specifically, treat pending and processing as the same in-progress state.

You can now favorite a video or image right in the fullscreen viewer in your Library. Click the heart button or press F, then keep arrowing through your work. No more backing out to the grid and trying to spot which of five similar clips you were just watching. This one came straight from a customer suggestion, so keep them coming.

Photos of people work again in Seedance 2 image to video. For about a day, the model's safety check was turning away any photo with a real face, so those clips stopped before they started. Uploads now take a different route and go through as normal. Anyone who hit this was refunded automatically.

Trim the start or end of any shot in Film Studio. Clips often take a moment to get moving, and cutting that off makes the finished film flow instead of feeling stop and go. Open a shot and use Trim start and Trim end, in quarter second steps.

Trimming happens when the film is stitched, so nothing is re-rendered and it costs nothing. Your clips stay exactly as they are, and you can change or undo a trim any time before you export.

Shots that continue from the previous clip keep their endings locked, so a trim can never break the handoff between two shots that were made to flow together.

Close the tab and your work still arrives fast. Most image and video models now tell us the instant your result is ready, so it lands in your history right away instead of waiting for the next background check.

Refunds got much faster. If a generation fails or stalls, your credits now come back within minutes instead of up to an hour and a half.

Developer API webhooks now fire on their own. If you set a webhook URL on an API job, we call it the moment the job finishes. Previously the webhook only sent after you polled for status.

API jobs finish even if you never check on them. Every job now completes, delivers, and refunds on failure server-side. Polling is optional, not required.

Rename any film in Film Studio. Click the title on the film's page, or the pencil beside it in your projects grid. Press Enter to save, Escape to leave it alone.

A renamed film carries its new name everywhere it already appears: the caption it shares to the Community gallery, its folder in My Media, and search.

Video Reframe works again. The 2:1 option was never accepted by the model, so almost every attempt failed. The aspect ratio list now matches what the model supports, including a new 9:21 Ultra Tall option.

Seedance 2 tools now check your clips and reference images the moment you pick them. If a frame is too small or a clip is too short, you see the exact numbers and the fix right away, instead of after a full upload and wait.

WAN Reference to Video no longer fails on small reference images. Undersized images are upscaled automatically so the job just runs.

The assistant and storyboard now fill in everything reference-based video models need. Those requests used to fail, and some models quietly dropped your starting image.

Attach a video clip in the assistant and it now goes to a model that can actually use it. Clips used to be treated as still images and bounced with a confusing file-type error.

Aspect ratio choices in assistant video requests now apply on every model. A few engines were quietly ignoring them and always rendering the default shape.

Grok Image Edit retries automatically when the backbone times out. If it still cannot finish, you get a plain message and you are not charged.

Pick a video where an image is needed and you hear about it instantly, along with the formats that work. Same for unsupported video formats and clips over a tool's length limit.

The video upload dialog could occasionally get stuck on Uploading. It now always finishes or tells you what went wrong.

Videos that finish after you close the tab no longer get lost. They now land in your recent creations automatically within a couple of hours, so you never pay for a video you did not receive.

Luma UNI 1 now accepts up to 8 reference images for edits, everywhere on the page. It used to stop you at 3 even though the model happily takes more.

Merge Videos now reminds you that a finished sequence is shown for 24 hours, so you know to save it to My Media or download it before the preview goes away.

Ask the agent for a video with nothing attached and you get a video. It used to reach for a model that needs a starting image and fail with a message about a missing image. If you do attach an image, the agent now keeps it rather than starting the clip from scratch.

Nano Banana Standard only offers the sizes it can actually produce. The 2K and 4K options were listed but never worked, and the credit hint quoted you for a 4K image you would not receive. All ten aspect ratios are still there.

The AI Ads page opens again. It had been failing to load.

When a file you send is too large, the message now tells you how big it is, what the limit is, and what to do about it.

Character Studio reference photos could show up more than once, with the primary image seeming to multiply as you clicked between characters. Duplicates no longer sneak in, and any character that already had them has been cleaned up automatically.

The brand blue is deeper and richer across the whole site. Buttons, links and highlights all sit on the same colour now, so nothing looks a half step off from the thing next to it.

The home page headline has a new treatment. One clean colour sweep across the words instead of a gradient and an underline competing for the same three words.

Section headings across the site and your dashboard are heavier and tighter. Easier to scan when you are moving fast.

Button text was harder to read than it should have been in dark mode. It now meets the accessibility standard for contrast.

The note telling you generations are kept for 24 hours was too faint to read on light backgrounds. It is legible now.

Lost-generation recovery now covers every model on the platform. If your connection drops mid-generation, the finished result is waiting in your history on your next visit.

Before deciding anything about a failed generation, we now confirm what really happened. Real failures refund instantly. Finished work gets delivered instead of apologized for.

Some video generations with sound failed with advice to change your prompt when a simple retry was the fix. The message now says so.

Your history holds a full 24 hours again. On a busy day, older work used to drop off the list early, so a clip you made a few hours ago could disappear when you refreshed the page. Nothing was ever deleted. We just lost track of it, and now we do not.

Clips that turned into a blank or broken player a few hours after you made them now play again. We keep our own copy of every result, and we hand that one over as soon as the original goes stale.

Generations with a reference image sometimes failed with a vague message about not being able to load it, even though the image was perfectly fine. The handoff was dropping about one time in three hundred. We now retry it for you instead of handing you the error.

Uploaded images are kept for 24 hours. If you come back to an older one, you now get told it has expired and to upload it again, instead of being asked to keep retrying something that could never work.

Finished work always finds its way back to you. If your connection drops or you close the tab while a video or image is generating, it still finishes on our side and reappears in your history on your next visit. No more paying for a result you never saw.

Refunds for failed generations are now instant. When a generation fails, we verify it right away and return your credits on the spot instead of making you wait up to two hours.

Refund messages now tell the whole truth. You only see credits refunded when they really were. And if a generation you thought failed actually completed, you get the finished result instead of an apology.

Approving scenes and saving storyboards in Film Studio works again on films edited after the wardrobe update.

Expired images and videos in agent chats now say so clearly instead of showing a broken tile.

Film Studio characters can now change outfits between scenes. Each character in film settings has a switch: keep one outfit for the whole film, or wear something new in every scene. You can also just write the costume change into your script and the planner follows it.

Shots that continue from the previous clip now stay sharp. They borrow that clip's last frame, which is lower resolution than a generated one, so quality used to drift down over a run of connected shots. That frame is now sharpened automatically. You can switch it off per film in Settings if you would rather keep those frames free.

Characters can carry a voice reference. Add a few seconds of a character speaking in Character Studio, and Film Studio sends it along whenever that character has a line, so they keep the same voice from shot to shot. Up to three voices per shot on Seedance 2 Reference.

Download all your storyboard frames as a zip, the same way you can download all your clips. Frames come out in shot order with the script alongside them.

Match previous shot now works on its own, and it is free. A matching shot takes the previous clip's exact last frame as its own frame instead of paying an image model to approximate it, so the cut is seamless and costs nothing. Every shot card shows whether matching is on or off and you can flip it right there. A matching shot waits for the shot before it to finish its clip, and Generate all frames skips it until then.

Films are cut into far fewer, longer shots. A scene in one place now plays as one continuous shot instead of being chopped into a shot per beat, so the room, faces, and clothing hold together. A film also runs only as long as its story needs. The length you pick sets the scope of the story, and nothing gets padded with filler to hit a number.

Conversations now stay in one shot. Back-and-forth dialogue used to be split into a new shot per line, and the room, faces, and outfits drifted at every cut. An exchange now plays out in a single shot with both speakers, and the film only cuts when something new happens. Video Analyzer imports group filmed conversations the same way.

Fine-tune wardrobe on any single shot. Every shot card's editor now has a Wardrobe section that shows what each character is wearing and lets you override it for that shot only.

A quick four-step guide on the Film Studio home page shows how a film comes together, from idea to export.

The script page now uses the full width of large monitors, matching the storyboard.

Deleting a shot now asks first and tells you exactly what goes with it, instead of making you click the same button twice.

If a shot ever gets stuck generating, a release button appears once it is clearly past due, so you can start it again without waiting it out. Stuck work still refunds on its own either way.

Film Studio now tells you when characters, props, or locations have no generated asset yet, so you know what might look different from scene to scene and can lock it in with one click.

Video Analyzer no longer fails whole runs over one unreadable moment. It retries, isolates the tricky stretch, and delivers the full breakdown.

Light mode is easier to read across the site. Banners, labels, and buttons that looked washed out on a light background now have proper contrast. Dark mode is unchanged.

No photos? No problem. When you create a character in Character Studio, you can now describe them in a sentence and generate four portrait options on the spot. Tap your favorite and it becomes the reference that keeps their identity locked across every scene, outfit, and video.

Pick a 5 minute film and you get a 5 minute film. Shot planning was quietly resizing your film to match how much dialogue the script happened to have, so a five minute pick could come back as thirty seconds of shots. Your chosen length is the promise again.

Scripts are now built from story beats instead of lines of dialogue. Silent moments count: an establishing frame, a reaction, a quick cut, a montage step, the closing image. Films get room to breathe rather than talking through every second, and scripts are written to a full arc with a real turn and an ending that lands.

The script writer remembers your last few finished films and stops recycling itself. New openings, new closing images, new character names, new locations. The genre and story you asked for are untouched.

The "some spoken lines didn't fit" notice was appearing on almost every film, even when nothing had been cut. It now counts what your film actually kept.

Outfits stay the same across a film. Each character's clothes are now pinned down colour by colour when your scenes are planned, so an apron that starts navy does not turn white halfway through.

If the automatic likeness check blocks a scene because a prompt or reference looks too close to famous characters, brands, or celebrities, you now see a plain explanation with what to try next. No more cryptic technical error. Applies to WAN scenes in Character Studio and Film Studio.

Seedance 2 Reference, WAN Reference to Video, and Kling Reference to Video no longer offer first frame and last frame shortcuts. Those tools build motion from your reference images, not from a start frame, so the shortcuts never applied. Frame extraction still works everywhere it belongs, including P-Video Animate, WAN 2.2 Animate, WAN 2.2 Replace, and WAN Video Character Swap.

Pixel Dojo film contests now live at pixeldojo.ai/contests. Follow a creative brief, share a finished film from Film Studio, and compete for cash prizes and a place in the winners gallery. The first brief is in production now. Watch that page.

Relive our first creator challenge, Turn Anything Into an Ad. Over 300 AI-made video ads, 986K verified views, and $3,750 paid to creators in 48 hours. The featured ads are on the contest page now, along with the six spotlight awards.

The extract first frame and extract last frame buttons now work properly on P-Video Animate, WAN 2.2 Animate, WAN 2.2 Replace, and WAN Video Character Swap. They used to fail quietly. Grab either end of a clip and send it straight to the Image Editor.

LoRAs and characters are now linked. Train from Character Studio and the finished LoRA attaches to that character automatically. You can also link any existing LoRA to one or more characters from the edit panel in your models library.

Filter your LoRA library by character, and see every LoRA a character owns right on their page in Character Studio and the Library.

The LoRA trainers' single image option got a serious upgrade. It now takes you to Character Studio, where one photo becomes a full training set. Generate shots of your character, pick your favorites, and send up to 30 straight into any of the seven trainers with one click.

Shots you send to training are saved to that character's folder automatically, so your training set stays in your library.

Meet Face Swap. Pick a target photo and the face you want in it, and the swap lands with the original pose, lighting, and background intact. One credit per swap, waiting in the Characters & Faces menu.

Two modes to match the job. Face only blends in the new face and keeps the target's hair. Full head carries the whole head, hair, hairline, and ears included.

Lock a seed to repeat a result you like, or leave it random for a fresh take every run. Developers get the same tool and both modes through the public API.

Playing a video in the community gallery could restart from the beginning a few seconds in while the file was still loading. Playback now runs straight through from the moment you press play.

A few users hit "Number must be greater than or equal to 1296" on every Seedream generation, no matter which settings they picked. A custom size saved on an earlier visit was silently blocking the whole page. That trap is gone, and any affected account heals itself on the next page load. Nothing to clear, nothing to reset.

Custom width and height are retired from the Seedream pages in favor of the size presets, which cover every ratio the models support. Developers can still send exact dimensions through the API.

Match previous shot now continues from the exact final moment of the previous clip, not just its opening frame. A lot can happen in five seconds of motion, so the next shot picks up precisely where the last one actually ended.

Camera settings now shape both halves of a shot. Framing like Close-up, POV, or Aerial is spelled out for the storyboard image so it actually listens, while movement like a slow push-in stays with the clip, where it belongs.

Film plans now choose a camera framing for every shot and pre-mark the shots that flow straight out of the one before, so continuity is set up before you render a thing.

Longer Seedream 4.5 and Seedream 5 Lite runs could end with a false timeout error blaming your prompt, even though the images finished seconds later where you couldn't see them. Generations now run in the background and always come back to your page, no matter how long they take.

Refreshing or leaving the page mid-generation no longer loses your Seedream images. Come back and they pick up right where they left off, finished and waiting.

If a Seedream generation genuinely fails, your credits now come back automatically, even if you closed the tab and never saw the error.

The Portrait Series Generator's Seedream 4.5 model got the same upgrade. Generation starts instantly instead of leaving you on a blank wait, and your portraits keep rendering even if you refresh or leave the page.

Videos longer than a minute now play and download in full from your Library and the gallery. The playback route was silently capped at 60 seconds, so a two-minute film looked truncated even though the complete file was always stored safely. Nothing was lost. Re-open any long video and the whole thing is there.

Scrolling the Character Studio page no longer erases what you typed. The prompt box moves between the top of the page and the bottom bar as you scroll, and crossing that line wiped your prompt, your attached elements, and your settings. Everything now stays put through scrolling, collapsing the bar, switching tabs, and even a page reload.

Match previous shot. Turn it on for any shot and its storyboard frame starts from the frame before it, so the room, the props, and where everyone is standing carry across the cut. Text alone could never hold a set still. This does.

A shot now has two boxes instead of one. The description renders the storyboard image, and a second box holds the movement and the spoken lines. Clips follow direction far more closely, because the motion no longer sits buried under scenery the frame already shows you.

Talk your changes into the whole film. Open the Script tab and say what you want in plain words, like putting a character in a red coat or making the lead a grandmother, and every shot updates at once. No more editing sixty shots by hand.

Every role your script names now gets a casting chip, and every chip has both doors: create the character on the spot, or bring one in from your library.

Share to Community now shows up the moment your export finishes saving, no refresh needed. And when you re-export a film after changes, the share option comes back for the new cut.

Silent shots stay silent. Video models sometimes invented dialogue for shots that had none written. Every shot without a spoken line now tells the model explicitly: nobody speaks, ambient sound only.

The Download button actually downloads now. It used to open your film in a browser tab and leave the saving to you. Same fix for the soundtrack download.

Drag shots into a new order. Grab any card on the storyboard and drop it where it belongs in the scene. The arrow buttons still work too.

Props get casting too. The script page now spots recurring objects in your story, a weapon, a locket, a vehicle, and offers them as one-click assets so they look identical in every shot. Anything you attach shows up with a live count of the shots it rides, and every shot editor has a new Props and Set section.

Wardrobe stays put. Once you say what a character wears, every shot renders that exact outfit, whether or not the shot description bothers to mention it.

Open in Film Studio failed with a bare Required error after analyzing a video. It now opens your film.

Imported shots used to describe the plot instead of the picture, saying things like a character is progressively arming herself, which produced the wrong image every time. Imported shots now describe only what is on screen in that moment.

Paste a video link anywhere you pick a video. The picker is three clean tabs now, and a slow link is treated as still checking instead of being rejected outright.

Rewrite the script any time, even after your scenes and shots exist. A Script and Storyboard switch sits at the top of every film, and if you edit the writing after planning, the storyboard tells you it is out of date and offers a one-click rebuild.

Every spoken line gets its own shot. Paste a script and the film sizes itself to fit it. If lines ever get trimmed to fit a length, one click extends the film and keeps every word. You trim the script to shorten a film. We never cut your writing for you.

Re-plan a single scene from the script while the rest of the film, and the clips you already rendered, stay untouched.

Film length is now a setting you can change any time, not something locked in when you create the film.

A much calmer film page. One clear next-step button instead of three loud ones, engine choices and settings tucked into one panel, and on your phone the action bar stays docked at the bottom so it never scrolls away.

Generate all shots really means all of them. The old four-at-a-time limit is gone. Click once and every clip renders in the background.

Films made with Grok Video or Happy Horse came out of the export silent, and adding a soundtrack left you with music over a film with no voices. Your clips always had their dialogue; the export was dropping it. Re-export any film and the dialogue comes back. Exports are free, so this costs you nothing.

A soundtrack is now clearly optional. You can export with just the dialogue and room sound from your clips.

Three new looks: Romance, Romantic Comedy, and Comedy. Each one carries its own lighting and scoring through every shot in the film.

When Export is unavailable, Film Studio now tells you which shots are holding it up and what each one still needs, instead of just greying the button out.

Export starts right away now. It used to re-download every clip in the film to inspect it before it could even begin, which on a long film meant a long wait before anything happened.

Scrolling through a creator's profile gallery kept snapping you back to the top. It now holds your place while the next batch loads in.

Profiles with a lot of work stopped loading after the first few hundred pieces, even though there was plenty more. You can now scroll a full profile end to end.

Profile galleries now use the same layout as the Community Gallery. Every piece shows its like and comment counts, and you can like, comment, and favorite right from the grid.

Attach several reference images to the agent at once. Send a photo of a person and the watch they are wearing, and both make it into the shot. Drag in a whole set and the agent uses all of them, in the order you attached them.

Your saved assets now work in the agent. Pick a character, product, location, or outfit from the new From My Assets menu and the agent refers to it by name and holds its likeness. Characters bring several stored angles along, which is what locks a face.

Some edits quietly ignored your reference photo and returned a brand new image instead. Your reference now reaches every edit model, so an edit is always an edit of your picture.

The agent now works from a hand-picked lineup of our best models instead of everything on the site. Plain image requests go to Seedream 4.5, video goes to Seedance 2, and the models Character Studio and Film Studio rely on are the same ones the agent reaches for.

Follow-up shots in a conversation sometimes jumped to a different model partway through and lost your subject's face. The agent now stays on the right model for the whole session.

API docs now match exactly what the API accepts. Reference image limits are published for every model, so you know the ceiling before you hit it, and three video endpoints that required an undocumented setting work again.

Editing through the PixelDojo MCP server now sends your reference images the way each model expects. Edits from Claude, Cursor, and other agent tools that used to fail go through. Update with npx @pixeldojo/mcp@latest.

Seedance 2 blocks references that show a famous face, a brand, or a copyrighted character. It used to report that your file couldn't be read, which sent people off re-exporting a perfectly good image or clip. You now get the real reason: the model's content filter blocked it, so try a different reference or switch to another model. Credits are still returned.

Reference to Video, Video Edit, and standard Seedance 2 no longer show raw technical error text with an internal request code when a reference is rejected.

Film Studio now writes with Grok 4.5. Scripts come back with stronger arcs, sharper dialogue, and shot plans that cover your story more faithfully, and planning is noticeably faster.

The agent thinks with Grok 4.5 too. It turns short requests into richer, production-ready prompts and picks the right tool faster.

Every tool's create bar has been redesigned. Your prompt gets the room it deserves, reference images and clips appear as tappable tiles with one-tap remove, and settings sit in tidy chips. It's the same layout on every tool, so nothing needs relearning.

On phones the create bar stacks cleanly: a full-width Generate button, controls always in reach, and nothing scrolling sideways.

Inpainting got a modern refit. The brush controls moved above the canvas where nothing can cover them, and the editor wears the new look. Painting itself works exactly as before.

A fast double-click on Generate could start two jobs and charge twice on some tools. One click, one job, one charge now.

On several tools the Subscribe button did nothing when you weren't subscribed yet. It now takes you straight to upgrading.

Saving an earlier image now records that image's own prompt and settings, not whatever the create bar happened to say at that moment. Saved videos also land in the right tool's Saved tab again.

The Enhance wand is now in Canvas. Tap it next to Generate to turn a rough idea into a rich, detailed prompt, tuned to your style, without leaving your session.

Enhance also landed on eight more tools, including Seedance 2, Krea, P-Video, and Ideogram Character. On reference video tools it keeps your character tags exactly as written while enriching the scene around them.

If a prompt can't be enhanced, you now get a clear message telling you why and what to do, instead of silence. Your credit is always returned.

When the enhancer hit a temporary glitch, some users saw a confusing technical error. Now your original prompt stays in place, your credit comes back, and you can simply try again.

Film projects, setup menus, and the Help panel now work properly from a keyboard and give screen readers the right names and selected states.

Film Studio's soundtrack controls no longer run off smaller phone screens, and the new-film composer starts tucked away on short displays so the workspace stays usable.

Finished storyboards load fewer videos up front. Offscreen clips wait until you scroll near them, making larger film projects lighter and faster to open.

When a LoRA training couldn't read your photos, it failed with a cryptic technical error and no clue what to do next. Now it tells you exactly what went wrong — images tucked inside a folder in your zip, or an unsupported format like HEIC — and how to fix it. Credits are always refunded automatically when a training fails.

The single-image trainer now double-checks every photo before adding it to your training set and skips any that didn't download properly, so a training can no longer start from broken files and fail minutes later.

If your training images fail to upload, you find out immediately — before the training starts and before any credits move.

Two more clip engines in Film Studio: Grok Video (2 credits per second) and Happy Horse 1.1 (3 credits per second). Both hold your storyboard frame and cast references. They render silent clips, so keep Seedance, Kling, or WAN for shots with spoken lines.

Happy Horse Reference-to-Video now defaults to the newer 1.1 engine: same quality family, 25 percent cheaper per second at 720p. Classic 1.0 stays in the version picker.

The Grok Imagine and Happy Horse prompting guides now cover reference-to-video properly: action-first prompting for Grok, and Happy Horse's character token binding, including the storyboard-anchor trick Film Studio uses.

Film Studio storyboards got a polish pass. Shot frames render larger on big screens, Cinematic films get extra-wide frames, shot details are easier to read, and every control is comfortable to tap on your phone.

Frame check now retries a mismatched shot up to 3 times, each with a smarter prompt rewrite. If it still can't match your script, the shot gets flagged so you know to take a look.

Redo clip now stays visibly locked while the new clip renders, and a note tells you the moment it's ready. No more wondering if anything happened.

The shot editor shows the frame and the clip side by side, and you can redo either one right from there.

Starting a new film on mobile now shows the cost estimate up front, the same as on desktop.

The Skills page now shows the film skill in action, featuring a real short film made end to end by an AI agent in a single call.

Qwen Image 2 Pro shots sometimes added a second person to the scene when your character's reference photos included friends or group shots. Shots now lock onto your primary reference photo, so one selected character means exactly one person in the result.

Meet Film Studio. Type one sentence and it writes a screenplay, storyboards every shot, renders each clip with lip-synced spoken dialogue, and stitches the finished film with a soundtrack. You direct every step. Find it under Generate Videos.

Cast your own characters. Attach characters, locations, and outfits from Character Studio and they stay consistent from the first frame to the last, wardrobe included.

Storyboard first, spend after. Every shot gets a still frame you approve before any video renders, so you only pay for shots you want.

One-click soundtrack. Generate an instrumental bed matched to your film's look and it mixes in under the dialogue when you export.

Watch ten real films made start to finish in the studio, then export yours, share it to the Community, or download every clip plus the script for your own editor.

Refreshing the page no longer wipes your composer. Your prompt, reference images, and settings come right back on every image and video tool. That includes the refresh after our new-version banner, so an update never costs you a draft you were still writing.

Results now stick around after a reload on Outpainting, Background Remover, Extract Frame, Character Stylist, Flux Redux, and Smart Resize. Your generations from the last 24 hours stay visible, the same as every other tool.

Seedance video generations occasionally trip an automatic audio check on perfectly innocent clips and used to fail with a cryptic technical code. You now get a plain explanation that it is usually a false positive and that generating again typically clears it. Credits are refunded either way.

A few video and upscaling tools could reject a resolution, duration, or scale setting that should have just worked, mostly over the API. WAN Reference to Video, WAN 2.2 Animate and Replace, and the Magnific upscaler now take those settings cleanly, and fall back to a sensible default when you leave one out.

Reve 2.1 is here. Generate, edit, and remix in one tool: no reference image means text-to-image, one means precise editing, and up to eight means remixing them into a single scene. A flat 4 credits per image, and it works on Canvas too.

Your prompts are now followed exactly, everywhere. Several models were quietly rewriting prompts before generation, which is why results sometimes drifted from what you asked for. We audited every image and video tool and turned that rewriting off across the site: WAN, Qwen Image, Z-Image, Hailuo, Ideogram Character, Flux, and more, on tool pages, Canvas, the Studios, and the API. What you type is what the model gets. API users can still opt in per request.

The Krea 2 LoRA Trainer is here. Teach Krea's aesthetic image model your face, product, or art style from as few as 10 photos. Training costs 60 credits per 1000 steps, and most models finish in a couple of minutes.

Use your trained LoRAs on the Krea Image tool for just 0.5 credits per image. Pick up to three from the new Add LoRA menu on the Turbo tier, dial in each one's strength, and your trigger word drops into the prompt automatically.

Finished LoRAs get an automatic preview image generated with the LoRA itself, so your models are recognizable at a glance in the Library. You can swap in your own thumbnail any time.

Every LoRA trainer got a design refresh. Cleaner training pages for Flux, Flux 2, Qwen, WAN, Z-Image, and SDXL with clearer pricing and the same one-click flow.

Single-image training now uses our latest image models to build your dataset. Choose Seedream 4.5, Seedream 5 Pro, Nano Banana Pro, Nano Banana 2, WAN 2.7, or Qwen Image 2 Pro, starting at 1 credit per variation with instant results.

Freshly started trainings now show up in your Library immediately instead of appearing a few minutes after submission.

The home page now leads with Studios. See real, unedited Marketing Studio ads and a Character Studio consistency strip before anything else. PixelDojo is complete workflows, not just a wall of models.

Agent users get a callout up top: one install command puts PixelDojo's skills inside Claude, Cursor, or Cowork, so you can render a finished ad or a consistent character with one call.

Pricing, testimonials, and common questions moved up so you can size us up without scrolling through five demo sections first. The demos and community showcase now live right below.

The ad composer slimmed down from seven rows to four. The scene description now lives behind an Edit scene line (the preset writes it for you), and the hook picker sits right on the script field. Less to read, same control.

One Quality control replaces the separate engine, tier, and resolution pickers. Pick your engine and its options in one place, with per-tier prices shown right on the buttons.

Every ad in your history has Run again and Use these settings. Re-render the exact same ad or load its style, product, script, and quality back into the composer for a tweak.

The composer remembers your product, character, style, and settings between visits. And switching ad styles no longer erases a hook or scene you edited.

Character and Location tuck behind an Add button until you use them, so adding your product stands out as the one required step.

The Saved and Community tabs on every tool page now lay out like a gallery. Images and videos show in their original shape instead of being cropped to squares.

The gallery layout no longer jumps around while media loads. Each tile reserves its final size up front and stays in its column, so what you are looking at stops moving.

Every tile has a fullscreen button. Tap it to watch a video or inspect an image at full size.

The make-public button on saved videos was hiding behind the sound toggle. It now sits beside it and says Make public, so sharing to the community gallery is one obvious tap.

The Saved tab now links to your Library instead of the old My Media page.

Hitting Save on an image or video card now flips the button to Saving with a spinner until it finishes, then confirms with Saved. No more wondering if the click landed, and no accidental double-saves. Works across every tool.

The Marketing Studio is now in the menu for everyone. Find it next to Character Studio at the top, under Generate Videos, and featured on your dashboard. Pick a product, pick an ad style, and render a video ad with voiceover in minutes.

The Generate Ad button no longer goes quiet when something is missing. It now tells you exactly what to do next, and if your product has no photos yet it opens the editor with the photo tools ready.

The Skills page now shows the ad maker in action with real example ads you can play, plus the API call that renders one from any product URL.

Clicking Generate could sometimes fail with an Unauthorized message even though you were signed in. It usually hit after your computer woke from sleep or a tab sat idle for a while. Your session now refreshes automatically and the request retries on its own, so the button just works.

Long video generations could end with Unauthorized in the result box, with credits taken and no video delivered. Progress tracking now stays authorized for the full length of every job across all video models, so your render arrives even if your session expires along the way.

Meet the Marketing Studio. Pick a product, add one of your characters as the presenter, choose a location, and a preset writes the whole ad for you, hook included. One click renders a vertical video with sound, ready for TikTok, Reels, or Shorts. Choose your engine: Seedance 2, Kling, or WAN, with live pricing as you switch.

Products join your asset library. Paste a store URL and the name, description, and photos import automatically, or add your own. Saved products work across your tools, not just in ads.

Twenty-four ad styles across UGC, unboxing, try-on, showcase, and commercial, each with a real 4K example video made by this tool. Pick a named hook, type the exact line you want spoken, and edit any scene text. The preset is a starting point, not a script you're stuck with.

The Marketing Studio is also an API and an agent skill. POST /api/v1/ads/run turns a product URL plus a style into a rendered ad, and the pixeldojo:ad skill brings it to Claude, Cursor, and any MCP client, so you can automate generate-and-post pipelines. Style catalog at /api/v1/marketing/presets.

Open any image in your Library or My Media and hit the new Use in button. It lists every tool that can take that image and opens your pick with the image already loaded in the right spot: upscale it, animate it as a video start frame, use it as an edit reference, or try clothes on it. Live on Magnific Upscaler, Seedream 4.5, VEO 3.1, Kling Video, and Virtual Try-On, with more tools joining.

Virtual Try-On's pickers now know what they're for. The garment picker shows only your outfit elements and the person picker only your characters, so the right choice is always one tap away.

Virtual Try-On's staged photos rendered as stretched banners that cropped faces. They're proper thumbnails now.

Meet the Library, your new home base. Your media, characters, scene elements, trained LoRAs, and saved prompts now live together as tabs in one place. Find it right next to Dashboard in the menu. Every old link and bookmark still works.

Your trained LoRAs launch loaded. The Create button on any model opens the right tool with that LoRA selected, its trigger word placed in the prompt, and the recommended strength applied. One click from model to generating.

Scene elements created without a reference photo now get an automatic preview image generated from their description, so your outfits, locations, and styles are recognizable at a glance.

The dashboard got a refresh. Character Studio is featured up top, and your recent creations moved higher so your own work is one of the first things you see.

The Add Image picker in every tool now has a Characters & Elements tab. Pick any reference image from your Character Studio characters, locations, outfits, and styles, and drop it straight into an upscale, edit, or video generation. No more re-uploading the same face.

Saved images and videos now remember the source image they were made from and their aspect ratio. This lays the groundwork for recreating edits with the original input loaded automatically.

If part of a Seedream 4.5 or 5 Lite batch failed, the credits for the failed images were quietly kept even though the app said they were refunded. Failed images in a batch now refund automatically, matching how Seedream 5 Pro already works.

Pinning a seed and requesting multiple images used to return near-identical copies while charging for each. On Flux 2 Flex, Flux.2 Studio, and WAN 2.7, each variation now gets its own seed derived from yours, so batches stay distinct and any single result can still be reproduced exactly.

Editing an image used to lock you to one result at a time. The image-count picker disappeared the moment you attached a reference. It now stays available, so you can generate several edit variations in one click and pick the best. Live on Seedream 4.5 and 5, GPT Image 2, Flux.2 Studio, Flux 2 Flex, Nano Banana 2, P-Image, WAN 2.6 and 2.7, and Grok Image.

Flux.2 Studio's edit mode showed a single-image price even when more images were selected. The price you see now always matches the credits you spend.

Adding a second character to a shot could produce a lookalike stranger instead of your actual character on some models. Scenes now tell the model exactly which reference images belong to which person, so every face in a multi-character shot matches its references. Your prompt stays yours. The matching happens behind the scenes.

WAN 2.7 no longer sneaks an uninvited extra person into scenes with a cast member.

Qwen Image 2 Pro caps a scene at 2 people with a clear note explaining why. Its reference limit couldn't hold 3 distinct faces accurately, so we stop you before the shot instead of after.

Images across the site were displaying at half quality due to an overly aggressive delivery setting. Everything now renders at 90-95% quality, so what you see on screen matches what you actually created. This covers tool results, Canvas, My Media, the community gallery, and every landing page.

Video thumbnails and preview frames got the same upgrade and now display sharp instead of soft.

Shots now use up to 6 of your character's reference images (up from 3) on models with room for them, and attached cast members bring a second reference when the budget allows. More angles means faces stay truer to your character.

If a reference image can't be stored when you save a character, you now get a clear message naming which image failed and what to do, instead of a save that silently loses that image a day later.

Reference images from the studio's first days could expire and show as broken icons. They now display as a clear "expired" tile you can remove and replace in one edit. New saves store every image permanently.

The API model catalog and example galleries now load dramatically faster. Preview images are served at display size instead of full resolution, cutting a typical catalog visit from over 60MB down to under 1MB. Big win on mobile connections.

Docs for 3D and audio models now describe their pricing and outputs correctly: meshes are priced per mesh, audio per clip, and 3D response examples show the right file type.

Vidu Q3's audio toggle in Canvas now works. Turning audio on used to be ignored quietly. Now your video comes back with sound.

Four video models in Canvas (PixVerse V6, Kling 2.6, Hailuo 02, P-Video Avatar) now show their proper settings panels: real sliders, labeled options, and toggles instead of plain text boxes.

Switching tools in Canvas (Generate, Edit, Upscale, Create Video) now clears the prompt box. Before, leftover text from the tool you just left could carry over and get used by the next one.

Tidied up the video settings in Canvas. When you animate an image, the result now simply follows your image's shape, and a couple of options that had no effect were removed.

Put more than one character in a scene. Open Add to scene and attach your other characters as cast, right alongside locations, outfits, and styles. Their reference images and names ride along so the model knows who is who.

Everything in a shot now shows as a chip in the prompt box: your main character, attached cast, and scene elements. One glance tells you exactly what is going into the generation.

Recall on a result now brings back the attachments too, not just the prompt. The exact cast and elements from that shot snap back into place, ready to tweak and rerun.

Recreate from My Media. Saved studio shots now reopen Character Studio with the character selected and the prompt and attachments already loaded.

Reference budgeting is now automatic. On models with tight reference limits, attached cast and element images get priority so a named face never silently drops out of the shot.

Model sales now apply inside multi-model tools too. If a model is on sale and you pick it from another tool's model menu, you get the discount and see the badge, same as on its own page.

The On Sale strip on your dashboard scrolls as a carousel when there are more deals than fit on screen.

Model sales now show in Character Studio. You were already being charged the sale price; the Generate button now shows it too, with the discount badge, for whichever model you pick.

Recall prompt on a result now fills the composer with that shot's prompt, ready to tweak and rerun.

In-progress generations can now be dismissed with the X on the card, in the studio and on every tool page. If a dismissed generation fails, your credits are still refunded automatically.

Generations that got stuck showing as processing after a refresh now clean themselves up.

Connecting PixelDojo to Claude Desktop, Claude Code, Cursor, and other AI apps now uses one short address: pixeldojo.ai/mcp. Paste it, approve in your browser, done.

Pasting the short address used to sign you in successfully and then fail to connect. Both the short and full addresses now work.

Consistent character generation now works when you attach a reference photo. It defaults to a sharper model too.

Upscale now correctly applies the scale amount you ask for, on every model.

The Agent's default video model is fixed, so a plain video request in chat or through an automation reliably completes.

Campaign packs that include a video no longer fail on that step.

Saved workflows now wait for a slow step to finish instead of giving up early.

/save and other follow-up actions in the Agent now correctly grab your most recent result, even when it was a video that took a little longer to render.

Type /save in the Agent (or say 'keep this') and the last result is stored in My Media permanently. Works for images and videos, from the Agent, the API, and MCP agents alike.

Library searches now look through your saved My Media first-class, not just recent generations. Your curated work is always findable, and saved copies never expire.

Saves made through the Agent remember which tool made them, so they appear in the right Saved tab with a working Recreate link.

Ask the Agent for a full audio scene: multi-character dialogue, sound effects, and music in one clip. Type /audio or just describe it.

The Agent can now search your past generations. Say 'the dragon from yesterday' and it finds it, then reuses it in your next edit or video. Free, no credits.

A fresh Agent home: bigger canvas, eight one-tap starting points including Character, Audio Scene, Upscale, and Campaign, each backed by a real output.

The same Agent skills now work from your own tools. The new connect card on the Agent page gets you set up in Claude Code, Cursor, and more with one command.

Connect AI agents without installing anything: point any MCP-capable client at our hosted server and every skill is there. The npx package still works too.

No more API key copy-paste for the hosted connection. Your client opens a browser window, you sign in and click Approve, and you are connected. Revoke access anytime from your API Keys page.

Agents can now search your generation history and reuse past outputs as reference images. Find yesterday's hero shot, make three new variants.

Save a multi-step chain once, then run it by name from any chat. Steps feed each other automatically, and you fill in the blanks at run time.

Agents can produce audio too: the new audio skill drives Seed Audio 1.0 for speech, dialogue, and fully produced scenes.

The Skills & MCP page got a full redesign: pick your platform (Claude Code, Cursor, Claude Desktop, Codex, Windsurf, or Cowork), pick a connection method, and follow three copy-paste steps. Cursor gets a one-click install, and a new FAQ answers the common setup questions.

Seed Audio 1.0 is live. Type a script and hear it spoken, or describe a whole scene like a late-night radio drama and get a fully produced clip with voices and sound effects.

Pick from 20 preset voices across English, Chinese, Spanish, and more, or let the model choose the right voice for your prompt. Speed, volume, and pitch are all adjustable.

Pricing is simple: 0.1 credits per second of audio. You only pay for the length of the clip you actually get.

Seedream 5 Pro, the new flagship of the Seedream family, is live. Sharper realism, stronger prompt following, and the best in-image text rendering we've seen. 1 credit per image at 1K, 2 credits at 2K, with results in seconds.

The Seedream 5 page now has a model picker. Pro is the default for finals; switch to Lite for fast 1-credit drafts. Your choice is remembered.

Seedream 5 Pro also joins the model pickers in Character Studio, Consistent Characters, Reality Polisher, and Canvas. Edits accept up to 10 reference images.

A new Seedream 5 prompting guide is up, with ten real generated examples, copy-ready templates, and the settings that matter for each use case.

Every example on the Skills page now shows the real images that exact call produced: the generated portrait, the same-character scene, the before-and-after edit, the four storyboard shots, the upscale detail crops, and a full campaign set.

The edit skill gets its own section: point at an image, say the change in plain language, and the subject stays put. Local files upload automatically.

Seedream 4.5 joins the Shots picker at 1 credit per image with instant results and room for 14 reference images. The fastest way to iterate on a scene.

WAN 2.7 brings budget-friendly scenes with strong identity preservation, also 1 credit per image.

Qwen Image 2 Pro adds crisp detail and the sharpest text rendering for signs, labels, and props at 2 credits per image.

That makes six models in the Shots picker, from 1 to 5 credits per image. Every one keeps your character's reference images and attached elements in play.

Ideogram Character now delivers every image you pay for. Requests for multiple images through the API or Character Studio were charged correctly but returned a single image. If a batch was affected, the unused credits were already refunded automatically.

When a generation fails in Character Studio, the card now stays visible with the reason and a note that your credits were refunded, instead of quietly disappearing.

Pinning the same shot as a reference twice no longer creates a duplicate. Pin buttons now show when a shot is already pinned.

A new tip in the studio explains that your reference images carry the identity, so prompts work best describing the action and setting.

The studio now teaches by example. Before your first generation, it shows a real character carried through locations, outfits, styles, and a finished turnaround sheet, so you can see exactly what it does.

The studio composer now collapses out of the way, just like every other tool, so you can browse your results on the full screen.

Draft videos in P Video now cost credits per second instead of a flat price. A short 5-second draft at 720p is now 0.5 credits, half the old price. Longer drafts scale with their length, so you always pay for exactly what you render.

Every API model page now has a live playground. Fill in the parameters, see the exact request as you type, and run a real test generation without writing a line of code. You see the same request and response your integration will use.

The cost of your exact settings shows before you run, next to your credit balance. Test runs are charged like real API calls, and failed runs are refunded automatically.

Results render right on the page: images, video, audio, even 3D files. Copy the working request as cURL and drop it straight into your code.

Testing an editor or upscaler? Upload a reference image right in the playground and it becomes a URL your request can use.

A handful of image models could leave a job stuck in pending forever when run through the assistant. Those now finish and deliver results reliably.

Character Studio is live. Create a character with a few reference images and every studio tool uses them automatically. No more re-uploading the same photo for every generation.

Scene elements make recreating a scene easy. Save locations, objects, outfits, and styles once, then attach them to any shot. Your character shows up in the same world every time.

The Shots tab puts your character in any scene you can describe, powered by Nano Banana Pro and Ideogram Character. The Sheet tab builds multi-view turnaround sheets in one click.

Love a result? Pin it as a new reference to make your character even more consistent, or expand your reference set with four fresh angles in one click. Everything saves to a folder just for that character.

Every API model page now shows a working code example for that exact model. Editors and upscalers show their image inputs, video models show valid durations, and text-to-speech shows a real speech request. Copy, paste, and it runs.

The AI-assistant docs got the same upgrade. Point Claude, GPT, or any coding agent at a model's llm.txt and it gets accurate parameters and the exact response shape, including audio and 3D models.

3D generations through the API now return your mesh file when the job completes. Before, some 3D jobs finished with an empty result.

Prompting guide links on API model pages now point to the right guide for each model family.

API catalog polish: model counts are consistent everywhere, searching and filtering keeps your place on the page, and every card shows the full endpoint path.

Ideogram 4 Instant is here. Near-instant text-to-image for just 0.2 credits, perfect for rapid drafts and trying lots of options fast.

Ideogram 4 Fast joins the lineup at 0.5 credits, tuned for speed with the same great text rendering and design Ideogram is known for.

Pick your mode on every generation. Instant and Fast for quick, low-cost drafts, or Turbo, Balanced, and Quality for the highest fidelity. Available on the tool page, in Canvas, and through the API.

Your private prompts stay private. We tightened the safeguards so a prompt you keep private can only ever be seen by you.

Searching your prompts is smoother and covers your whole collection, with no flash of old results while the new ones load.

The Edit button now opens your prompt straight into the Builder, and drafts you are working on are kept so moving around no longer loses your progress.

My Prompts loads faster, fits neatly on mobile, and shows clearer states when a search finds nothing or something fails to load.

Flux 2 Flex, Flux Studio, Creative Upscale, and Seedance Video Edit are now open to browse without an account. Take a look around before you subscribe.

The AI Images page got a tune-up: the featured gallery now starts on All Models, page numbers are accurate everywhere, and the whole page loads complete on the first paint.

The AI Videos page got the same tune-up. A showcase clip that had stopped playing is back, and model counts now match reality: 26+ video models and growing.

The whole Character Studio toolkit is now open to browse without an account, from Consistent Characters to Virtual Try-On. Plus a new FAQ covering how character consistency works.

The home page loads its full content on first paint and every model count on it is now accurate: 29+ image models, 26+ video models, 70+ total.

The LoRA trainers page loads complete on first paint, and the training wizards for Flux 2, Z-Image, and WAN 2.2 are open to browse before you subscribe.

The usage report now runs on your local timezone. Today means your today, and hourly charts line up with your clock instead of a faraway server's.

Three new date ranges: Last 7 Days, Last Month, and Last 6 Months. Perfect for spotting longer trends in your credit usage.

Tap any bar in the usage or refund charts to pin its details below the chart. Especially handy on phones.

Switching date ranges updates the report in place instead of flashing a loading screen, plus clearer chart labels, cleaner model names, and a batch of small fixes across the page.

Comment threads in the community gallery hold together better. Replies stay attached to their conversation instead of dropping out of the thread.

The community gallery loads faster, and switching filters or searching no longer flashes older results before the new ones arrive.

Searching your LoRA models now checks your whole library, not just the models already on screen. If a model exists, search will find it.

Editing a model's details no longer risks breaking it. Changing the name or trigger word keeps everything else intact.

My Models loads faster and stays up to date as trainings finish, with cleaner empty and error states and a Retry button if something fails to load.

Strengthened the privacy and security around your trained models and their files.

Clearing search in My Media now works. The X button, deleting the text, and switching folders while a search is active all reset your view reliably.

My Media loads faster, and your folders and their counts are ready the moment the page opens.

More polish across My Media: sort your work by Most Voted, use every card action on touch devices, and get a Retry button if the gallery ever fails to load instead of a blank screen.

When a model goes on sale, you'll see the discounted price right on the Generate button along with the sale end date. Discounts apply automatically. No codes to enter.

Deals are easy to spot. Discounted models get their own picture cards on your dashboard, a percent-off badge in the All Tools menu, and a callout on the home and pricing pages while the sale runs.

Your savings are visible as they happen. Sale generations show the credits you saved on the result card, and your dashboard totals your savings for the month.

The Canvas model picker now shows exact credit costs for each model instead of estimates.

WAN 2.1 returns as a unified video tool — type a prompt for text-to-video, or drop in a starting image to animate it. 480p or 720p, clips from 1 to 5 seconds, at just 1.5 credits per second.

Bring your own LoRA: paste any HuggingFace, CivitAI, or .safetensors URL to apply custom video styles — one of the community's favorite reasons to use WAN 2.1.

Seedance 2 Mini is here — the same Seedance 2 you love at roughly half the credits. Just 1 credit per second at 480p and 2 at 720p, for clips up to 15 seconds with native audio.

Pick your price-quality trade-off on every generation: Mini for drafts and high-volume work, Fast for everyday clips, High when you need 1080p or 4K. Available on Seedance 2, Seedance 2 Reference, Canvas, and the API.

Seedance 2 Reference now checks your reference clips before you spend credits: if your videos run past the 15-second combined limit, you get a clear message with your actual total instead of a failed generation.

The Refer a Friend page now features our affiliate program. Earn 30% recurring commission on every subscriber you send our way, paid out monthly. It stacks with your 100-credit referral bonuses.

Prompt Enhance now has its own wand button right in the prompt box. One tap turns a rough idea into a rich, detailed prompt. No more hunting through the menu.

Enhance now learns your style. It looks at the prompts behind the images and videos you saved, downloaded, and liked, then writes new prompts that sound like you.

Your own creations now guide every enhancement. Enhance pulls examples from your past work, including private images, so suggestions match what you actually make.

Smarter example matching behind the scenes. Enhance now finds better reference prompts for each model, so results fit every tool's strengths.

Free tools are now easy to spot. Tools that cost zero credits, like Background Remover and Style Transfer, show a green FREE badge in the menu, and a Free filter helps you find them all.

Every category in the All Tools menu now puts the most-used models first, based on what creators actually generate with. The best tools are at the top, not buried.

Tool search got smarter. Results are ranked by how well they match, and if nothing matches you now see the most popular tools instead of a dead end.

Your Recent list now shows only creation tools. Pages like My Media no longer push your recently used models out of the list.

Every tool now has one consistent name everywhere: menus, search, and the API catalog. NEW badges also retire automatically, so a badge actually means new.

The My Stuff menu no longer shows the same shortcut twice, and pinning tells you when your pin bar is full. Arrow keys now work across the whole tools menu too.

The help assistant now knows every tool and guide on Pixel Dojo. Ask which tool makes videos, what a model does, or where to find a prompting guide, and it points you to the right page with a link.

Answers now stream in as they are written, so you get help right away instead of watching a spinner.

Refreshed the help panel with one-tap starter questions and a cleaner look. Your conversation now stays put if you refresh the page.

Credit refunds can no longer get lost if something goes wrong at the exact moment a refund is issued. Your balance always ends up right.

Heygen Avatar now settles your credits against the real length of the finished video. If your clip comes out shorter than estimated, the difference goes back to your balance automatically.

Audio lip-sync no longer costs a flat 120 credits no matter what. It holds a deposit up front, then refunds down to 2 credits per second of the actual video.

Scripts in Chinese, Japanese, and Korean now get accurate cost estimates instead of being priced as a single word.

Merging clips with Keep Audio on now actually keeps the audio. Videos made with sound, like WAN 2.7 clips, no longer come out silent after a merge.

Flux Image to Image is back in the sidebar after briefly going missing. All its settings work exactly as before.

Video jobs that stall out now refund their credits automatically. No more waiting on support for a stuck generation.

Seedance 2 Reference now checks your reference video length up front, so you find out before spending credits instead of after a failed run.

Turning audio generation off on Seedance 2 Reference no longer crashes the page. The same fix covers the Keep Audio, Captions, and prompt toggles on Kling Video Edit and the avatar tools.

Reality Polisher now scores your image with multiple independent AI checks instead of one. Obvious flaws like an extra finger no longer slip through with a perfect score.

Each polishing pass now fixes the exact flaws found in your image instead of applying generic corrections. Results land noticeably harder.

Pick your image's medium, from photo to anime to watercolor and 40 more, or let auto-detect lock it in. Either way it stays fixed for the whole run so your art style never drifts toward realism.

Polishing rounds are faster, scores are fairer for images without people, and the best result so far is always the starting point for the next round.

Refreshing mid-run no longer loses finished iterations, and stopping a run no longer shows a false failure.

When a generation gets blocked by content moderation, you now see the actual guidance instead of a generic failure message.

The home page got a tune-up. The Start Creating button now greets you right at the top on every device, and the page loads lighter than before.

On phones, the Sign Up button is now always in reach in the top bar.

The home page is also much shorter and easier to scan. Model deep dives now live on the AI Images, AI Videos, and LoRA Training pages where they belong.

More speed-ups under the hood. The home page sends less code to your browser, the FAQ and footer appear instantly instead of loading in, and the video showcase on the AI Videos page now switches clips without a blank flash.

Continuing a saved video on WAN 2.2 now brings its LoRAs along. Click First frame or Last frame on a saved video and the LoRAs you made it with are re-applied automatically, with the same strengths. No more re-picking them by hand.

Character Studio is your new home for character work: keep one character consistent across images, build multi-view turnaround sheets, try outfits on them, animate them into talking avatars, and train a custom model of them. Find it in the top menu.

Open any image or video in My Media and tap Recreate to jump straight into the tool that made it, with the prompt already filled in. Creations made with a LoRA bring the LoRA along too, so you can remix in one tap.

The image detail view got a cleaner, calmer look that gets out of the way of your art.

The main menu now always opens to the Home tab, so you land in the same familiar spot every time. Quick Access also moved up above Recent, putting your go-to destinations right at the top.

On video tools, hover any saved or community video and you will see First frame and Last frame buttons. Click one to pull that frame into the composer as your start image, then keep the video going from there. The easiest way to pick up a clip right where it left off.

The new Character Sheets tool turns one or more reference photos into a full multi-view turnaround: front, 3/4, side, and back views of the same character in one image. Add several references for even stronger consistency. Perfect for animation reference, model sheets, and building training sets.

Choose photorealistic or anime, and full-body or portrait framing. Pick from Nano Banana Pro, Nano Banana 2, WAN 2.7, or Qwen Image 2 Pro to match the look and cost you want.

Save an image or video made with a LoRA and My Media now remembers which LoRAs you used, showing them right on the saved creation.

Consistent Characters now includes Grok Imagine, Flux Kontext Dev, WAN 2.7, and Seedream 4.5. Pick the look and speed you want while keeping the same face across every variation.

Consistent Characters now lets you set the aspect ratio of your portraits. Choose square, portrait, landscape, wide, or tall so your character sets come out in the shape you need. Available on every model that supports it.

Every image and video in the gallery now has a Create your own button. Tap it to jump straight into the exact tool that made it, with the prompt already filled in and ready to tweak.

The leaderboards got a full redesign — a cleaner look with a top-three podium and a clear ranked list. Every entry now shows its likes, reuses, and saves at a glance.

Popular now ranks by a combined score: likes, plus how often a creation is reused, plus how often it's saved. The rankings reflect everything the community loves, not just views.

A new Most Saved board highlights the creations people bookmark to their favorites the most.

Google Gemini Omni Flash is here. Turn a text idea, a photo, or even an existing video into a short clip with built-in sound, no separate audio step. Pick your length and shape and go.

Bring a still photo to life, or remix a clip you already have. Drop in an image to animate it, or hand it a video and describe the change you want.

Love a clip but want a tweak? Hit Continue on any Omni Flash result and just describe the change. It builds a fresh take on that clip, no re-uploading.

Nano Banana 2 Lite gives you a faster, lower-cost way to create images at 2 credits each, sitting right next to the full Nano Banana 2 model.

In a tool's Saved tab, you can now make any creation public with a single tap. It's submitted to the community gallery through the same quick review as the My Media page, then appears for everyone to discover.

The Saved and Community tabs now link straight to your full My Media library and the full community gallery, so you can jump to the big view in one click.

The history, saved, and community tabs in each tool now do more than reuse a prompt. On tools that edit or take a reference, tap any creation to drop its image or video straight into your input, so you can build on a result in one click.

Star the creations you love and turn on the Favorites filter to see just those in your tabs. Your go-to references and best results stay one tap away.

Leave a comment on community creations right from the tool you're working in, without heading over to the main gallery.

Set up a public profile and your gallery posts show your name instead of appearing anonymous. A quick reminder now appears on the gallery so you can add yours in seconds, and people browsing the community can discover more of your work.

The account menu is clearer about which is which. 'Manage Login' covers your sign in details, and 'Manage Profile' takes you straight to your public profile and gallery settings.

Your negative prompt is where you tell the AI what to leave out. Listing things to avoid there could sometimes trip a content warning and stop your generation before it started. Now it doesn't, so you can spell out what to skip freely and your work runs right away.

WAN 2.2 Image now only offers the resolutions it can actually produce. Picking a higher one used to fail right at the start. The choices match what works, so your image generates the first time.

WAN Video Character Swap now works with saved and shared videos, not just fresh uploads. It reads the clip length for you, so you no longer hit a confusing error when you pick a video from your library.

If you build with our API, you can now check your remaining credit balance with a single request. Your apps and agents can confirm there's enough to run before they start, and prompt a top-up when it's getting low.

The Community tab on each tool now shows the full set of example creations, with no repeats. Some examples were being skipped or shown twice, so you couldn't always see the full range of what a tool can do.

Every tool now shares the same streamlined layout. Your prompt and settings sit in one tidy bar at the bottom, and your results take center stage above it, so the whole site feels consistent no matter which tool you open. This now covers the video and avatar tools too, including the talking-avatar tools that take a voice clip.

Adding an image now works the same on every tool. One button opens a single window where you can drag in a file or pick from your media, instead of two separate buttons doing two different things. The same clean picker shows up across the whole site, including tools that take a video too.

When a generation can't start because of something you can fix, like a prompt that runs too long, you now get a clear message that says exactly what to change and the limit to stay under, instead of a vague 'validation failed.' Same helpful wording everywhere, including Canvas.

The Canvas model picker now shows a quick icon next to each model, the same ones you see in the main menu. A pencil means the model can edit images and a speaker means it makes sound. Spot the right model at a glance.

The menu on your phone now opens on the first tap. Picking a tool from the menu sometimes needed two taps. One tap takes you straight there.

More tools stay responsive while they work. Across the image tools, the Generate button no longer locks up, so you can line up your next image without waiting on the last one to finish.

Generations are steadier when a model is briefly busy. If the backbone has a quick hiccup, we now retry for you behind the scenes instead of bouncing your request back, so fewer creations end with a 'please try again.'

Big reference photos just work in Seedream 4.5. A very high resolution image used to get turned away for being too large. Now it is automatically sized to fit, so your upload goes straight through.

Generating more than one image at a time with Seedream 4.5 works again. Choosing 2, 3, or 4 images had quietly stopped the Generate button from doing anything. Pick any count and create away.

When a generation can't start yet, for example a prompt that needs adjusting, the tool now keeps you on your settings instead of jumping to an empty results view. You stay right where you can fix it and try again.

Your Agent can now make talking videos. Give it a portrait and a script and it brings the face to life, lip-synced to your words. Try it with /avatar, or just ask it to make someone talk.

Add a voiceover without leaving the chat. Type /voice and what you want said, and the Agent returns natural-sounding narration you can play, download, and drop into your videos.

Need a soundtrack? /music turns a quick description into a short, original track. Set the mood, pick a length, and play it right in the chat.

Videos in the community gallery now play correctly when you open them. Clips from our newer video models were showing a still frame instead of the player. They play right away again.

Canvas has a cleaner, more focused look. Your work now fills the screen front and center, and everything you need sits in one simple bar at the bottom: the step you're on, the model, settings, and Generate. A slim row of your steps sits just above it, so you can jump back to any result with a tap.

Save and download now live right on each result, so finishing a piece is one tap. The same simple bar works on any screen, so creating on your phone feels just like desktop.

Starting a fresh canvas is now a clear, visual choice: pick Generate Image, Edit, Upscale, or Create Video to jump straight in, each shown with an example of what it makes.

Rodin is here, a new way to turn images into real 3D models. Add one to five reference photos of the same subject and Rodin builds a textured 3D mesh you can spin, zoom, and download. Great for game assets, product visualization, and 3D printing.

Choose Fast mode for quick prototyping or Standard mode for higher production quality, and switch on HighPack for sharper 4K textures when you want extra detail.

Krea Image is here, a flagship aesthetic image model that art-directs itself. It turns a short prompt into a polished, stylish image in seconds. It comes in three tiers from one tool: Turbo for fast, cheap generation, plus Medium and Large for higher fidelity with a creativity dial and style reference images.

The dashboard and the tools menu now lead with our two easiest starting points: the all-in-one AI Canvas and the AI Agent. Jump straight into either, or pick a specific tool from the menu. New here? Start with one of these and you can make something in seconds.

The PixelDojo Agent has a fresh start screen with a gallery of real looks you can make. Tap one to load it and make your own, or jump straight into editing, video, and product photoshoots. Just describe what you want and it makes it.

The start screen now leads with what the Agent can make: Create Image, Edit, Make Video, and Ads and Marketing, each shown with a real example and the video card actually plays. One tap drops you straight into that kind of project.

The Agent is much nicer on phones. Bigger tap targets, a cleaner input that stays clear of the keyboard, and generated results you can save or download right where they appear.

A fresh Canvas now shows you how it works at a glance. Generate an image, then edit, upscale, and animate it, with each result flowing into the next tool. Tap a step to jump straight to it, or pick a one-click starting prompt with a model already loaded, so you can make something in seconds.

Canvas can now chain more of your favorite models. Qwen Image 2 Edit and Qwen Image Edit for instruction-based edits, WAN 2.7 and WAN 2.2 image-to-video for animating a still, and Z Image for generation, all available right inside a Canvas session.

Canvas now opens with the most-used model for each step already selected, so you can just type a prompt and generate. The model picker lists models by how popular they are, so your top choice sits right at the top, marked Popular, instead of alphabetically.

Canvas works much better on phones. The Generate button stays visible when the keyboard is open, and the model picker opens as a tall, scrollable sheet so you can see far more models at a glance.

Seedance 2 image-to-video can now take both a start frame and an end frame. Add both and your clip animates from the first image to the last — perfect for precise transitions and morphs. The end frame is optional; drop in just a start frame for regular image-to-video.

Seedance 2 can now generate true 4K video (3840×2160). Pick 4K on the High quality tier for text-to-video, image-to-video, and reference-to-video — and on the Standard tier in Seedance 2 Video Edit. Fast still tops out at 720p.

4K runs at a higher rate than 1080p — about 23 credits per second, or 27 when you include a reference video — so you can budget before you generate.

Extending a clip with WAN 2.7 Video now picks a valid output length for you and keeps it longer than your source, so the whole clip is included. If your source clip is over the 10-second limit, you're told up front — instead of the generation running, failing, and refunding partway through.

Seedance 2 reference-to-video now checks your reference video's length before you're charged and shows the 15-second limit clearly, rather than failing after submission.

Uploading an unsupported image format (HEIC, AVIF, or GIF) now shows a clear message asking for JPG, PNG, or WebP right away — instead of letting a generation run, fail, and refund.

When a video you add to a Seedance video tool can't be used, the error now tells you the real reason and limit — frame size up to 1080p, 24–60 fps, MP4 or MOV — and what to change. Before, a rejected video was mislabeled as an image with the wrong size rule.

Reference image errors now show your file's actual dimensions, so you can see exactly how far outside the 300–6000px-per-side range you are before resizing.

The API model catalog is easier to browse — filter by output type (Image, Video, Audio, 3D) and by what each model does (text-to-image, image editing, text-to-video, image-to-video, upscaling) to find the right model fast.

Audio model previews now play. Text to Speech and Text to Music examples were showing a blank video player on their model pages — they now appear as a proper audio player you can listen to.

More models now show real example outputs on their API pages, so you can see what they produce before you integrate.

Happy Horse 1.1 is here. On the Happy Horse video and reference-to-video tools, a new Model selector lets you switch between 1.0 and the newer 1.1 engine for text-to-video, image-to-video, and reference-to-video — at the same 720p/1080p and 2–15 second durations.

Happy Horse 1.1 costs less to run — 3 credits per second at 720p and 4 credits per second at 1080p, down from 4 and 6 on 1.0. Pick 1.1 from the Model selector to pay the lower rate.

Boogu Image is here — a bilingual (English & Chinese) text-to-image model with crisp detail and 2K output. Generate up to 4 images at once.

Boogu Image also does instruction-based editing — give it a source image and describe the change you want, right from the same page.

When you upload several images at once into an editing or reference tool, all of them now appear as cards in your workspace. Previously only the first showed up even though the rest were still attached to your edit. Fixed across the image editors — Flux 2 Flex, Flux Studio, Google Nano Banana, GPT Image (1.5 and 2), HiDream, Luma, Qwen Image 2, Seedream 4, and WAN 2.6 / 2.7 Image.

When a WAN 2.7 Video generation can't run because of your input — an image that's too large, an audio clip that's too long, or an unsupported size — you now get a plain-English message telling you exactly what to fix, instead of a generic error code. If a generation fails, your credits are still refunded automatically.

When you call a specific model variant through the API by its own name, it now charges that variant's own credit price and generates with that exact variant — instead of quietly falling back to the family's default. Fixed across FLUX 2 (Klein 4B/9B, Pro, Max, Dev, Dev + LoRA), Qwen Image and Qwen Image 2, Recraft V4.1 (Standard/Pro in raster, SVG, and utility), Hailuo (Standard/Fast), Google Gemini and Nano Banana, Kling Motion Control (Standard/Pro), WAN 2.6 and WAN 2.2, and GPT-Image 1.5 (Low/Medium/High). For example, FLUX 2 Klein 4B now charges 0.1 credits instead of 1.5. You'll pay the right amount and get the model you picked.

PixelDojo has a new logo — say hello to our dragon. You'll spot the refreshed mark across the site and in your browser tab.

Every tool that takes multiple reference images now lets you pick several from your library in one go — open the picker, check the ones you want, and hit "Add" — instead of adding them one at a time and reopening the picker for each. This covers the reference-to-video tools (Seedance 2 Reference, Kling Reference to Video, Grok Imagine R2V, Happy Horse R2V, WAN Reference to Video, Veo 3.1), the video editors (Seedance 2 Video Edit, Kling Video Edit, WAN Video Edit), and the image editors (Seedream 4, GPT Image 2, Google Nano Banana, Qwen Image 2, WAN 2.7/2.6 Image, and more). Dragging in multiple files at once works too.

Seedance 2 Reference lets you batch-select multiple reference videos the same way.

When editing an image with WAN 2.7, choosing the 1:1 aspect ratio now returns a true square. Previously it kept the source image's dimensions instead of cropping to square.

Image editing now has an "Original" aspect-ratio option that keeps your input image's proportions — and it's the default for edits, so your results match your source unless you pick a specific ratio.

Seedance 2, Seedance 2 Reference, and Seedance 2 Video Edit now work with photos of real people. Use a real face as your start frame or in your reference images and your video generates instead of getting rejected — great for character work, cameos, and bringing a specific person into a scene. Your uploads are still checked for safety; we've just removed the overly strict block that was turning away ordinary photos of people.

In the Recent generations panel, each item's remove button is now a red trash icon, clearly separate from closing the panel — so reaching for the corner to dismiss it no longer wipes out your latest result by mistake. The close button at the top right is now the obvious way out. Removing an item still only clears it from this list; your saved files are never touched.

A few tools in the navigation menu were missing their icons — they now display correctly.

Ideogram 4 is here — text-to-image with best-in-class text rendering, stunning realism, and bold creative designs. It's the one to reach for when the words have to be crisp and correct: posters, logos, branding, advertising, book covers, and product mockups. Pick the tier that fits your job — Turbo (1 credit) for fast iteration, Balanced (2 credits) as an everyday default, or Quality (3 credits) for hero, print-ready work. Generate at high resolutions across square, portrait, landscape, and ultrawide shapes, or let Ideogram choose the best fit for your prompt. Find it under Generate Images.

Ideogram 4 is built into Canvas too, so you can chain it with edit, upscale, and animate in a single session — and all three tiers are available on the developer API.

MAI Image is here — a new text-to-image model with standout prompt adherence and natural lighting. It renders exactly what you describe with clean, photoreal detail, making it a dependable all-rounder for product shots, editorial portraits, food, interiors, and concept art. Choose from 11 aspect ratios (1:1 through 21:9, or 'auto' to let the model pick the best fit), generate up to 4 at once, and export PNG, JPEG, or WebP. 1.5 credits per image — find it under Generate Images.

A full MAI Image prompting guide with real example galleries — product, portrait, cinematic landscape, food, and concept art — is live under Guides.

WAN 2.6 Image now generates at its full, correct resolutions with a wider set of aspect ratios — square, portrait, and landscape, plus a true 21:9 ultrawide. Pick the exact shape you need and get properly sized output.

Editing with WAN 2.6 now keeps your original image's proportions by default, so your edits come back the right shape instead of being cropped or stretched.

Added a seed option to WAN 2.6 Image for repeatable results — reuse a seed to get consistent variations of the same prompt.

Grok Imagine Video now has a 1.5 model for image-to-video — animate any photo and it creates matching audio (music, sound effects, and ambient sound) in the same pass, no separate editing needed. Just switch to image-to-video mode and pick it from the Model option. Still 2 credits per second.

The Grok Imagine prompting guide has a new audio section for 1.5 — how to cue music, sound effects, ambient tone, and short dialogue, plus the AUDIO: block syntax, real example clips you can play with sound, and copy-ready templates. Find it under Guides.

Credits now settle accurately on instant-finishing jobs — a successful generation no longer triggers a mistaken refund, and jobs that return multiple results won't over-refund. Your balance reflects exactly what you used.

Qwen image generation and editing now finish reliably — your work no longer gets stuck spinning if we ship an update while an image is being created. Every job lands as a finished result, or refunds automatically if something goes wrong.

Image tools are now far easier to use on mobile — pick your model, aspect ratio, and number of images with tap-friendly controls, see the credit cost right on the Generate button, and on phones the controls slide up in a tidy composer drawer so the screen stays focused on your work.

This rollout reached a dozen-plus image generators, upscalers, and editors, so they all look and behave the same way no matter which one you open — no more relearning the controls from tool to tool.

P Video Animate — motion transfer for still images. Upload a reference video and a character image, and it copies the video's motion, timing, and camera moves onto your character. Under Generate Videos. 1 credit/sec at 720p, 2 credits/sec at 1080p.

Flux Pro Virtual Try-On — premium, high-fidelity try-on under Characters & Faces. Dress a person photo in any garment, with an optional styling note ("tucked in", "open over a tee"). Preserves pose, lighting, and background. 2 credits per result.

Krea V2 (Medium + Large) — text-to-image with a creativity dial (raw → high) and style reference images. Medium at 1 credit for fast iteration, Large at 2 credits for the highest fidelity, across eight aspect ratios.

Riverflow 2.0 now generates cleanly on every aspect ratio at 1K, 2K, and 4K — the ratios that used to error out are fixed.

Mobile tool pages: the Generate button stays visible and tappable, new generations scroll into view automatically, and recent generations moved into the My Stuff menu to keep the screen clear for creating.

Fresh prompting guides with real example galleries for Krea V2, Flux Pro Virtual Try-On, Riverflow, Luma UNI 1, and P Video Animate — all under Guides.

Riverflow 2.0 — Fast + Pro in one tool. Fast at 1 credit (1K/2K) for cheap iteration. Pro at 3 credits (1K/2K) or 7 credits for full 4K — the only model on the site shipping 4K output right now. Text-to-image with optional reference image to steer style and composition (the model writes a fresh frame from your prompt; references aren't modified). Find it under Image Generation.

Luma UNI 1 — Standard + MAX in one tool. Standard at 1 credit for fast iteration, MAX at 3 credits for the quality-first variant. ~3K resolution across 7 aspect ratios, up to 8 reference images per edit, optional manga style preset. Also under Image Generation.

P-Image Upscale — sub-second upscaling up to 128 megapixels. From 0.1 credit (4 MP) up to 3 credits (128 MP). Toggles for detail boost and an AI-image realism pass that cleans up plastic-looking outputs. Find it under Upscale & Enhance.

Searching the Prompt Library and then clicking the Image or Video filter chip was kicking users to an error page. The filter now applies cleanly on top of any search — pick a query, narrow by type, and the matching prompts show up the way they should. Same fix covers tag and model combinations along the way.

Sourceful Riverflow 2.0 is now live in the Image Generation section. One tool, two quality tiers: Fast is 1 credit per image (1K or 2K). Pro is 3 credits per image at 1K/2K and 7 credits at 4K — the only tier on Pixel Dojo right now that ships full 4K output.

Click any image in your history to use as a reference — Riverflow uses it to steer style and composition while generating a fresh frame from your prompt. (It's reference-guided generation, not inpainting/editing — the source image isn't modified.) Ten aspect ratios available: 1:1, 4:3, 3:4, 16:9, 9:16, 3:2, 2:3, 5:4, 4:5, and 7:3 (ultra-wide).

The resolution picker is smart about the model you're on — 4K is unavailable while Fast is selected (Fast doesn't ship 4K), and switching from Pro 4K back to Fast snaps you to 2K automatically instead of leaving an invalid combo. Cost preview updates live as you change model or tier so there's no math required.

Luma UNI 1 is now live in the Image Generation section. One tool, two quality tiers: Standard is fast and clean at 1 credit per image, MAX is the quality-first variant at 3 credits per image. Pick the tier in the form, the cost updates live as you change it.

Both text-to-image and image editing in the same page. Click any image in your history to add it as a reference, then describe the edit — Luma supports up to 8 reference images per generation. Ships at ~3K resolution across 1:1, 16:9, 9:16, 3:2, 2:3, 2:1, and 1:2. Optional 'manga' style preset for the look you'd expect.

The in-app 'What's New' modal was clipping the last item or two behind the Got it footer on shorter screens — laptops, smaller windows, and phones in particular. The modal now keeps comfortable breathing room above and below regardless of viewport size, the footer always stays in view, and mobile browser chrome (Safari's address bar in particular) no longer pushes anything off-screen.

When there's more content below the visible area, a bouncing chevron appears at the bottom of the modal alongside a soft fade — no more guessing whether you've seen everything. The cue disappears once you've scrolled to the last item.

The modal was rendering two close buttons stacked in the corner. Now there's one.

P-Image Upscale is now live in the Upscale & Enhance section. Sub-second image upscaling to any target size up to 128 megapixels. Pricing scales with output size: 0.1 credit for 4 MP, 0.5 credit for 16 MP (≈4K), 2 credits for 64 MP, 3 credits for 128 MP. Toggles for Enhance Details (sharper textures) and Enhance Realism (cleans up AI-generated images that look too plastic).

The Upscale & Enhance lineup now has three clear lanes: P-Image for fast and cheap bulk work, Clarity Pro for photorealistic with identity preservation, and Magnific for creative/premium. Pick the right one for the job instead of paying premium prices on every upscale.

When we ship a new version of Pixel Dojo while you have it open, the old page used to silently reload itself a second or two after you next clicked something — which sometimes ate the video or image you just started generating. It now shows a quiet 'A new version of Pixel Dojo is available' banner instead. You choose when to refresh, so nothing in flight gets dropped.

The app now checks for new versions in the background and surfaces the upgrade banner the moment a deploy lands, instead of waiting until you click something that tries to load a missing chunk. Less surprise, no broken interactions.

WAN 2.7 Video now reserves a slot in your history the instant you hit Generate, before the network call lands. If anything interrupts the dispatch — a reload, a network blip — the job is no longer 'lost'; it shows up as 'Preparing…' and either resolves or clears itself cleanly. Other generation tools get the same protection as we roll them over.

Qwen Image, Qwen Image 2, and Kling Image no longer deselect your source images the moment you hit Generate. The images you picked stay selected so you can keep iterating on the same shot — tweak the prompt and hit Generate again without re-picking everything. Use the Exit edit mode button (or Clear history) when you actually want to start fresh. Matches how the other edit-capable models already behave.

The looping hero video at the top of the home page no longer hangs on a loading spinner — it loads and plays as soon as the page is ready. Underneath, we also hardened the player so any future bad video URL falls through to a Play button instead of silently spinning.

The Agent now reads what your prompt is actually about — product shot, portrait, logo, cinematic, anime, character — and picks a model that's good at that flavor. Before, every text-to-image prompt routed to whatever was most popular overall, which meant a logo request and a product photo both landed on the same generic default. Now a logo goes to a text-rendering-strong model, a product shot goes to a product-photography model, and so on.

Model recommendations now use site-wide usage data (every surface — chat, Canvas, dashboard, API, MCP) instead of Canvas-only data. The top models reflect what creators across the whole site converge to, not just one surface. A single heavy API caller can't dominate the rank either — each user's contribution is capped at 5% of total signal.

When the Agent picks a model for you, you'll see WHY in the response — 'used seedream-4 because it's tuned for product photography; also good: nano-banana-pro, gpt-image-2'. No more black-box defaults.

Every Agent result now shows a 'Try next' chip strip — Animate, Add effect, Photoshoot, Upscale, Edit again — that one-click chains the next skill onto your current image. Pre-fills the composer, attaches the asset for you, and the inline mode/effect picker pops automatically. No more remembering what / commands exist or what step comes next.

Brand-new Agent chats now show a categorized starter grid (Image / Edit / Video / Marketing) with 8 ready-to-send prompts spanning every skill — including the newer /photoshoot, /effect, and /hooks paths that were easy to miss. The /hooks card carries a 'Free' badge so you can experiment with paid-social hooks without thinking about credits.

Campaign tool card was showing a fake 'Prediction not found' error on campaigns that were actually still rendering. Wrong job-tracker path was sending the campaign ID to the image-prediction lookup, which 404'd and surfaced as a red error. Campaigns now poll their own status endpoint, show partial progress as assets land ('Campaign running — 3 of 5 assets ready'), and only flip to failed if every sub-job genuinely fails.

If you just generated a video, then asked the Agent to 'edit this' or 'animate it', the .mp4 URL was getting passed as a still-image reference — which silently 422'd at the model with 'cannot read image'. The Agent now walks back to the most recent IMAGE asset in your history, skipping videos, since every edit / storyboard / upscale / character / photoshoot / effect skill consumes a still.

WAN 2.2 (Standard, Plus, Extended) was rejecting public-API calls that didn't explicitly include a 'mode' field with 'mode: Required'. Now infers the mode from your input: pass an image_url and it picks image-to-video, otherwise text-to-video. Same convention every other multi-mode video model uses.

Public /pricing page now shows three tiers side-by-side. Lite ($10/mo, 160 credits) for casual creators, Pro ($25/mo, 400 credits) for working creators, Pro Max ($50/mo, 800 credits) for power creators. Every tool, every model, every new launch — included on every plan. Pick the volume that fits how much you create.

Pro Max is here for the creators who keep buying $50 credit packs one at a time. Same 800 credits, but as a subscription — predictable monthly billing, no more 'I should top up before tomorrow's deadline'.

Every plan unlocks the whole studio. Lite gets the same 70+ models, LoRA training, Canvas, and developer API as Pro and Pro Max — you're choosing how much you create per month, not which tools you get. Pro keeps its 6-month and annual billing options behind a small disclosure below the card so they don't compete with the tier comparison.

Mobile /pricing layout stacks the three cards with Pro on top to preserve the anchor — same hierarchy you see on desktop, just vertical. No squishing the 'Most Popular' card off-screen.

SEO + page meta on /pricing now lead with 'Plans from $10/mo'. Schema.org markup includes all three tier offers so search engines can show the right starting price.

Pixel Dojo Lite is here — $10/month for 160 credits. Same per-credit price as Pro ($0.0625/credit), just a smaller plan. Currently surfaced as a save option when you go to cancel and pick 'It's too expensive' — coming to the public pricing page soon.

Cancel flow now leads with real save options when you pick 'It's too expensive'. Instead of just a 'Manage plan' button that punted you to a portal, you now see a clear sapphire-glow downgrade card showing $25 → $10 with side-by-side credit counts (160 vs 400), then Pause as the secondary option, then a smaller 'Manage plan' link. Hierarchy you can scan in two seconds.

Downgrade now has a proper email flow. Click 'Switch to Lite' and you get an immediate confirmation showing exactly when your current plan ends — plus a reminder email at that date with a one-click link to start Lite. Your trained models, LoRAs, and media library stay intact through the transition.

Modal stays open with a confirmation state after you schedule a downgrade — 'Your Pro stays active until [date]' — instead of closing instantly with a fading toast. Easier to absorb what just happened and what comes next.

Pause and Downgrade are now mutually exclusive so you can't accidentally do both. If you're already paused, the downgrade card shows 'Resume to downgrade'; if a downgrade is scheduled, the pause card shows 'Cancel pending downgrade first'. The blocked option stays visible so you know it exists.

Click the credit number in the top bar to open a quick view of your balance, this month's credits used, and your recent refunds. No more digging through menus to see where credits went or whether a failed generation actually got refunded.

Separate Buy Credits button now sits next to your balance, so topping up is always one click away — even when you're deep inside a tool or Canvas session.

Small green dot lights up on the credit pill whenever refunds have landed that you haven't seen yet. When a generation fails or times out we put the credits back — now you actually know it happened.

Dashboard now shows a monthly usage card right after Recent Creations: credits used, credits refunded, your top tools by spend, and a one-click link to the full report.

Refund notifications now fire as soon as a generation fails — see the credits come right back in real time. Multiple refunds landing at once bundle into a single notification so you don't get spammed.

MCP skills handle missing inputs more gracefully. The character skill now returns a clear message when called without a reference image on edit-style models (instead of looping with a cryptic backend error), and the generate skill correctly maps parameters for grok-video-extend.

Magnific / Creative Upscale at very high scale factors now returns a clear, actionable error when the output exceeds size limits — telling you exactly what to try (lower scale factor or smaller source image) instead of a raw parse error.

API access now works with credits alone — no active subscription required. If you ever saw 'An active subscription is required to use the API' from /api/v1/generate even though you had plenty of credits (or paid via a credit pack instead of a recurring plan), that error is gone. Behavior now matches what /api-platform/documentation has always said: signed-in users can create an API key and use it as long as they have credits.

Twenty-one tools picked up an in-tool prompting-guide banner today — open one and you'll see a small 'Read the guide' card at the top of the settings sidebar (or in the empty state) linking straight to that model's dedicated prompting guide. Video side: Runway Gen-4.5, LTX 2.3, Pixverse V6, Pixverse v5.6, Seedance 1, WAN 2.5, P-Video. Image side: Hunyuan Image 3, Dreamina 3.1, Ernie, PonyXL, WAN Image, ImagineArt, Qwen Image, P-Image, WAN 2.7 Image, WAN 2.6 Image, FLUX Studio, GPT-Image 1.5, Z Image Turbo, FLUX 2 Flex. The guides have been live at /guides/* for a while — they're just one click away from the tool page now.

FLUX Studio prompting guide at /guides/flux-studio-prompting-guide — documents the every-FLUX-1-variant workspace (Schnell, Dev, Krea Dev, FLUX Pro, 1.1 Pro / Pro Ultra, Kontext Pro / Max). Four worked examples plus a tier-selection guide for matching variants to use cases.

WAN Image prompting guide at /guides/wan-image-prompting-guide — the fast Replicate-backed WAN image variant. Documents the optional LoRA fal-lora variant and how it pairs with WAN 2.7 Image for hero shots. Four examples covering portrait, landscape, product, and stylized.

WAN 2.6 Image prompting guide at /guides/wan-2-6-image-prompting-guide — intermediate-gen WAN with the distinctive fixed pixel-dimension preset system (1280×1280, 1280×720, 768×1024, etc) vs free aspect ratios. Useful when you need exact downstream output dimensions.

P-Video prompting guide at /guides/p-video-prompting-guide — Pruna's three-mode video model (text-to-video, image-to-video, audio-to-video). Documents the cost-efficient tier choice and the audio-to-video mode for music-driven motion and lip-sync work.

Ernie Image prompting guide at /guides/ernie-prompting-guide — Baidu's image flagship with multilingual prompt support (English, Chinese, Japanese) and native CJK text rendering. Four worked examples spanning a Chinese tea master portrait, Shanghai Pudong skyline, Chinese calligraphy sign, and traditional garden pavilion.

FLUX 2 Flex prompting guide at /guides/flux-2-flex-prompting-guide — the resolution-flexible FLUX 2 variant where you pick 0.5/1/2/4 MP per generation. Documents how to iterate cheap at 0.5 MP and finalize at higher MP, plus how it relates to FLUX 2 Pro / Max / Dev.

Pixverse v5.6 prompting guide at /guides/pixverse-prompting-guide — the dependable previous-generation Pixverse workhorse. Quality tiers from 360p to 1080p, optional audio, predictable motion. Four examples covering glassy lake sailboat, beach character walk, fountain pen signature, and skater handrail grind.

WAN 2.5 prompting guide at /guides/wan-2-5-prompting-guide — fast image-to-video flash variant in the WAN family. Documents how the speed-optimized I2V tier compares to the premium WAN 2.7. Four worked examples spanning rice terrace drone, maple forest character walk, ink-drop macro, and sunset surfer vertical.

PonyXL prompting guide at /guides/ponyxl-prompting-guide — documents the score-tag quality system (score_9, score_8_up, score_7_up that you must lead every prompt with) and booru-style tag prompting conventions. Four worked examples across anime portrait, fantasy knight, chibi cat wizard, and anime mecha.

Z Image Turbo prompting guide at /guides/z-image-turbo-prompting-guide — covers what's distinctive about Z Image Turbo's synchronous response (URL returns inline, no polling round-trip needed), and why that matters for high-volume programmatic workflows. Four production examples.

WAN 2.7 Image prompting guide at /guides/wan-2-7-image-prompting-guide — the image-side sibling of WAN 2.7 Video. Documents Standard/Pro tier choice and how the same prompt language carries between WAN Image and WAN Video for consistent aesthetics across modalities.

P-Image prompting guide at /guides/p-image-prompting-guide — at 0.1 credits per image it's the cheapest image option on the platform. Documents the moodboard / high-volume use case where you generate 30-50 variations for ~3-5 credits total, then upscale the keepers via a premium tier.

Hunyuan Image 3 prompting guide at /guides/hunyuan-image-3-prompting-guide — Tencent's 80-billion-parameter image model. Four worked examples spanning painterly fantasy, photoreal cultural portrait, Brutalist architecture, and mechanical watch macro. Covers the Fast/Quality toggle.

Google Gemini Image prompting guide at /guides/google-gemini-image-prompting-guide — Google's image flagship with three tiers (Standard at 1 credit, Pro at 3-6) and unique native 4K support. Examples for cinematic 21:9 landscape, documentary portrait, food editorial at 4:5 Instagram-spec, and stylized illustration.

ImagineArt prompting guide at /guides/imagineart-prompting-guide — three model tiers (2.0 preview, 1.5 Pro, 1.5) in one workspace, nine aspect ratios including the unusual extremes 1:3 (ultra-tall mural) and 3:1 (ultra-wide banner). Four worked examples cover cinematic, fashion editorial, pop art, and ultra-tall mural composition.

Grok Imagine Video prompting guide at /guides/xai-video-prompting-guide — xAI's video model. Four examples (Mojave Mustang, Irish cliff character, chocolate truffle macro, snowboarder vertical) covering cinematic motion, atmospheric scene composition, and 480p/720p tier choice.

Qwen Image prompting guide at /guides/qwen-image-prompting-guide — covers the Plus / Max tier choice, snapshot pinning for campaign consistency, and clean text rendering in Latin and CJK scripts. Four worked examples including a multi-line bookshop sign with proper kerning.

Dreamina 3.1 prompting guide at /guides/dreamina-prompting-guide — 4MP cinematic image generation with atmospheric defaults, flat pricing across 1K and 2K, and custom dimensions (256–2048 each axis) for spec-size output. Four worked examples covering architecture, food editorial, fashion portrait, and atmospheric fantasy.

Pixverse V6 prompting guide at /guides/pixverse-v6-prompting-guide — documents the unique multi-clip mode (first frame + last frame generates the transition between them), the six built-in style presets (anime, 3D animation, clay, comic, cyberpunk, none), and the moderation quirks worth knowing.

Seedance 1 prompting guide at /guides/seedance-1-prompting-guide — smooth natural motion, six aspect ratios including the standout 21:9 ultra-wide cinematic, native audio with sound-cue steering, and the end-frame interpolation mode for motion between two anchor images.

GPT-Image 1.5 prompting guide at /guides/gpt-image-1-5-prompting-guide — covers literal prompt adherence, in-image text rendering with quoted strings, transparent-background PNGs (real alpha channel, no cutout pass needed), and when to reach for it over FLUX or Seedream. Five worked examples spanning chalkboard typography, editorial illustration, macro photoreal, transparent logo, and cinematic landscape.

Ideogram Character prompting guide at /guides/ideogram-character-prompting-guide — how to get consistent character variations from a single reference image, no LoRA training required. Includes a worked example of one character pushed across fantasy armor, anime, film noir, and sci-fi astronaut styles. Reference-image quality rules included.

Runway Gen-4.5 Video prompting guide at /guides/runway-gen4-video-prompting-guide — cinematic motion, camera-language fluency, six aspect ratios from 21:9 cinematic through 9:16 vertical. Four worked examples (aerial drone push, character tracking shot, product macro, vertical reel) plus the gotcha that 1:1 aspect requires image-to-video mode.

LTX 2.3 prompting guide at /guides/ltx-2-prompting-guide — the broadest-surface video model on the site. Covers the Fast vs Pro tradeoff, durations up to 20 seconds, resolutions through 4K, optional native audio, and three modes (text-to-video, image-to-video, video extension for 30+ second pieces). Four worked examples and a full settings table.

Eight fresh prompting guides shipped today — FLUX, Recraft V4.1, Seedream 4.5, HiDream O1, Hailuo 2.3, Kling Image v3, Kling 2.6 Pro, and WAN 2.7 Video. Each one is built from a shared format with real generated examples (not stock clips), a copy-ready prompt library, settings reference, and a model-specific FAQ. Browse them all at /guides.

Four of the eight are video model guides — Hailuo 2.3, Kling 2.6 Pro, and WAN 2.7 Video each ship with embedded sample clips you can scrub through before writing your own prompt. The shared template will speed up every video guide we publish from here.

Four image model guides for FLUX, Recraft V4.1, Seedream 4.5, and HiDream O1 — cover style ranges, aspect ratios, the model's prompt sensitivities, and what to ask for vs what to leave out. Useful whether you're brand-new to the model or you've hit a wall trying to nail a specific look.

Seedream 3 has been fully retired — the model was no longer reachable from the dashboard but a few public API and Canvas surfaces still advertised it, which meant API calls to /api/v1/models/seedream-3/run were returning service-unavailable errors. Cleaned up the catalog so what you see is what you can actually generate with. If you were using Seedream 3, move over to Seedream 4.5 or Seedream 5 Lite — both are live and stronger across the board.

Hunyuan Image 3 moved to a new backbone so we can ship it with the safety checker fully disabled at the provider — content blocks should no longer interrupt your generations. Pricing changes accordingly: 2 credits per image (was 1).

Hunyuan Image 3 settings simplified — replaced the Steps + CFG sliders with a single Fast / Quality mode toggle (Fast is the default, flip it off when you want max fidelity). Aspect ratio presets expanded from 3 to 7 (added 16:9, 9:16, 4:3, 3:4, 3:2, 2:3).

Hunyuan Image 3 is now available — Tencent's 80-billion-parameter text-to-image model with high-fidelity output and fine-grained control. 1 credit per image. Generate up to 4 images at once.

Hunyuan Image 3 exposes Steps (10–50) and Guidance / CFG (1.5–15) controls — adjust denoising iterations and how literally the model follows your prompt. Default values (20 / 7) work for most cases; nudge up CFG when you want the output closer to the prompt.

Hunyuan Image 3 ships with three aspect ratio presets — 1:1 (square), 3:1 (ultra-wide cinematic), 1:3 (ultra-tall portrait). More aspect ratios to follow once we lock down the supported pixel mappings.

Bria 3.2 prompting guide is live at /guides/bria-3-2-prompting-guide — five example prompts spanning Photography and Art mediums, plus six prompting tips for matching medium to language, anchoring quality, and getting clean product shots.

Bria 3.2's 14:17 and 17:14 aspect presets were sending unsupported pixel dimensions and erroring on submit. The pixel mappings have been corrected — those aspects now generate cleanly.

Bria 3.2 is now available — text-to-image at 1K resolution across 9 aspect ratio presets (1:1, 3:2, 2:3, 7:4, 4:7, 9:7, 7:9, 17:14, 14:17). 1 credit per image. Generate up to 4 images per prompt and pick the output format (PNG, JPG, or WEBP) before you hit go.

Bria 3.2 medium toggle — bias outputs toward Photography or Art with a single click. Useful when prompts read ambiguously and you want to nudge the result one direction or the other without rewriting the prompt.

Bria 3.2 image and prompt enhancement — optional toggles for richer detail / sharper textures and for letting the model expand your prompt with descriptive variations.

Vidu Q3 is now available — text-to-video and image-to-video across four resolution tiers (360p, 540p, 720p, 1080p) and five aspect ratios (16:9, 4:3, 1:1, 3:4, 9:16). Pick exactly the frame and resolution you need before generating instead of upscaling or reformatting later. Credit cost scales with resolution: 1 credit/second at 360p and 540p, 2 at 720p, 3 at 1080p.

Optional synchronized audio on Vidu Q3 — off by default for clean tracks you'll post-score, one toggle on when you want ambient and diegetic sound layered in with the visuals. End your prompt with a 'Sounds of …' line to steer the soundscape.

Vidu Q3 image-to-video — upload one source still and Vidu writes motion onto it from your prompt. Source aspect drives the output frame automatically, so portrait photos animate as portrait videos without reformatting.

Vidu Q3 prompting guide at /guides/vidu-q3-prompting-guide — six example prompts spanning cinematic aerials, character close-ups, product spots, multi-shot action, atmospheric wides, and vertical I2V. Plus a settings reference table and a six-item FAQ.

Image-to-video on WAN 2.7 was silently dropping the source frame in some flows — you'd attach a photo, ask it to animate, and the result would be a completely different subject. The variant didn't pin its mode internally, so a default kicked in that took the text-only branch and ignored the image. Now pinned correctly: your starting frame is honored every time, whether you call it from chat, Canvas, the dashboard, or the public API.

Audited every multi-mode video model for the same bug pattern. Same fix applied to Happy Horse 1.0 (both text-to-video and image-to-video variants) and Kling Video v3 (all four — Standard Text, Standard Image, Pro Text, Pro Image). Worst case caught: Kling Pro Image was doubly broken — would have been billed as Standard pricing AND ignored the image. All fixed.

If you've ever pasted a long, detailed prompt and gotten hit with "Prompt must be 2,500 characters or less" or similar — that's gone. Every generation tool now lets your full prompt through. If a specific model genuinely can't handle the length you'll see that model's own error message (which is more accurate than a guessed cap), but the upfront client-side rejection is no longer in the way.

Same prompt-honor fix from earlier today extended to Dreamina — its 'Enhance Prompt' toggle was on by default and silently rewriting your wording before generation. Now off by default like the rest.

Hover-info icons added to the prompt-rewrite toggles on Dreamina, Hailuo, and WAN 2.2 too. Every "enhance / optimize / expand" checkbox across the site now tells you upfront what it actually does.

Audited every generation tool for silent prompt truncation. No tool ever shortens what you typed before sending it to the model. If your prompt is too long for a particular model you'll see a visible error so you can trim — never a silent edit.

If you ever felt like WAN 2.7 (or one of the other WAN, Qwen, or Z Image tools) wasn't following your prompt — you were right. Every generation was silently running your wording through an upstream auto-rewriter that expanded and stylized it before the model saw it. That's now off by default across all of them. What you type is what the model gets. You can still turn it back on if you want the AI to embellish short prompts.

Every "Enable prompt expansion" toggle now has a hover-info icon that explains exactly what it does (the model rewrites your wording) so the trade-off is visible at decision time.

Cleaned up the default negative prompt on video tools. Removed "excessive motion" and "flickering" — those were silently fighting action-heavy prompts. Only always-harmful negatives (anatomy defects, watermarks, compression noise) remain. If you want stillness, you can still add it to your own negative prompt.

Images you generated with the SDXL LoRA tool weren't recording which LoRA produced them — so when you opened the details popover in My Media on one of those images, the 'LoRAs Used' section stayed blank even when you'd clearly picked one. The save flow was dropping the LoRA metadata on the floor. Every new save from the LoRA Image Creator now correctly stores the LoRA name and strength, and they show up in the details panel where they belong. Older saves can't be retroactively tagged, but everything going forward will be.

A new 'My Stuff' button sits next to 'All Tools' in the top bar and opens a dropdown with everything personal — Dashboard, Canvas, Agent, My Media, My LoRAs, My Prompts, Community Gallery, and your Usage Report. One click from any tool page. These destinations used to live two clicks deep inside the All Tools drawer (or buried in the avatar menu).

Your full pinned tools list and the last five tools you visited also show up inside My Stuff — so you can jump between everywhere you actually work without leaving the page you're on. The three-icon pinned strip in the top bar stays exactly where it was.

Yesterday's 'Save All' truthful-toast fix only covered the image tools. Now every video tool — Kling Video, Veo 3.1, Hailuo, Seedance, WAN, Grok Video, LTX, Pixverse, Runway, P Video, Omnihuman, and the rest — uses the same single-round-trip save with per-item results. Select 10 videos, hit Save, you get an accurate count back. Mobile-safe with keepalive enabled.

Same fix applied to LoRA Image Creator. Bulk-save reliability now matches every other tool on the site.

If you ever clicked 'Save All' on a batch of generated images and saw the success toast but didn't find them in My Media later — that was a real bug, not a glitch. The save flow has been rebuilt around a single server round-trip with truthful per-image results: the toast now says 'Saved 7 of 10 — 3 failed' if any of them didn't land, and the failed ones stay selected so you can retry without re-picking them.

Mobile (especially iPhone) was where this hit hardest. iOS Safari was killing in-flight save requests whenever you switched apps or locked your screen mid-batch. Saves now run keepalive-true and the server-side fan-out keeps everything going even if you background the tab — no more silent drops.

Recraft is now V4.1 — better photorealism, smoother gradients, and noticeably better text rendering vs V4. Same prompt understanding and design taste, just sharper across the board. Available at /recraft.

Three output variants under the same tool: Raster (standard image), Vector SVG (editable, layered), and a brand-new Utility variant tuned for product mockups — flat lighting, front-facing composition, clean scenes. Perfect for packaging, icons, and e-commerce hero shots.

Pricing simplified and cheaper across the board. All Standard generations are 1 credit (was 1–2). All Pro generations are 5 credits (was 6–8). Pick the variant you need without doing pricing math first.

Aspect ratio menu expanded from 8 to 14 options. Added 2:1, 1:2, 5:4, 14:10, 10:14, and 6:10 alongside the standard square/portrait/landscape set.

Kling Avatar used to charge the full 60-second rate up front regardless of how long your audio actually was — a 2-second clip cost 120 credits at Standard. Now the cost panel reads your audio file the moment it uploads and charges only for the real duration. That same 2-second clip is 4 credits now (or 6 on Pro).

Kling Avatar V2 is live. Upload a portrait and an audio file and get a lip-synced talking video — works on realistic humans, stylized characters, cartoons, and animals. Great for talking-head explainers, narrated product demos, and giving any character a voice without filming.

Two quality tiers: Standard for fast turnarounds (2 credits/second) and Pro for sharper facial detail and smoother motion (3 credits/second). Video length matches your audio automatically.

Optional prompt slot lets you guide actions, emotions, and camera movement — the audio still drives the mouth and timing, the prompt adds direction on top ('warm enthusiasm', 'subtle nods', 'wide shot').

Public API calls to MuleRouter-backed video tools now surface the same helpful error messages the in-app tools already do. When the model is briefly offline upstream, you'll see 'This video model is briefly offline while we top up our backbone. Your credits were refunded — please try again in a few minutes.' — not the previous internal architecture message. Applies broadly: any tool with a server-side override now bubbles its real error (size limits, format issues, transient outages) through /api/v1, not just MuleRouter models.

Type /photoshoot or /effect and a picker pops up with the named choices — hero / lifestyle / with-model / levitating / restyle for photoshoots, and dolly-zoom / 360-spin / fpv / crane-up / whip-pan / parallax / vertigo / bullet-time for effects. Each entry has a short description so you can see what it does before picking. Tab or Enter to select; the value splices into your prompt and you keep typing the scene.

Typing something that isn't a valid mode/effect after /photoshoot or /effect (like pasting a URL) used to fail with a cryptic 'invalid enum value' error after sending. The picker now shows 'No match' inline as you type so the mistake is obvious before you waste the send.

/photoshoot turns any product image into a brand-quality shot. Five modes — hero (clean studio), lifestyle (real-world setting), with-model (person interacting), levitating (dynamic floating with splashes), and restyle (same product, fresh aesthetic). Each mode bakes in the prompt scaffolding a product photographer would write — you just pick the mode and add direction.

/effect adds a named cinematic effect to a video shot — dolly-zoom (Vertigo), 360-spin, fpv (first-person drone), crane-up, whip-pan, parallax, vertigo, or bullet-time. Pass the effect name and what the camera is looking at; the camera-move prompt is handled for you.

/hooks generates 10 paid-social ad copy hooks for a product or concept — TikTok / Reels / Shorts ready. Pure text, no image generation, zero credit cost. Mixes curiosity, social proof, transformation, contrarian, and FOMO angles so you can A/B test scroll-stoppers.

/model picker now works inline anywhere in your message, not just at the start. Type /model mid-prompt to pop a filterable list of every available model — narrows as you type, picks splice into your text without wiping what you'd written.

The + button gained a 'Pick a model' shortcut and an 'Insert skill' submenu. Both insert at your cursor instead of replacing your prompt — so you can write half a sentence, click +, choose a skill or model, and keep typing.

Follow-up skills (/edit, /character, /storyboard, /upscale, /photoshoot, /effect) now automatically use your last generated image as the reference. 'animate the dragon' just animates the dragon — no re-attaching, no re-picking.

'Just make me an image' was sometimes routing to an image-edit model by mistake, which would error because no reference image was attached. The Agent now correctly defaults to a true text-to-image model when you don't supply one.

Slash commands in the Agent. Type / and pick a skill directly — /generate, /edit, /character, /storyboard, /upscale, /from_url, /campaign — to skip straight to the tool without going through the chat picker. Tab or Enter to confirm.

Pin a specific model with /model <apiId> <prompt> — e.g. /model wan-2.7-i2v animate her smiling. Great when you know exactly which model you want and don't want the Agent to decide for you.

Attach reference images two ways: 'Upload from computer' or 'From My Media' — pick any image you've already generated on PixelDojo without leaving the chat.

Attached images now render as a real thumbnail inside your chat bubble instead of a clunky '[attached image: URL]' line. Looks like a proper chat — and follow-ups like 'animate this' thread the image through automatically.

Pasting a product URL (/from_url) now renders a rich preview card — hero image, product name, description, and source — instead of a wall of plain text. Much easier to confirm at a glance that the Agent grabbed the right product before you tell it to make hero shots.

The Agent's default text-to-image pick was Dreamina even when WAN and Nano-Banana are actually more popular. Defaults now follow real Canvas usage rank instead of alphabetical order — so 'just make me an image' lands on what creators actually keep, not whichever model came first in the alphabet.

Reading a product page from a URL no longer renders the success card with a red error border. It's a successful action, so it now looks like one.

When you ask the Agent to 'animate this' on an image you uploaded or just generated, it now picks WAN 2.7 i2v by default — far more faithful to the source subject than the more creative video models, which sometimes reinterpret the image entirely.

Every Agent generation card now shows the source image it used, so you can confirm at a glance that 'edit it' / 'upscale that' picked up the right reference.

The Agent now offers the same skill set as the public /skills page — edit, product extraction from URL, and full campaign generation (hero + lifestyle + video pack from one product link) are all available in chat.

Reference images now thread through reliably when you say 'edit this' or 'animate it' as a follow-up — no more dropping the image silently. Credits also refresh in the top bar the moment they're deducted, and conversation history persists cleanly when you leave and come back.

Asking the Agent to 'make a video' with WAN 2.2 no longer errors with 'mode: required' — text-to-video vs image-to-video is auto-detected from your prompt.

Qwen Image, Qwen Image 2, and their Edit variants now accept your full prompt — we removed the 800-character soft cap and the 'we trimmed your prompt' warning. Long, detailed prompts go through unmodified. If you regularly write past 800 chars for fine-grained control, this is for you.

Meet the Agent. Describe what you want — 'cinematic samurai portrait', 'edit this to add neon rain', 'make a 6-second video of waves' — and it picks the right model from across the 70+ tools and runs it. Live at /agent from the sidebar, available on subscribed plans. Generations cost the same credits as the tools they call; the chat itself is free.

The Agent auto-tunes to real PixelDojo usage. Defaults land on the models creators actually keep for each task — text-to-image, edits, video, upscales, characters — so 'just make me an image' produces what's working today, not yesterday's pick.

Six built-in skills cover the common asks: generate, edit, character (consistent identity across shots), storyboard (video), upscale, and a status checker for long-running jobs.

Videos render in the background and appear in chat the moment they're ready — same auto-polling pattern the video tool pages use. No need to ask 'is it done yet?'.

Drag in a reference image to anchor edits, character generations, or upscales. Each completed asset shows the model that made it with Save / Download / Open full size on hover. Conversation history saves so you can resume any chat.

Right-sized the Prompt Library headline so the featured strip stays visible without scrolling on a standard desktop. Still bold and editorial — just no longer eating the whole viewport.

Removed a redundant 'AI-powered' badge from the search bar — the sparkle icon already says it, and the extra chrome was competing with the Create button.

When semantic search has no real matches for your query, the empty state now says so honestly and points you to popular prompts instead of leaving you stranded with a generic 'not found' message.

Semantic search on the Prompt Library now only returns prompts that actually relate to your query. Weak matches are filtered out before they reach you — search for 'tesla' and you'll see Tesla-themed prompts (or an honest 'no semantic matches' when the library doesn't have any), not a random mix of motivational and business prompts pretending to be relevant.

Newly created prompts are now searchable the moment you save them. Previously, only prompts that were part of a periodic backfill were findable by semantic search — anything saved between backfills was invisible until the next sweep. Embedding now runs at save time, so brand-new prompts show up in relevant search results immediately.

The Prompt Library landing page got a real makeover — bigger editorial typography, a massive widescreen hero showcase autoplaying the standout video prompt of the week, and a row of supporting picks underneath. Great prompts now feel like featured work, not search results.

A 'Create Prompt' shortcut now lives right in the library hero so you can jump straight into the builder from anywhere in the library.

Semantic search now labels its results as 'Top N matches' instead of a misleading total — the AI search surfaces the most relevant prompts, not every possible match, and the UI now reflects that clearly.

A brief render error that took the Prompt Library landing page offline for a window today — now resolved.

The Prompt Library now uses AI-powered semantic search. Type what you're looking for in plain English and it finds prompts by meaning — not just keyword matching. 'Cinematic portrait at dusk' finds dusk-mood prompts even if they don't use those exact words.

Every prompt detail page now shows up to six semantically related prompts below the main one, so you can discover variations and adjacent styles without starting a new search.

A featured strip at the top of the library showcases curated prompts with rendered video and image previews — hover to see them come to life before you click through.

Outpaint no longer fails with a bare "Unprocessable Entity" error when you expand from the edges. The submit path now drops zero-valued sliders before the request reaches fal.ai, so an "edges" expansion with one or two active sides goes through cleanly instead of being rejected as an empty payload.

When fal.ai does reject an outpaint request, you now see the actual validation reason — which field or value was wrong — instead of just "Unprocessable Entity." Same translation we already use for Smart Resize and other fal-backed tools, now wired into Outpaint.

Click Enhance now works with model-specific examples on every PixelDojo tool, not just the top 10. Niche image and video models now pull from their own track record instead of guessing from generic data — a real precision boost for less-mainstream tools.

The pool of prompts the enhancer learns from just got dramatically larger. Every thoughtful prompt now counts toward retrieval, not just the most-liked or admin-approved ones — those are still weighted higher, but no longer the only ones in the running. Many more candidates per request.

Deleting a single image or video from My Media now fully cleans up the underlying file. Before, the gallery entry would disappear right away but the file lingered in storage.

Click Enhance now learns from real prompts that have produced great generations on PixelDojo. Instead of using a generic template, the enhancer pulls a handful of high-quality examples that actually worked on your chosen model and uses them to shape the rewrite. Think of it as having access to the prompt notebooks of every successful generation on the platform — 40,000+ curated entries plus everything in the community gallery.

Per-model styling. The enhancer now knows that what works for Flux is different from what works for Veo 3, and pulls relevant examples from each model's track record. Your enhanced prompts are tuned to the model you're using, not a generic image template.

Gets better the more you use it. Every time you save, download, or upscale a generation, that signal feeds back into the enhancer. Prompts that produce things you keep get weighted higher next time someone enhances a similar idea.

Drop-in upgrade. If retrieval can't find anything useful for your specific prompt, you get the same enhancement you've always had — no regression, no errors. The new system is additive.

The Canvas co-pilot now recommends models based on what the community is actually using — not what its training data thinks is famous. Top models per chain type (generate, edit, upscale, animate) are pulled live from real Canvas usage and refreshed every hour. As new models climb the leaderboard, the co-pilot picks them up automatically. You can still name any model in your prompt ("use Flux," "with Seedream") and it will honor your pick.

Clarity Pro Upscaler is here — photorealistic upscaling up to 16× with a creativity dial that goes from strict fidelity all the way to expressive re-imagining. Realistic skin, preserved identity, and output up to 64 megapixels (~8K). Find it in Upscale & Enhance.

The community gallery and all user profiles are now fully public — no account required to browse, discover, and get inspired. Share any gallery link with anyone.

Gallery image descriptions are now generated the moment you open a piece. Open any image in the gallery and a smart description appears automatically, giving every creation more context.

The gallery detail page got a layout refresh — likes, views, share, and vote are now in a single compact row so the image takes center stage.

Kling Video multi-shot mode was silently blocking generation in some cases due to a form validation issue. It now submits correctly every time.

Multi-Shot Mode on Kling Video 3.0 — write a separate prompt for each shot (up to 6) and Kling stitches them into one seamless clip. Toggle it on, describe your scenes, choose Custom or Auto pacing, and generate. Great for short narratives, product demos, or any video that needs distinct beats.

Creator Studio has been retired. If you had it bookmarked, the link now takes you straight to Canvas — the better tool for multi-model workflows.

A small welcome card on the dashboard now leads with the most important thing to know: "Generations live here for 24 hours." Save what you want to keep, otherwise it's gone tomorrow. Inline links to your last-24-hour Recent pill, My Media, and Buy Credits. Dismiss it once and it's gone for your account across every browser and device.

Tool empty states (the screen you see before you generate anything on most tools) now include the same 24-hour notice. Belt-and-suspenders so you see the message before you spend credits, not after you wonder what happened to your generation.

The Recent floating pill (bottom-left of the dashboard) just got a real job. Every row now has three icon buttons on the right: Save, Download, and Remove. Save promotes that generation into your permanent My Media library; Download grabs the file directly to your computer; Remove hides the row from Recent (it never deletes the file or anything in your gallery).

Recovery: if you accidentally delete a generation from My Media, Recent still has it for 24 hours. Hit Save in Recent and it goes right back into your library.

Saved items now show a green checkmark in Recent so you can see at a glance which generations are already in your permanent library and which are still just "in flight."

ImagineArt 2.0 (preview) is here. The next-gen entry in the ImagineArt family lands with sharper realism, stronger composition, and the most faithful prompt following yet. 1 credit per image.

The /imagineart page now hosts the whole ImagineArt lineup — 1.5, 1.5 Pro, and 2.0 — in one workspace. Pick the model from the panel on the right, no more page-hopping. The menu link and page title are now just "ImagineArt" instead of "ImagineArt 1.5."

Pasting a really long prompt into Qwen Image (or Qwen Image 2, or either of their Edit variants) used to bounce with a validation error — the model only reads 800 characters and we were rejecting anything longer outright. Now we trim the prompt to fit and pop a one-time warning toast telling you it happened, so the generation just runs. Same treatment for negative prompts (which cap at 500). If you're calling the API directly, the response now carries a `warnings` array describing what got trimmed.

Search in All Tools is much smarter. Multi-word queries now AND across tokens (typing "edit face" actually returns face-editing tools instead of zero results), category names are part of the searchable text (typing "face" surfaces everything in Characters & Faces, not just tools with "face" in the title), and a small synonym map covers common variations — "edit" → inpaint / modify / fix, "upscale" → enhance / sharpen, "video" → clip / animate, etc.

All Tools sidebar feels calmer. The 11 category icons used to all show in full color at rest — a parade of pink / red / blue / purple / etc that fought for attention with no clear hierarchy. Icons now render at half opacity at rest with a much subtler colored chip behind them, then snap to full color on hover or when their category is active. Same color identity, way easier to scan.

Reverted the mobile All Tools Pinned and Recent sections from two-column to single-column. The two-column was harder to read because tool labels and descriptions got tight at narrow widths.

The search box is now a permanent affordance in the top bar — click it (or press ⌘K) to instantly search across every tool, style, and model. Replaces the static Dashboard / Canvas / My Media / My LoRAs / My Prompts / Gallery / API row that used to sit there. Those destinations live one click away in All Tools' Quick Access section, and on the dashboard cards. Search is the right primary affordance when you have 100+ tools.

Up to 3 of your pinned tools now show as compact icon buttons in the top bar (on screens md and wider). Pin from any tool card and it appears in your bar; unpin to remove. Pin order is stable so your bar doesn't shuffle. The icon for the tool you're currently on lights up — your bar tells you where you are.

Theme toggle moved out of the bar and into the user menu (click your avatar). Theme is rarely changed once set, and freeing the bar slot leaves more room for what you actually use day to day.

Mobile All Tools is way less to scroll. Browse Categories now shows up first (instead of being buried under Pinned, Recent, Quick Access, and Featured Tools). Pinned + Recent fit two-per-row instead of stacked. Featured Tools stays on the dashboard hero where you can already see them. Net result: about 60% less vertical scroll before you reach the catalog.

Removed the duplicate "Account" category from All Tools (it had Usage Report, Profile, Subscription, Feedback — all of which live in the user menu next to your avatar, where you'd expect to find account stuff).

All top-bar buttons are now 40px tall (was 36px), comfortably above the 44px-ish touch target threshold once you account for the padding around the icon. Easier to tap on iPad and laptop trackpads.

Open All Tools and the Home tab is now structured as a real launchpad: your Pinned tools at the top, your Recent generations next, then a Quick Access strip with Dashboard / Canvas / My Media / My LoRAs / My Prompts / Gallery / API, then a curated Featured Tools section. Pinning your first tool no longer makes the Featured suggestions disappear — both coexist.

The drawer's left-rail tab is now "Home" instead of "Quick Start," with a calmer Home icon. The "Featured Tools" hero on /dashboard and the "Featured Tools" section in the drawer now share the same curated list, so when we promote a model, both surfaces update.

When you start typing in the drawer's search box, the left rail no longer keeps the previously-selected category lit up — search results are global, so the highlight matched nothing. Sidebar entries dim while a search is active.

The Recent floating pill (bottom-left of the dashboard) now self-heals. If a generation it remembers has been deleted from your gallery or the temp preview expired, the broken thumbnail row auto-hides and the count updates instead of showing an empty box.

Drawer chrome cleanup: section labels are sentence case (Pinned / Recent / Featured Tools / Found N tools) instead of WIDELY-TRACKED UPPERCASE; duplicate Account / Usage / Credits links removed from the drawer footer (they live in the user menu and the credits pill).

Removed the two large pulsing blue / sapphire ambient circles that animated behind every dashboard page. They were 5%-opacity decoration; the dashboard now feels a touch calmer and more tool-focused without them.

A floating Recent button appears bottom-left of every dashboard page once you've generated something. Click it to see every successful generation from the last 24 hours across every tool — image, video, edit, the lot — with a thumbnail, timestamp, and one-click jump back to the tool that made it. Built for the all-too-common moment of "I refreshed and now I can't find my video."

Qwen Image edits no longer ignore your aspect-ratio selection. Picking 1:1, 16:9, 4:3 (or any other ratio) now resizes the output instead of silently mirroring the source image's dimensions. Caught and reported by jazze#77 — thank you.

If your connection drops mid-generation — common on mobile or shaky Wi-Fi — you used to see a bare "Failed to fetch" toast with no context, leaving you wondering whether your credits had been silently spent. The toast now reads "Connection lost — your credits are safe" and explains that if a generation actually started, credits refund automatically within 2 hours. Check your generation history before retrying. Applies across every tool page (every browser: Chrome, Safari, Firefox, Edge).

The public API for /api/v1/models/wan-image/run now matches its own documented default — you can omit numberOfImages and get 1 image, instead of bouncing with "validation_error: numberOfImages: Required." The docs always said default: 1 but the schema was requiring the field; both sides now agree.

Generations that fail validation now show what went wrong instead of silently doing nothing. If your prompt is too long, your image too large, or a value out of range, a clear toast tells you exactly which field and what to change. Previously the Generate button just returned with no feedback — a long-standing papercut across Qwen Image, Qwen Image 2, WAN Image, ImagineArt, Z-Image Turbo, Dreamina, Ernie, and Ideogram Character.

Running out of credits now shows "Insufficient credits" with a link to top up, instead of a generic "generation.failed" toast. ~50 tool pages were checking the credit-error response in a way that quietly missed the real shape — they all detect it correctly now.

Error toasts no longer leak internal codes like "INPUT.INVALID_PARAMETER" or "CREDITS.INSUFFICIENT" as the title. The friendly message that was supposed to render now actually renders.

Every parameter on every API model docs page now has a description. Hundreds of fields used to render as just a name and a type with no explanation; they now carry plain-English context — what the field does, what the valid values mean, when to use it. Affects /api-platform pages, the OpenAPI spec, the LLM-readable docs at /llm, and Canvas's "More Settings" panel. All five surfaces auto-update from a single source so future tools inherit the descriptions for free.

Edit tools — Qwen Image Edit, WAN 2.7 Image Edit, GPT-Image 2 Edit, P-Image Edit, OpenAI Image 1 Edit, HiDream Edit, and a few others — now ship a multi-image example alongside the single-image curl. The capability was always there, but the docs only showed one URL; it was easy to miss that you can compose up to 3-9 reference images in a single call. The new examples make it obvious.

HiDream is now HiDream O1 — a brand-new image backbone replacing the previous L1 family. Two variants live behind one tool: Full (the production quality target) and Dev (a faster, distilled version). Pick between them with a one-click pill right above the prompt.

Pricing simplified to a flat 1 credit per generation and 1 credit per edit. No more dual-credit tiers — every HiDream call costs the same regardless of variant.

Aspect ratio moved inline with the composer. Pick Square HD, Square, Landscape 4:3 / 16:9, or Portrait 3:4 / 9:16 right next to the Generate button — no more buried Settings modal. The picker uses real aspect-ratio icons so the layout you'll get is visible at a glance.

Image editing got the same upgrade. The Edit endpoint now uses HiDream O1's image-conditioned mode (Full and Dev variants), still 1 credit per call, and accepts up to four reference images for subject-driven edits if you call the public API directly.

Cleaner workspace — the standalone Settings modal and the post-generation auto color correction option are gone. Both were holdovers from the old engine that didn't have a home in the new flow.

Public API users: /api/v1/models/hidream/run and /api/v1/models/hidream-edit/run are live with the new model + image_size schema. The OpenAPI spec and LLM docs picked it up automatically. The legacy /api/hidream and /api/hidream/edit endpoints continue to work for existing integrations.

Qwen Image 2 now defaults to 2048×2048 (was 1024×1024). If you weren't picking a size manually, every new generation gets 4× the pixels — noticeably sharper out of the box. The previous default was using a quarter of what the model can actually produce.

Qwen Image 2 unlocks a high-fidelity "Max" resolution tier: 2048×2048, 2688×1536, 1536×2688, 2368×1728, and 1728×2368. The 2K and 1K tiers stay as alternatives for faster generations.

Qwen Image Plus and Max gained three controls that were previously hardcoded: smart prompt rewriting (the model adds detail to short prompts), watermark on/off, and a custom negative prompt. All available in the public API and Canvas More Settings.

Qwen Image Edit (both the legacy and 2.0 variants) now respect a watermark toggle. Output sizes are also tightened to a documented set — invalid combinations get a clean error instead of a confusing 500.

Sending more than 1 output to Qwen Image Plus or Max used to bounce with an unhelpful upstream error — the underlying model is fixed at 1 image. You now get a clean message at request time telling you to use LoRA mode if you want multiple outputs in a single call.

Power users on the public API can now pin to dated model snapshots (e.g. `qwen-image-2.0-pro-2026-04-22`, `qwen-image-plus-2026-01-09`) for reproducible behavior across model rev bumps. The rolling names like `qwen-image-2.0-pro` keep tracking the latest.

Polished the look of the Qwen Image and Qwen Image 2 pages in light mode — Generate and Subscribe buttons now use the sapphire site CTA color, and the resolution picker reads cleaner. The new "Max" tier shows what it actually represents on hover (about 4 megapixels, the model's documented sweet spot).

WAN 2.7 Pro now generates true 4K images. Switch to Pro on /wan-2.7-image and set resolution to 4K — output jumps from 2048² up to 4096². Text-to-image only (the model caps editing at 2K).

Resolution and aspect ratio are now independent controls. Pick your size (1K / 2K / 4K) and your shape (1:1, 16:9, 9:16, 4:3, 3:4, 5:4, 4:5) separately, instead of hunting through a combined "size" dropdown with twelve mixed values. Standard model gets 1K and 2K; Pro adds 4K.

Edit Mode accepts up to 9 reference images at once, up from 4. Bring your whole mood board.

Interactive edit: in Edit Mode, toggle on "Interactive edit (bounding boxes)" and drag rectangles directly on each reference thumbnail to restrict changes to those regions. Up to 2 regions per image. Useful for outfit swaps, sign edits, or selective restyles without touching the rest of the image.

Custom color palettes in the Settings menu — bias a generation toward 3–10 specific colors with weighted ratios. Hex picker, percentage sliders, and an auto-balance button that distributes ratios to a clean 100%. Good for brand work or hitting a specific mood.

Seed control in the Settings menu. Pin a number for reproducible runs; leave blank for random.

All of the above is also live on the public API, the OpenAPI spec, and the LLM docs — scripts and agents pick everything up automatically.

The public API for Kling 2.6 Pro and Seedance 1.5 now accepts the video duration as a plain number (e.g. "duration": 5) as well as a string ("5"). Previously, sending the number — which most HTTP clients do by default, and which our own Seedance curl example showed — bounced back with "duration: Expected '5' | '10', received number." Both shapes work now.

Seedream 4.5 custom output sizes that fall outside the model's supported range used to fail with a cryptic provider error ("image size must be at least 3686400 pixels. Request id: 0217…"). You now get a clear message with your actual width × height, the real limit (about 3.7–16.8 megapixels), and a concrete suggestion like "increase to 2560×1440".

When the generation service behind P-Image, P-Image Edit, P-Video, and P-Video Avatar is briefly overloaded, you now see a plain "the generation service is busy right now — please wait a moment and try again" instead of a bare technical error like "PRUNA API error: 504."

When Canvas Co-pilot's planner refuses your request because the underlying model's safety filter flagged it, you now see "Co-pilot can't plan that — your request was flagged by the safety filter. Please rephrase your goal …" instead of the generic "Co-pilot couldn't draft a plan. Try again." The previous wording asked you to retry an identical prompt that would keep failing.

Nano Banana reference-image dimension errors used to surface as "Runware: Invalid reference image width. The reference image width must be an integer between 1 and 12288." — leaking the provider brand name and the internal parameter name. You now see a clean message that explains each side must be between 1 and 12288 pixels and recommends 4000px on the long edge as a safe target. Same cleanup applies to prompt-length rejections (the message now says exactly which range is supported, no brand prefix). Centralized in the Runware client so every Runware-backed tool — Image Relighting, Google Gemini Image, Pixverse v6, Consistent Characters, Veo 3.1, and Nano Banana — gets the cleaner messages.

WAN 2.5 in Canvas now accepts an empty prompt when you're animating an image. Previously, dropping an image into a WAN 2.5 step and hitting run without typing a description bounced back with a "Prompt is required" error from your own UI — even though the underlying model is happy to animate the image with neutral motion. The schema now treats the prompt as optional in image-to-video mode (text-to-video still requires one).

If your card got declined and you fixed it by starting a fresh subscription (instead of updating the card on the failing one), your account could end up marked as expired even though the new payment had gone through and the next renewal was already on the calendar. The two billing events were arriving seconds apart and the older one was overwriting the newer one. Subscriptions now stay active in this case, and a small number of customers we found in the bad state have already been put back to active automatically — no action needed on your end.

Canvas now has an AI co-pilot. Press / on the canvas (or click the spark icon on the right edge) and ask it for anything — "make 4 lifestyle variants of my active shot," "build a launch from shop.example.com/runner," "animate this into a 6-second reel." It drafts a plan, shows you the cost upfront, and one click runs the whole chain. Parallel-safe too: 4 lifestyle gens fire at the same time instead of waiting in a queue.

Plans collapse to a clean title + cost + Run button. Tap "Show steps" to see the numbered plan, or "Show graph" for a tiny DAG view of how steps depend on each other. Power-user detail is one click away when you want it; out of sight when you don't.

100-credit hard cap per plan, enforced both at the AI planner and on the server. If a plan would cost more, you get a clean rejection with a hint to ask for something smaller — no surprise bills.

Co-pilot runs use the same execution path as a manual generate — same credits, same automatic refunds on failure, same job poller, same telemetry. Failed nodes can be retried without re-running the ones that already succeeded.

Greenfield keyboard map: / and ⌘K both open the co-pilot. ⌘M opens the model picker. Esc collapses. The right-edge icon rail is 56px when collapsed so the existing single-generate workflow looks and feels exactly the same — power users don't pay for a feature they didn't ask for.

Grok Imagine now has a Quality tier alongside the original Fast tier. Pick Quality for noticeably sharper detail, more accurate compositions, and stronger text rendering — ideal for thumbnails, hero images, ad creative, or any final visual you'd ship to a client.

Choose 1K (1024px on the long edge) or 2K (2048px on the long edge) output on the Quality tier. 1K is faster and cheaper; 2K is the high-resolution deliverable. The default is 2K.

The Quality tier supports editing too — upload an image, describe the change, and get a high-fidelity edit at 1K or 2K. Works the same way the Fast tier edits do, just with crisper output.

Pricing stays simple: Fast tier is still 1 credit per image. Quality is 2 credits at 1K or 3 credits at 2K. The credit chip on the Generate button updates live as you switch tiers and resolution.

Canvas now offers Grok Image Quality as its own tile in the model picker — drop it into any chain when you want a cleaner final pass without leaving the canvas.

When Grok Image Edit can't generate an output — usually because xAI's safety filter quietly refused the edit, or the source image is in an unsupported format — you now see a message that names the likely cause and tells you what to do (rephrase the edit instruction, or re-upload the source image as a JPEG or PNG, then retry). Previously it surfaced a bare "Prediction failed:" with nothing after the colon, which gave you zero clue what went wrong.

pixeldojo:campaign — drop a product URL into your AI agent (Claude Code, Cursor, Codex, Cline, Windsurf, Zed, and any other MCP client) and get back a hero image, four lifestyle variants, and a vertical video, all from one call. About 50 credits, two minutes end to end. Perfect for a quick-turn ad pack or a social drop.

pixeldojo:from_url — paste any product page URL (Shopify, WooCommerce, BigCommerce, or anything with structured data) and get back the product name, description, and images. Used as the first step inside Campaign, but also handy on its own when an agent needs structured product info without copy/paste.

pixeldojo:generate now accepts a preset slug. Pass preset: "cinematic-portrait" or preset: "product-floating-clean" and skip the prompt-writing step entirely. The 60 expert-tuned presets in our library now work the same way through your agent as they do on the website.

Job status checks now show how long a job has been running. After 5 minutes your agent gets a "this is unusually long" warning; after 12 minutes it gets told to start a fresh request instead of polling forever. Fewer agents stuck in endless poll loops on the rare hung job.

The Skills page at pixeldojo.ai/skills now shows the new agentic tools front-and-center, with code snippets for each. New shortcut: paste pixeldojo.ai/skills.md into your AI agent and say "install this" — the page has a built-in directive that tells your agent exactly which command to run.

Nano Banana jobs that appeared stuck on "still processing" — even after the actual generation had completed — now resolve correctly. The status check was looking up jobs through the wrong backbone for the Pro and Pro v2 tiers, so it never saw the completion. Old stuck jobs unblock immediately if you retry; new jobs return URLs as expected. If you generated a Nano Banana image in the last day and it never came back, retry it now.

When fal.ai rejects a Smart Resize request (status 422), you now see the actual validation reason — e.g. which target size or input field was wrong — instead of the bare "Unprocessable Entity" status text. Same translation we already use for the other fal-backed tools, now wired into the Smart Resize submit path so you can fix the input on the first retry.

The rotating agent chip on the Skills page now shows the actual brand mark for each integration — Claude Code, Cursor, Codex, Cline, Windsurf, and Zed — instead of two-letter initials. The chip also sits properly inline with the headline now, lined up with the cap of the surrounding letters instead of dropping below the baseline.

The install steps under the hero are tidier — Step 01's title fits on one line so all three columns line up, and the agent list across the page (hero subtitle, install card, and "Works with" pills) is consistent end-to-end.

The hero headline on the Skills page now cycles through the agents that support PixelDojo — Claude Code, Cursor, OpenClaw, Codex, Hermes, NemoClaw, and Cowork — each with a small color-coded chip so you can immediately spot your setup.

The install section is now three numbered steps (01 / 02 / 03) instead of icon cards, making the path from zero to working agent quicker to skim.

The new /presets page lets you browse all 60 workflow presets in one place — from product photography and fashion to storyboards and text rendering. Every card links directly to the matching tool with the preset loaded.

The preset library grew from 24 to 59 entries, all with real generated example images. Each thumbnail was produced with that exact preset so you know what output to expect before you click.

The Skills page now has a live terminal demo showing exactly what the agent conversation looks like: you describe what you want in plain English, it routes to the right model, and you get back an output URL. Makes the workflow concrete instead of abstract.

The hero on the Skills page now cycles through real agent use cases — things like "Build a 4-shot product storyboard" — so it's immediately clear what you can hand off to your LLM.

A "Build with PixelDojo" card is now on the main dashboard, linking developers straight to the Skills page and API Platform from day one.

Seedream 4 and Seedream 5 Lite now translate ByteDance's "The request failed because the image format is not supported by the API. Request id: 0217..." into a clean message that names the likely cause (HEIC, AVIF, GIF, or a corrupted file) and tells you what to do (re-upload as JPEG, PNG, or WebP). The internal request id no longer leaks to users.

Reference-image size errors on Nano Banana and Consistent Characters are clearer. The previous message — "Please re-upload an image between 5kb and 20MB" — fixed the unit casing and now explains both failure modes: if your image is too large, compress or resize it (under 4000px on the long edge usually works); if it's smaller than 5KB, the file may be corrupted or didn't finish uploading. Includes the supported formats (JPEG, PNG, WebP) so you know exactly what to send back.

When Seedream rejects a prompt for sensitive content or copyright reasons, you now see a clean "Generated content was blocked by the model provider — try a different prompt or input media" instead of the raw "The request failed because the input text may contain sensitive information. Request id: 0217..." envelope. Credits are refunded the same way; the change is just the message you see.

If you ever saw the orange "You're out of credits" banner while your header still showed plenty of credits, that's gone. The banner could latch on after a brief subscription-status hiccup or a top-up and stay visible until you dismissed it manually — even with hundreds of credits in your account. It now re-checks the live state on every render and hides itself the moment your balance is healthy again.

@pixeldojo/mcp ships on npm. Run `npx @pixeldojo/mcp init` for copy-paste install snippets, set PIXELDOJO_API_KEY, restart your agent, and you've got five named tools your LLM can call: pixeldojo:generate (any prompt → image or video), pixeldojo:character (consistent characters across shots), pixeldojo:storyboard (multi-shot scenes from one brief), pixeldojo:upscale (enhance any image), and pixeldojo:status (check on long jobs). Works with Claude Code, Cursor, OpenClaw, and any other Model Context Protocol host.

Smart polling for long jobs. Fast generations come back in the same call. Jobs that take longer than 30 seconds — LoRA training, long-form video — return a job ID and a status URL, and your agent calls pixeldojo:status to check on them. No more 5-minute MCP timeouts that look like failures. Polling uses exponential backoff (1s → 2s → 4s → 8s) so we don't hammer your account's rate limit.

Your API key never appears in your agent's logs. The MCP server strips any pd_xxx token from every log message before it leaves the process — even debug output. Same for error messages: a 401 surfaces as "Set PIXELDOJO_API_KEY" with a link to the keys page, never as a stack trace with your token in it.

Recraft tool page now respects the ?preset=<slug> URL param. Click a preset thumbnail anywhere on the site and the recraft form opens with the prompt and aspect ratio already filled in. The other preset-having tools follow next.

PixelDojo Skills replaces the developer-agent landing at /agents. The new /skills page leads with a one-line install for AI-agent tools (Claude Code, Cursor, OpenClaw), covers four named skills your agent can call (generate, character, storyboard, upscale), and shows real preset thumbnails you can try without writing a prompt. Old /agents and /agents.md links keep working — they redirect to the new home.

Try-a-preset on every tool docs page. Visit any model's page on the API Platform and you'll see a row of curated starting points — cinematic portraits, product-on-marble shots, magazine covers, and more. Click one and we'll open the tool with those parameters loaded, so you can play before you write a single line of prompt. Twenty-four presets ship today across the most-used image and video models, with more on the way.

Use-case filter chips on the API Platform. Browse 130+ models by what you actually want to make: marketing, character, cinematic, fashion, photorealism, text-in-image, upscale, video, and more. The filters are deep-linkable — share /api-platform?useCase=character with a teammate and they land on the right shortlist. Empty results now ship with a one-click clear-filters button instead of a dead-end message.

The Skills page tells the truth about scale. The hero now reads 130+ models — matching what's actually on the API Platform — instead of the old 40+ figure. The model showcase pulls live counts from the registry, so the number you see is the number you can call.

Sapphire-aligned design across the developer landing. The fuchsia-violet-cyan gradient look is gone, replaced with the sapphire palette that already drives the rest of the product. Cleaner hero, cleaner CTAs, less visual noise — the page feels like a workspace tool, not a marketing template.

If a renewal payment failed and your subscription ended, the Subscribe page could keep telling you "You're Already Subscribed" — sometimes for an hour or more — and the Subscribe button stayed locked out. The page now reflects billing changes within a minute, so re-subscribing after a card decline works the way you'd expect.

Submitting WAN 2.2 with an unsupported quality_mode + resolution combo (e.g. quality_mode "plus" with resolution "720p" — Plus only supports 480p and 1080p) used to fail opaquely as "credit_error: Invalid credit amount" after the request was already accepted, sounding like an account issue. You now get a schema validation error upfront that names the exact resolutions your chosen quality mode supports — e.g. "Resolution '720p' isn't available on the plus quality mode — pick one of: 480p, 1080p."

Switching models in Canvas no longer leaks settings from the previous tool. Picking Qwen Image 2 with the Pro variant and then switching to WAN 2.7, WAN, or ImagineArt used to send the previous tool's `model` value to the new one, failing with "Invalid enum value, expected 'wan-2.7-standard' | 'wan-2.7-pro', received 'qwen-image-2.0-pro'". Same root cause for sizes (1280×1280 / 1536×1536) and durations (numeric vs. string) leaking across model switches. Canvas now resets all advanced settings whenever the active model changes — re-selecting the same model preserves your current picks.

Seedream 5 Lite custom-size mode now works end-to-end. Picking the "Custom" image size with your own width and height used to be rejected at the schema gate as "Invalid enum value, expected 'auto_2K' | 'auto_3K' | 'square_1_1' | … received 'custom'" even though the server already supported it. Custom dimensions now flow through cleanly, matching how Seedream 4.5 already worked.

Sending more than 14 reference images to Nano Banana used to surface the raw provider error "Runware: Invalid number of elements for 'referenceImages' parameter" — confusing brand and parameter names with no guidance. You now get a clean message that includes your actual count and tells you exactly how many to remove (e.g. "You sent 20 reference images. Nano Banana supports up to 14 — please remove 6 and try again.").

WAN 2.5 in Canvas no longer rejects 3s and 4s durations and now lets you go up to 15s. The duration slider used to span 3-10s, but the model's actual range is 5-15s, so picking 3 or 4 failed with "Number must be greater than or equal to 5" and the 11-15s range was hidden. Slider now matches the model exactly, and stale durations from sibling tools (WAN 2.2, Hailuo, LTX) auto-clamp into range.

WAN Reference to Video in Canvas no longer fails when you switch in from a sibling tool with a different resolution or aspect-ratio picker. Stale '480p' values from WAN 2.2 and unsupported aspects like '3:2' used to slip through and trigger "Invalid enum value" — Canvas now defensively falls back to 720p / 16:9 when an incompatible value is in flight.

P Video Avatar is live. Upload a portrait, write a short script (or paste an audio URL for lip-sync), and get a talking-head video. 30 named voices, 10 languages — English (US/UK), Spanish, French, German, Italian, Portuguese, Japanese, Korean, Hindi. 1 credit per second at 720p, 2 credits per second at 1080p. Find it under Characters & Faces in the All Tools menu.

Switching to Hailuo 02 in Canvas after a model with a different resolution picker (WAN 2.5, PixVerse, etc.) used to fail with "resolution: Invalid enum value, expected '768p' | '1080p', received '720p'" because Canvas carried over the previous model's value. Canvas now exposes Hailuo's exact resolutions and falls back to 768p when an incompatible value is in flight.

LTX 2 in Canvas no longer rejects odd-numbered durations. The duration slider used to step by 1 (so picking 7s or 11s failed with "Duration must be one of: 6, 8, 10, 12, 14, 16, 18, 20 seconds"); it's now an explicit list of valid lengths. Picking longer than 10s on the Fast tier also auto-locks 25fps + 1080p so the cross-tier requirement is always satisfied.

Runway Gen-4.5 in Canvas now constrains duration to 5s or 10s the same way the standalone Runway Video tool does, so stale 6s/8s values from sibling models no longer leak through and fail with "duration: Invalid input".

Happy Horse Reference to Video is here. Upload up to 9 reference images (characters, props, costumes), describe your scene with character1 / character2 tokens, and the model generates a video with each subject locked to its image. 720p or 1080p, 3-15 second clips, 4 credits per second at 720p and 6 at 1080p.

Happy Horse Video Edit lets you transform existing footage with a prompt — style transfer ("make this look like a Studio Ghibli film") or local replacement ("swap the sweater for the one in the reference image"). Pair a source clip with up to 5 reference images, choose to keep your original audio or have new audio generated. Same pricing as the rest of the Happy Horse family — 4 credits per output second at 720p, 6 at 1080p. Find both in Generate Videos under WAN Family.

Happy Horse 1.0 is now live. Generate fast, expressive video from text or a starting image at 720p or 1080p, with clip lengths from 2 to 15 seconds. 4 credits per second at 720p, 6 credits per second at 1080p. Find it under Generate Videos in the All Tools menu.

Public API calls to qwen-image-2.0, qwen-image-2.0-pro, qwen-image (Plus/Max), the Qwen Image Edit model in image-relighting, and the Qwen Image 2 Pro model in consistent-characters were hanging indefinitely. The dashboard worked fine because it polls a different endpoint, but the public API's polling didn't recognize the internal job ids these tools were returning. They now return image URLs directly when called via the API — first /api/v1/jobs/{id} GET serves the result. Dashboard / Canvas / Creator Studio behavior unchanged.

The Qwen Image Edit, Qwen Image 2 Edit, and HiDream Edit endpoints are now reachable via the public API at /api/v1/models/{qwen-image-edit, qwen-image-2-edit, hidream-edit}/run. Sister set to Generation now matches every dashboard tool 1:1 — every customer-facing capability has a public apiId.

Each /api-platform/{apiId} doc page now shows a "See also" section linking related apiIds (gen ↔ edit pairs across 16 tools: Flux, GPT Image, Nano Banana, Kling, P-Image, Qwen, Seedance, WAN, XAI, HiDream). Customers building integrations can discover companion endpoints without leaving the docs.

Canvas's "More Settings" panel now auto-populates for any new tool — derived from the tool's schema enums + inputParameters. Add a tool.config.ts and the tool gets a working settings panel for free, no manual registration in the Canvas params table.

Two new CI gates so /api-platform docs and Canvas can't drift from the schemas: canvas-param-schemas-parity (every Canvas option must be schema-acceptable) and no-passthrough-schemas (schemas reject unknown fields, customers see clean 400s instead of silent drops).

CLAUDE.md now formally documents the single-source-of-truth principle: tool.config.ts is canonical. Adding a tool means writing one file — every doc surface, the public API, Canvas, Creator Studio, and OpenAPI all derive from it automatically.

Closed the rest of the schema-vs-docs drift across the API platform: 11 more tools (Hailuo 02, Image Relighting, Pony XL, Qwen Image 2, Recraft, Seedream 4.5, Seedream 5 Lite, WAN 2.6 Image, WAN 2.6 Image Edit, WAN 2.7 Image, WAN 2.7 Image Edit) had documented enum lists for parameters like resolution, image_size, scheduler, output_format, and aspect_ratio while the schema accepted any string. The schema now enforces the same set the docs advertise — bad values fail fast with a clear 400 instead of bouncing off the provider after credits are deducted.

Added a reverse-direction parity check to CI (tests/contracts/tighten-loose-string-audit.test.ts): every documented enum value must be enforced by the schema. Combined with the existing forward check (every advertised value must be acceptable), the docs and schemas can't drift apart again.

Aspect-ratio and resolution fields on the public API now reject invalid values up-front instead of being passed through to the provider for a confusing 422. ~16 tools (Flux, Flux 2, Flux Edit, Google Gemini Image, Nano Banana, Dreamina, Seedream 3, P-Image, P-Image Edit, Qwen Image, WAN 2.2 / 2.5 / 2.7 video, WAN Image, Ideogram Character, HiDream, Image Editor, Video Reframe, Video Transform, WAN 2.2 Animate) had loose `z.string()` validation that let unsupported values through silently — they now use the same enum the dashboard offers.

WAN 2.5 / 2.7 video API now accepts the documented `720p`/`1080p` form (the dashboard's value) — previously the schema only accepted uppercase `720P`/`1080P` even though every doc said lowercase. The API now matches the docs and continues to send the upstream-correct uppercase form to DashScope on your behalf.

HiDream resolution is now a proper enum on the API (`1024x1024`, `1360x768`, etc.). The dashboard's human-friendly form (`1024 × 1024 (Square)`) still works, so you can use either.

Runway Gen-4.5 was 500'ing on every Canvas chain because the duration picker emitted "5" or "10" as text but the schema expected a number. Both formats now work, so animating a Canvas frame with Runway goes through on the first try.

LTX-2 was occasionally rejecting Canvas runs with "FPS must be one of: 24, 25, 48, 50" when the FPS dropdown was set — same number-as-text mismatch. Coerced at the schema, so any FPS option you pick now works. Also added a guard that drops stale resolution values (like 480p left over from a previous WAN step) and falls back to 1080p instead of failing the run.

Consistent Characters and Google Nano Banana now show "Your reference image is outside the allowed size range. Please re-upload an image between 5KB and 20MB and try again." or "We couldn't fetch your reference image — the link may be unreachable or the format unsupported. Please re-upload (JPEG or PNG) and try again." instead of a raw Runware JSON envelope with internal taskUUIDs and documentation URLs.

Image-edit endpoints for GPT Image 2, GPT-Image 1.5, and OpenAI Image 1 are now available via the public API: POST /api/v1/models/gpt-image-2-edit/run, /gpt-image-1-5-edit/run, and /openai-image-1-edit/run. Send a prompt plus image_urls (or input_images) and the same edit pipeline the dashboard uses runs end-to-end — image URL sanitization included. Existing /api/v1/models/{wan-2.7-image-edit, wan-2.6-image-edit, kling-image-edit, p-image-edit, xai-image-edit}/run already worked the same way; this fills in the orphans.

Each new edit apiId is documented on /api-platform with example curls and shows up in /llm/{apiId}.txt for AI-assistant-driven integrations.

API documentation now matches the schemas the public API actually validates against. Previously several tools' /api-platform pages and /llm/{model}.txt docs advertised parameters the schema would silently drop — for example `num_outputs` listed on Kling, Qwen, WAN, Seedream, Imagineart, P-Image, and others when the underlying schemas use `num_images`, `numberOfImages`, or `n` (or only generate one image per call). Renames + removals applied across ~15 tools so the names you read are the names the API accepts.

Google Nano Banana 2 and Nano Banana Pro (text-to-image variants under /api/v1/models/google-gemini-image/run) advertised `reference_images` for guided generation but the schema didn't accept the field. Reference images now work end-to-end on those variants — same image URL sanitization and Runware backbone the dashboard uses.

/llm/{model}.txt now derives parameters from each tool's Zod schema when no hand-written list exists — so reference-image fields, custom dimensions, and other inputs always show up in the LLM-friendly docs even for tools we haven't manually documented yet. Matches the same fallback /api/v1/models/{model} already does.

Heygen Avatar API parameters renamed to match the schema: `expression` → `expressionHappy`, `background.type` → `backgroundType`, `background.value` → `backgroundValue`. The dashboard already used these names; the API docs now agree.

Deleted ~600 lines of stale hand-rolled API model entries that had drifted from each tool's actual schema (e.g. a static `nano-banana` apiId that competed with `google-nano-banana`, a `seedream-4` entry advertising `reference_images` while the live schema accepted `image_urls`). Every public apiId now comes from a single source of truth: the tool's own tool.config.ts.

If you upload an unsupported image format (HEIC, AVIF) or a corrupted file to Seedream or Magnific, you now see "Could not read your reference image — the file format may be unsupported (HEIC, AVIF) or the image may be corrupted. Please re-upload as JPEG or PNG." instead of a raw provider JSON dump or Python traceback.

When a referenced image URL returns 404 / 403 / 401 (link expired or access denied), Grok Imagine Edit and other image-editing tools now say "We couldn't load your image — the link returned 404 (not found). The image may have expired or moved. Please re-upload and try again." rather than "Failed to fetch image: 404".

Canvas was offering a 480p option for WAN 2.5 that the backend doesn't support (silently failed every run). Removed it from the Resolution dropdown. WAN 2.7 Video and Seedance 2 also now defensively normalize stale resolution values from previous Canvas steps so chains like "Kling Image → Seedance 2 video" no longer carry over the wrong setting.

WAN family + Qwen models running on Alibaba DashScope now show "Alibaba DashScope is temporarily unreachable. Please try again in a moment." instead of a raw "fetch failed" error when the upstream service is briefly down. Same treatment for ByteDance / Seedream when its servers are overloaded — credits are still refunded automatically.

WAN 2.2 image-to-video was occasionally rejecting valid runs with "sample_shift: Must be greater than or equal to 1" from Replicate. The schema now bounds the field correctly and clamps any out-of-range value, so the run goes through cleanly.

Reference-image fields on public API endpoints (Nano Banana's image_input, Seedream's image_urls, Veo 3.1's reference_images, and every other tool that accepts source or reference URLs) were being silently dropped because the public API bypassed the per-tool payload shaping that tool pages, Canvas, and Creator Studio already used. Public API calls now run through the same submission pipeline — including image URL sanitization, multi-tier routing (e.g. Veo 3.1 Lite correctly uses its own backbone), and field-name translation — so a curl to /api/v1/models/{id}/run behaves identically to the same model on the site.

Synchronous image models (Seedream 4.5, Grok Imagine, Z-Image Turbo, Recraft sync, Nano Banana Pro) now return completed output on the first /api/v1/jobs/{id} poll — no more spinning while waiting for a provider job that never existed.

API calls to LTX-2 sub-models like /api/v1/models/ltx-2-pro-i2v/run were silently being validated as text-to-video instead of the mode the endpoint name implies, so callers got confusing errors about "text-to-video" rules while posting to an image-to-video endpoint. Each variant's mode is now applied automatically from the endpoint you call — a minimal {prompt, image_url} call to ltx-2-pro-i2v just works.

LTX-2 fps validation used to return "fps: Invalid input" with no list of valid values. It now says "FPS must be one of: 24, 25, 48, 50." Aspect ratio errors similarly spell out the supported values and when each applies.

The Community Gallery and user profile galleries are now behind sign-in. Browsing generations, exploring model pages, clicking into individual images, and viewing other creators' profiles all require a PixelDojo account going forward — the same account you use for the dashboard. Signed-out visitors get bounced to sign-in and routed back to the page they wanted.

Video Upscaler and Video Transform were throwing "Prediction ID is missing in the response" on every submit after a recent behind-the-scenes change. Both tools are working again — upload your clip and go.

If a P-API model (P-Image, P-Video) can't reach its upstream service, you now see "Pruna AI service is temporarily unreachable. Please try again in a moment." instead of the raw "fetch failed" error. Credits are still refunded automatically.

Canvas was failing silently on Flux Image-to-Image and Face Enhance when you tweaked More Settings like Guidance Scale, Prompt Strength, or Scale Factor — the inputs emit text and the schema expected numbers. Both now accept either, so your Canvas runs go through on the first try.

Canvas was offering a 21:9 aspect ratio for Grok Imagine that the backend doesn't support, so that pick always 500'd. Canvas now only shows the 7 aspect ratios Grok actually accepts.

If you pick Flux Pro or Flux 1.1 Pro with 21:9 or 9:21, you now see "Flux 1.1 Pro does not support 21:9 — try 16:9, or switch to Flux Schnell or Flux Dev for ultra-wide ratios." instead of a raw 422 dump from Replicate. No credits spent when the pre-check rejects.

Creative Upscale was occasionally 500'ing from Canvas when Creativity or Resemblance sliders landed outside the provider's range. Values are now clamped to what Clarity Upscaler accepts (Creativity 0–1, Resemblance 0–3) so the run goes through cleanly.

Prompt safety checks now run server-side on every tool, Canvas, Creator Studio, the public API, and LoRA training fields — not just the dashboard pages that had the client-side check. Three layers: a public-figure name list, a deterministic phrase scanner, and an AI moderation backstop. Anything flagged returns a clear, actionable message about what to change.

The image safety check that previously ran only on a couple of tools now applies to every tool that can produce explicit imagery, and uses a more reliable classifier than before. Photos that depict children are rejected with a specific message. Other tools (upscalers, background removers, analyzers) are unchanged.

Every image stored on PixelDojo is now eligible for CSAM hash scanning at the storage layer. All upload paths — generation outputs, reference images, and LoRA training datasets — route through the same scanning coverage.

When an uploaded image is over our 12 MB cap, we now show your file's actual size and tell you to resize or compress — no more guessing why the upload failed. Applies to every tool that takes an image.

On Seedream 4, if your reference image is bigger than the 36-megapixel limit or the host is slow to respond, you now get a clear message with the actual dimensions and next steps — instead of a raw technical error.

You can now paste Civitai model URLs from the new civitai.red domain anywhere we accept Civitai links — LoRA training, model imports, and HuggingFace uploads all recognize it automatically.

Z Image Turbo — the sub-second photorealistic model — now has the sticky mobile composer. Describe, pick aspect ratio, add an optional LoRA, and generate up to four at a time. Eighteen tools on the new shell.

ImagineArt 1.5 now has the sticky mobile composer. Describe your image, pick aspect ratio and model tier, and generate up to four at a time from a single pull-up panel. Seventeen tools on the new shell.

Portrait Series Generator (Consistent Characters) now has the sticky mobile composer. Pick a source portrait, choose a model, set a background, and generate up to 12 portraits — all from a single pull-up panel. Sixteen tools on the new shell.

Lip Sync, Video Autocaption, and Video to Sound all got the sticky mobile composer. Fifteen tools on the new shell now. The interior UI on these three is still pre-design-system (custom headers and buttons) — a polish pass will line them up with the rest.

Text To Music now has the sticky mobile composer. Describe the music, drag the length slider, tap an example prompt, generate — all from a single pull-up panel. Eleven tools on the new shell.

Hunyuan 3D now has the sticky mobile composer. Pick text or image input, tweak advanced settings, and generate — all from a single pull-up panel. Ten tools on the new shell.

Subject Control, PuLID Characters, and Image to 3D have all been removed. Subject Control overlapped with Virtual Try-On and Style Transfer. PuLID and Image to 3D were both on disk but never made it into the sidebar menu — nobody could navigate to them. Consistent Characters covers identity preservation, Hunyuan 3D covers 3D generation.

Dreamina 3.1 in Canvas now only offers 1K and 2K for resolution — the two values the model actually accepts. If your Canvas session had 4K or HD carried over from another model, Dreamina runs were failing at submit; stale state is now politely remapped to 2K instead of erroring.

Magic Lighting now has the sticky mobile composer. Upload, pick a model, tap a lighting preset, and generate — all from a single pull-up panel. Eight image tools on the new shell.

Style Transfer now has the sticky mobile composer. Source, style, prompt, and advanced settings all live in one pull-up panel. Seven image tools on the new shell now.

Change Camera Angle now has the sticky mobile composer. The orbit visualizer, sliders, presets, and generation options all live in one panel you can slide up from any scroll position. Six image tools on the new shell now.

Virtual Try-On now has the sticky mobile composer. Upload your garment, pick your model, and generate from any scroll position — five image tools on the new shell now.

Portrait Upscaler now has the sticky mobile composer. Four image tools on the new shell so far — upload, set your scale factor, and generate from any scroll position on mobile.

OmniHuman's 'Generate from Text' button works again. A plumbing change a week ago had disconnected the text-to-speech response from the page — you'd click Generate Audio, nothing would appear, no error. Now it polls to completion and shows the audio ready to use. Caught while sweeping other tools for the same class of bug as Background Remover.

Background Remover now actually shows your finished image again. A recent plumbing change had disconnected the final result from the page — you'd click Remove Background, wait, and get sent back to the empty state with no image. Fixed, verified end-to-end.

Extract Frame and Image Analyzer now have the same sticky mobile composer as Background Remover. Tap the bar at the bottom to open controls, tap again to see your result. Three image tools on the new shell now; more following on the same cadence.

Canvas signing now rotates on a 15-minute window, matching Creator Studio. If a signed link ever leaked from the page, its useful life is minutes, not hours. Zero visible change to how Canvas works for you.

On mobile, Background Remover now has a sticky composer at the bottom of the screen — tap to open the upload controls from any scroll position, tap again to collapse and see your result. Same pattern video tools have used for a while. First image tool to get it; the rest will follow.

Under the hood, image and video tools now share one shell. Three duplicate components (ToolWorkspaceShell, ToolPageHeader, ToolEmptyState) replace six parallel ones (VideoToolShell, VideoPageHeader, VideoEmptyState + image siblings). Every page you touch lives on the same plumbing now — future polish lands everywhere at once.

Every Creator Studio frame and video generation now flows through the same reliability backbone as tool pages and Canvas. If a model has an issue, we spot it and recover the same way whether you hit it from a tool page, Canvas, or Creator Studio. Zero visible change to your workflow — just one shared plumbing instead of two.

Flux LoRA trainers — all six of them — now use the proper Flux purple gradient on their Generate buttons and chrome. Orange-red and yellow-orange drifts are gone. Every Flux-family tool, training or generating, speaks the same color.

Softer corners across marketing and landing pages — hero cards, pricing tiles, community gallery, model showcases, and tutorial blocks now all share the same corner radius as the rest of the product. 42 files touched. The dashboard-to-landing transition is visibly one product now.

Image tools and video tools now share one empty-state component. Video tools keep their icon pills and info tips (like "Lite supports 720p or 1080p in 16:9 and 9:16"), image tools keep their colored-dot pills, and the visual style is consistent across both. Under the hood this eliminated a whole duplicate component tree.

The shared component library had a longstanding bug where every 'purple' accent was silently rendering as sapphire (the primary brand color) instead of actual purple. Across the Flux family (Flux Studio, Flux 2 Flex, HiDream, PonyXL, LoRA Image Creator, Feedback) that meant empty states, prompt boxes, aspect ratio pickers, credit cost displays, and mode toggles were all speaking the wrong color. Fixed at the source — 16 shared components across tool-components, video-components, history-grid, pricing-plans, admin, and my-images. Every purple-accented surface now renders actual purple.

Kling Image and Text to Music Generate buttons moved to amber — the proper Kling-family and audio-category color. Matches the rest of the Kling and audio surfaces across the product.

xAI Image's Generate button cleaned up from a sapphire-to-rose mix to pure rose — consistent with xAI Video and the rest of the xAI family.

Magnific Upscaler, Creative Upscale, and Google Gemini Image are now the category-correct blue on Generate and chrome, instead of sapphire, yellow-amber, or blue-purple mixes.

Ernie and P-Image are now pure cyan — matching Qwen, Recraft, ImagineArt, and the other cyan-family tools. No more stray rose or sapphire on their chrome.

LoRA Image Creator's cost display and Generate button are now the Flux-family purple gradient throughout instead of drifting back to sapphire.

The Subscribe CTA on the Qwen Image 2 page moved from a mixed emerald-teal gradient to solid sapphire — matching the product-wide rule that site-level actions like Subscribe are solid, while per-tool Generate buttons carry the gradient.

Video utility tools (Video Reframe, Video Transform, Video Upscaler, Lip Sync, Merge Videos) now share a consistent green across their Generate buttons and chrome — matching the category identity.

Subject Control, a character tool, now uses rose — the character-tool family color — instead of sapphire.

The Subscribe page CTA moved from a sapphire-to-blue gradient to solid sapphire — site-level actions like Subscribe are solid sapphire across the product now, gradients are reserved for the single Generate button inside each tool.

Flux 2 Flex and HiDream Generate buttons are now the same purple as Flux Studio and PonyXL — matching the documented Flux-family color. Sub-states (editing, upload, etc.) all agree now instead of drifting between blue, sapphire, and green.

OpenAI Image 1's Generate button moved from a blue/sapphire gradient to emerald — the proper OpenAI-family color, consistent with GPT Image 2 and GPT Image 1.5.

Video to Sound's Generate button now uses amber — the audio-tool family color — instead of blue. Matches Text to Music and other audio surfaces.

Softer corners on cards, sheets, and empty-state tiles across 17 tool pages — from Flux Studio to Seedance 2 to Purchase Credits. The whole product sits at the same corner radius now instead of a handful of pages looking bubble-rounded.

Flux Studio's Generate button now uses the proper Flux-family purple gradient instead of blue — matches PonyXL and restores the family visual identity. The Subscribe-to-create CTA was cleaned up to solid sapphire (site-level actions are sapphire across the product now).

Removed a shimmer sweep animation from Generate buttons on Flux Studio and Seedream 4, plus the focus glow behind their prompt boxes. Tool surfaces are meant to be quiet; your prompt and your outputs hold focus now.

Toned down all-caps-shouting button labels on Billing, Settings, My Prompts, Merge Videos, and Usage Report. Labels are readable at a glance now instead of broadcasting at you.

Creator Studio's chrome now matches the rest of PixelDojo. Buttons and labels across the project header — Save, Characters, Scene, Shot Settings — use the same quiet treatment you see on every tool page. Your project title reads in title case instead of all caps.

Softer corners on the pre-project picker cards and the shot settings sheet. Same radius as the rest of the product, so Creator Studio sits next to Canvas and your tool pages without looking like a different app.

The in-project workspace is calmer. Removed a decorative background glow so your storyboard holds focus. Ambient motion still lives on the pre-project picker where it belongs.

Seedance 2, Seedance 2 Reference, and Seedance 2 Video Edit have moved to a new backbone. Same models, faster and more reliable queuing — no more jobs getting stuck on "queued."

Most Seedance 2 generations now cost fewer credits. Fast 480p is down to 2 credits/sec and Fast 720p to 4 credits/sec. High 720p is 4 credits/sec (5 with a reference video), and High 1080p is 10 credits/sec (12 with a reference video). The quality popovers and Canvas cost displays reflect the new rates automatically.

Seedance 2 Video Edit keeps Standard tier and full 480p / 720p / 1080p output; reference images and reference videos work the same way you're used to.

Your dashboard now recognizes whether you've generated before. If you're a returning creator, the top of the page becomes a compact welcome strip with Create Image and Create Video ready to click — and your Recent Creations move up to become the visual hero. First-time visitors still see the full onboarding intro so they know where to start.

The home page buttons are now solid sapphire across the board — no more yellow Buy Credits pill, no yellow hover flash on Create Image, no gradient Subscribe button. Cleaner, more consistent, and it lets your art stay the hero.

Featured Tools on the dashboard now show honest category labels — Image, Video, Editing — instead of marketing claims like "Best Quality" or "Most Flexible." You can see at a glance what each tool is for.

Video tutorials got their own home at /tutorials. The Master AI Creation video library moved off the main dashboard to a dedicated page, so your workspace stays focused on getting work done. You can jump to them anytime via the new Videos link in the Prompting Guides section.

The returning-visitor welcome strip keeps a Buy Credits (or Subscribe for free accounts) button right alongside Create Image and Create Video — styled as a subtle sapphire-accented pill so it's always one click away without competing with your creative actions.

GPT Image 2 Prompting Guide is live at /guides/gpt-image-2-prompting-guide — a hands-on playbook with seven generated examples covering near-perfect text rendering, dense infographic layouts, multilingual signage (Latin + CJK), world-aware photorealism, product mockups, character consistency, and UI mockups. Every prompt is copyable.

Every prompting guide now has a clean breadcrumb trail — Home › Prompting Guides › <Title> — so you always know where you are and can jump back to the index with one click.

The site footer has a new "Prompting Guides" column led by an "All Prompting Guides" link to the index, plus direct links to every major guide including GPT Image 2, Nano Banana Pro, Flux 2, Qwen Image 2, Seedream 5, Veo 3.1, Kling 3, and Seedance 2.

The /guides index now lists every guide we've published.

The AI Image landing page now leads with GPT Image 2 across its hero, FAQ, and structured data, and features a new Prompting Guides section so you can browse model-specific playbooks while choosing a generator.

Uploading an unusually large reference image to Nano Banana Edit (Pro tier) used to return a raw provider JSON blob. You'll now get a short, actionable message telling you the image dimensions are outside the supported range and to resize to within 1–12288 pixels per side before retrying.

GPT Image 2 is live at /gpt-image-2 — OpenAI's next-generation image model, with native 4K UHD rendering and noticeably sharper typography. Great for hero art, editorial covers, and print-ready marketing.

Six explicit render sizes in the size picker: 1024×768, 1024², 1024×1536, Full HD 1920×1080, QHD 2560×1440, and 4K UHD 3840×2160. Each 4K tile wears a small amber "2×" chip so you know when you're on the premium tier.

Tiered credit pricing that tracks compute honestly. Low quality is a flat 0.5 credits per image at any size. Medium is 1.5 credits (3 at 4K UHD). High is 5 credits (10 at 4K UHD). The quality popover shows the live per-image cost for the size you picked, so there are no surprises.

GPT Image 2 also supports edit workflows — upload or pick reference images and describe your change. Multi-image refs, optional edit masks, and the same 4K rendering carry through.

Available in Canvas as a full generate step with size, quality, and format controls in More Settings, and exposed in the public API as the gpt-image-2 model ID.

The Seedance 2 duration control in Canvas now shows the three durations the model actually supports — 5s, 10s, 15s — as discrete chips instead of a free-range slider. The slider let you pick 6 or 7 and then the model rejected the job; those nudges are no longer possible.

PixelDojo is now discoverable by AI agents and LLM-powered coding tools out of the box. The homepage and API reference advertise their machine-readable documentation (llms.txt, LLM-optimized reference, OpenAPI spec, ai-plugin manifest) through standard HTTP Link headers — so agents like Claude Code, Cursor, and ChatGPT can find the API contract on first visit without any extra setup on your side.

The API reference at /api-docs now speaks both HTML and Markdown from the same URL. Visit it in a browser and you get the rich HTML page; have an AI client fetch it with Accept: text/markdown and you get the LLM-optimized reference directly — now with the correct text/markdown content type so strict agents accept it on first try.

Added an API catalog at /.well-known/api-catalog (RFC 9727) — a single machine-readable index pointing to the OpenAPI spec, HTML and Markdown documentation, AI plugin manifest, and health endpoint. Agents that follow the well-known convention can now find everything they need in one request.

Published an Agent Skills index at /.well-known/agent-skills/index.json exposing three tool-worthy skills — generate-image, generate-video, and list-models — each backed by a markdown manifest with SHA-256 integrity. An agent runtime that speaks the Agent Skills Discovery RFC can now call PixelDojo image and video generation as first-class skills.

Added OAuth 2.0 Protected Resource Metadata at /.well-known/oauth-protected-resource (RFC 9728) with supported scopes and a pointer to the API key page. Agents can auto-discover how to get credentials for the PixelDojo API instead of needing bespoke setup docs.

Homepage structured data now identifies PixelDojo as a SoftwareApplication with a feature list, category, and offer details alongside the existing Organization and WebSite blocks. Cleaner ingestion by search engines and AI crawlers — and more accurate knowledge cards in answers about what PixelDojo actually is.

Declared our AI content preferences. The site robots.txt publishes Content Signals indicating the marketing site and documentation are open for AI training and inclusion in AI answers. User-generated gallery content and profile pages opt out of AI training via noai/noimageai — your posted creations are crawled for search and inspiration, not harvested for training.

Agents can now auto-discover how to sign in. Added OpenID Connect discovery at /.well-known/openid-configuration and the matching OAuth 2.0 authorization server metadata at /.well-known/oauth-authorization-server, mirrored from our Clerk-backed auth. Agents that speak either flavor can find the issuer, endpoints, and supported scopes without any bespoke config.

Markdown negotiation now works site-wide on every agent-facing surface — homepage, /agents, /api-platform, and /api-docs. Fetch any of them with Accept: text/markdown and you get the LLM-optimized reference; browsers keep getting the HTML pages. Moved the routing into middleware so it runs ahead of static-page caching, which the previous version got stuck behind.

Homepage Link headers are now served on every response, not just HTML ones. The original guard only emitted them when Accept: text/html was present, which meant agent crawlers using Accept: */* saw nothing. They now see the full RFC 8288 link graph (api-catalog, agent-skills, oauth-protected-resource, openid-configuration, service-doc, service-desc, and more) on any request to /.

Credit deductions are now fully transactional. Every debit and its ledger entry commit together as a single atomic unit, or not at all — so even if the database blinks mid-generation, your balance and our record of where it went stay perfectly in sync. Eliminates a whole class of "my credits got deducted but the job never ran and I never got a refund" edge cases.

API keys are now stored as one-way hashes, not the raw key. The full secret lives only in your hands — we cannot display it again after creation, and a database leak couldn't expose it either. Existing keys keep working seamlessly: they auto-upgrade to hashed storage the next time you use them. No rotation needed unless you've misplaced your copy.

Webhook HMAC signatures now use a dedicated signing secret separate from your API key. The signing secret is generated per-key and kept isolated — meaning webhook verification keeps working even if you rotate the API key, and vice versa.

Tool pages feel more like a professional studio and less like a consumer toy. Image and video output cards have subtler shadows and a lighter hover lift, so the art stays the hero and the chrome gets out of the way.

First-run delight: the first image, first video, and first Canvas chain step you generate in a session gets a small scale-in — then the animation goes quiet so it doesn't fire on every click.

This page (What's New) got an editorial redesign. Bigger display headlines, typewriter-style dates, magazine spacing. Designed to be read, not skimmed.

Per-second credit pricing on Seedance 2 Reference to Video and Seedance 2 Video Edit has been adjusted to reflect their higher compute costs. Seedance 2 Reference: Fast 5 cr/s at 480p and 9.5 at 720p; High 8 at 480p, 16 at 720p, 24 at 1080p. Seedance 2 Video Edit: Fast 6.5 cr/s flat; Standard 5 at 480p, 9.5 at 720p, 16 at 1080p. Plain Seedance 2 (text/image-to-video) pricing is unchanged.

1080p output is now available on Seedance 2, Seedance 2 Reference to Video, and Seedance 2 Video Edit — on the High / Standard quality tiers. Fast tiers stay at 480p / 720p.

Seedance 2 Video Edit brings back its Standard quality tier — higher fidelity for prompted video edits, especially at 720p and 1080p.

Pricing on all three Seedance 2 tools recalibrated to cleaner whole numbers. Seedance 2 / Reference: Fast 3 cr/s at 480p and 6 at 720p, High 5 at 480p, 10 at 720p, and 15 at 1080p. Seedance 2 Video Edit: Fast 4 cr/s flat; Standard 3 at 480p, 6 at 720p, 10 at 1080p.

Seedance 2 generations no longer hang on "queued" forever after the backbone migration — the status endpoint URL namespace was wrong in the first cut. New jobs now poll correctly.

Picking a video from "My Videos" in Seedance 2 Reference to Video no longer fails with "Failed to fetch" — the library picker was re-validating CDN-hosted videos and tripping browser CORS. Library videos skip the re-fetch now (they were already validated at upload time).

PixelDojo now respects your system-level "reduce motion" preference. If you have motion sensitivity settings turned on in your OS, decorative shimmer, pulse, and spin animations now pause, and transitions shorten. No setting to toggle in the app — your OS choice just works everywhere.

Seedance 2, Seedance 2 Reference to Video, and Seedance 2 Video Edit now run on a faster backbone — same prompts, same credit prices, faster and more reliable queuing. Expect fewer stalled jobs during peak traffic.

WAN 2.7 Video now accepts prompts up to 8,000 characters (previously 5,000). Long cinematic prompts and prompt-enhance expansions no longer trip the character cap.

Seedream 4 now retries transient ByteDance "service encountered an unexpected internal error" responses automatically with backoff. Generations that would previously fail after a provider hiccup now recover on their own.

Ideogram Character's resolution picker now lists only sizes the model actually accepts — the old options (512×512, 768×768, 1280×720, 1920×1080, 1024×768, 768×1024) were silently rejected by Replicate. Replaced with nine provider-valid resolutions covering square, 4:3, 3:4, 3:2, 2:3, 16:9, 9:16, and ultra-wide/ultra-tall.

Veo 3.1 Standard no longer offers 1:1 as an aspect ratio — Google's API only accepts 16:9 and 9:16, so picking square was guaranteed to fail. The dropdown now matches what actually works.

Grok Image (xAI) now accepts prompts up to 5,000 characters (previously 2,000), matching P-Image, WAN 2.5, WAN 2.7 Video, and Seedance 2.

Canvas resolution picker for WAN 2.5, WAN 2.7 Video, and WAN Reference to Video now shows a proper 720p / 1080p chip selector instead of a free-text field — no more "Expected '720P' received '720p'" surprises. Stale sessions with lowercase values still submit successfully.

Prompt length limits unified across every generation tool. Modern image and video models accept 65K–131K tokens (hundreds of thousands of characters), so our per-tool caps of 1K–8K chars have been lifted to a single generous 250K-character ceiling. Detailed multilingual prompts, prompt-enhance expansions, and long cinematic descriptions no longer hit arbitrary limits on Ernie, Kling, Grok, Hidream, FLUX Edit, WAN, Seedance, or any other tool.

P-Image Edit now accepts prompts up to 5,000 characters (previously capped at 2,000), matching its sibling P-Image tool. Detailed edit instructions and longer Canvas chains no longer hit a surprise validation error.

WAN 2.2's public API schema now documents all LoRA parameters. Standard mode supports two LoRAs (lora_weights_transformer + lora_weights_transformer_2 with matching scale fields), and Extended mode supports full dual-LoRA with four slots — 1 and 3 target the high-noise transformer, 2 and 4 target the low-noise transformer. Scales accept 0–4.

Seedance 2 now accepts prompts up to 5,000 characters (previously capped at 2,000), matching the longer-prompt tools like WAN 2.5 and WAN 2.7.

The API Platform pricing page now shows real credit costs for 17 more models that were stuck displaying "0 credits" — Consistent Characters, FLUX, Flux Edit, Nano Banana Edit, Grok Imagine R2V, Ideogram Character, Kling Image V3, Kling Image Edit, Kling 2.6 Pro, Lip Sync, OmniHuman 1.5, Flux Redux, Seedance 1, Seedance 2 Reference, WAN 2.7 Image Edit, WAN Reference to Video, and WAN 2.7 Video Edit. Every variant now surfaces the correct per-request price.

When an image upload fails in an image-to-image tool or Canvas, you'll now see the real reason (like a session or subscription issue) instead of a generic "Failed to upload" message that made it look like the file itself was the problem.

Seedance 2 Reference to Video is live. Guide Seedance 2.0 with up to 9 reference images, 3 video clips, and 3 audio tracks — tag them in your prompt as [Image1], [Video1], [Audio1] to lock characters, motion, and sound in a single generation. Fast (3/5 credits/sec at 480p/720p) and High (3/6 credits/sec) tiers, native audio sync, 3–15 second clips.

Seedance 2 Reference in Canvas now shows proper Quality, Resolution, Duration, Aspect Ratio, and Generate Audio dropdowns instead of raw text inputs that could silently fail validation.

P-Image, WAN 2.5, and WAN 2.7 Video now accept longer prompts (up to 5,000 characters) — previously anything over 2,000 characters was rejected with a validation error.

GPT-Image 1.5, Qwen Image 2, and Ideogram Character now show proper dropdowns in Canvas More Settings — Size, Quality, Model, Speed, and Style controls replace raw text fields that sometimes caused silent failures.

Seedream 4 edit mode no longer rejects requests that pass a single reference URL instead of an array — the API now accepts both forms gracefully.

Pose Control (FLUX) works again — submissions without a LoRA were being rejected upstream because the empty strength value was sent as a string instead of a number.

Grok Imagine Video in Canvas now has proper Duration, Resolution, and Aspect Ratio dropdowns instead of raw text fields that silently failed validation.

Earlier validation on Veo 3.1, WAN 2.6 Image Edit, and Seedream edit. Veo 3.1 (standard) now catches 1:1 aspect ratio with an input image upfront and asks for 16:9 or 9:16. WAN 2.6 Image Edit caps reference images at 4 when selecting. Seedream 4 and Seedream 5 Lite edit cap reference images at 14 when selecting. All three avoid confusing post-submit errors.

Ernie — Baidu's multilingual text-to-image model is live. Write prompts in English, Chinese, or Japanese with built-in prompt expansion. HD is 1 credit, UHD is 3 credits, and the Turbo variant is 1 credit flat.

"Continue in Canvas" actually sends your image into Canvas now. If you'd ever used Canvas before, the button used to take you there empty-handed.

Seedream 4 and Seedream 5 Lite edit mode accept pasted or newly-uploaded images again — base64 uploads were timing out before reaching the backend.

Veo 3.1 Lite videos finish and display correctly. Jobs were completing on the provider but never got picked up on our side.

Pose Control (FLUX) with multiple LoRAs works now — was 0% success after the multi-LoRA update.

SeedEdit 3 is back online after a provider API change broke every request.

LTX 2 in Canvas now has proper Tier, Duration, Resolution, and FPS dropdowns instead of raw text inputs.

WAN Reference to Video in Canvas gets proper Duration slider plus Resolution, Aspect Ratio, Model, and Shot Type dropdowns — no more typing values that silently fail.

WAN 2.6 Image Edit in Canvas now has Size, Prompt Enhancement, and Negative Prompt as proper form controls. Previously free-text fields were tripping validation.

Magnific Upscaler now catches oversized images upfront with a clear "Image is too large to upscale (max 10 MB)" message instead of a cryptic upstream error.

Nano Banana edits with "Match input image" aspect ratio no longer 400 on the upstream API — always sends explicit width and height now.

Seedance 2 image-to-video no longer complains that you need an image after you already uploaded one — mode and image state stay in sync.

Omnihuman's Fearful, Disgusted, and Surprised emotions produce audio now (were sending wrong values to the voice engine).

Flux Edit's guidance slider in Canvas now stays inside the range the model accepts — no more mid-drag validation errors.

Empty-prompt submissions across 30+ tools now show a clean error. Advanced JSON mode, Enter-key submits, and rapid clicks all get caught client-side instead of hitting the API.

Content-moderation rejections across the Seedream and Seedance families are correctly labeled as content issues instead of provider outages — cleaner signal when something is actually broken.

Public API calls to Heygen Avatar, Flux variants, Kling, Veo, Seedream, Seedance, WAN, Recraft, Ernie, and ~60 other models no longer return a spurious "Invalid credit amount" error. Per-request pricing for these models now flows through to the API correctly.

The /api/v1/models listing and API Platform docs pages show real credit costs for dynamically-priced models instead of "0 credits".

Your Usage report loads noticeably faster — added indexes that make per-user lookups across credits and refunds much quicker, especially for longer time ranges.

Canvas's model picker now shows every model tier as its own option — pick Nano Banana Standard, Pro, or 2 directly instead of hunting through settings.

Multi-image editing in Canvas — models like Nano Banana Edit, P-Image Edit, and WAN 2.6 Image Edit accept up to 5 reference images. Click '+ Add reference' next to your source to attach more.

Canvas's More Settings panel is way friendlier. Aspect ratio and resolution are dropdowns. Image slots have a proper picker. Fields you've already filled from a canvas step get hidden so you're not duplicating work.

When you pick an image source for a video tool, mode auto-flips to image-to-video and the image slot is pre-filled. No more empty fields asking what to do.

P-Image Edit is now available in both the public API and Canvas — fast image editing with up to 5 reference images.

Fixed the bug that silently sent edits to the wrong endpoint — Canvas edits now land on the correct model every time.

Admin Tool Health dashboard — per-tool success rate, error breakdown, and latency for the last 7 days.

Light mode contrast fixes on the Canvas source indicator and temporary-output warning — readable in both themes now.

WAN 2.5 Image-to-Video now disables Generate until you've uploaded an image — no more silent failures.

WAN 2.7 Video Extend mode waits for your source clip's duration to finish processing before letting you Generate.

Nano Banana Pro and Nano Banana 2 work correctly via the public API again — requests now route to the right backend.

Image editing tools added to WAN 2.7 — paint over areas and regenerate with AI.

Usage history now shows which tool and provider were used for each generation.

All API error messages are now clearer and more consistent across every endpoint.

Strengthened security across the platform — all endpoints now require proper authentication.

Homepage loads 40% faster. Font loading improved by 97%.

Removed 90+ unused files and 48 unnecessary dependencies for a leaner, faster app.

Credits are now deducted upfront for LoRA training jobs — no more surprise charges.

Automatic refunds are more reliable — failed jobs always return your credits.

Brand-new Sapphire color scheme across the entire platform for a cleaner, more modern look.

Updated accent colors for each tool family — image, video, and editing tools now have distinct visual identities.

New typography with Satoshi for headlines and Geist for body text — easier to read everywhere.

Credit refunds section added to your usage report so you can track every refund.

Refund tracking is now more transparent — see exactly when and why credits were returned.

Resolved edge cases where timed-out jobs weren't automatically refunding credits.

Stuck jobs are now automatically detected and refunded every 10 minutes.

Admin dashboard now has better tools for monitoring job health across the platform.

WAN 2.7 Video is here — next-generation video creation with improved motion and detail.

WAN 2.7 Image generation added with updated model configurations.

Video continuation support — extend your WAN 2.7 videos with last-frame extraction.

Better resolution handling for video models with smarter aspect ratio defaults.

Streamlined credit purchase flow — fewer steps to top up your account.

Better error messages during checkout so you always know what happened.

WAN 2.6 Image model added to the lineup.

Reference-to-video now supports WAN 2.7 with better subject consistency.

Auto-resubmit for failed video downloads — no more lost generations.

Prompt enhancement defaults updated for QWEN Image 2 and Flux 2 for better results out of the box.

HeyGen avatar tool now has smarter credit calculations based on video length.

Seedance 2 is live — create videos with realistic motion and native audio awareness.

Updated AI model roster with the latest versions and pricing.

Video prompting guides added to the homepage for quick access.

Seedream 4 now supports custom image sizes and aspect ratio selection.