Video Extension
Generated on PixelDojo with Grok Imagine Video Extend. Produced by PixelDojo's generation pipeline.
Five models on Pixel Dojo take a clip you already have and add new footage to it: Grok Imagine Video Extend, WAN 2.7 Video in continuation mode, Seedance 2.5 Video Edit in extend mode, Flux 3 Video with a video input, and WAN 2.7 Video Edit with an overridden output length. The cheapest is Grok Imagine Video Extend at 2 credits per second of new footage, and it is the only one that bills the added seconds alone. The others price the whole output, and Seedance also bills the source clip you fed in. Below is each one's source limit, length cap and exact credit formula, read off the tool code.
The five models, source limits and credit formulas
Grok Imagine Video Extend
Purpose built for this one job. Source is an MP4 between 2 and 15 seconds, up to 100 MB. New footage runs 2 to 10 seconds and defaults to 6. The charge is 2 credits per second of new footage, rounded up, so the default is 12 credits and the maximum for one call is 20. The source clip length does not change the price.
WAN 2.7 Video, continuation mode
Set the mode to video continuation and pass a clip. The source has to be 2 to 10 seconds, and the requested output length must be strictly longer than the source. Output runs 2 to 15 seconds at 720p or 1080p. It bills the full output at 2.5 credits per second at 720p and 4.5 at 1080p, so stretching a 5 second clip to 10 seconds at 720p is 25 credits.
Seedance 2.5 Video Edit, extend mode
The only one that can add footage before the clip as well as after it. Source runs 4 to 30 seconds and up to 200 MB, output 4 to 30 seconds. Billing counts the source seconds plus the output seconds at 1.25 credits per second at 480p, 3 at 720p and 7.5 at 1080p. A 5 second clip extended to 10 seconds at 480p is 15 billable seconds, or 18.75 credits.
Flux 3 Video with a video input
Passing a clip switches it into video to video, which extends or transforms the source. The rate jumps from 5 credits per second at 720p to 10, and from 8 to 12 at 1080p. Output is capped at 15 seconds in this mode against 20 for text to video, with a 5 second floor. Aspect ratio is ignored because the source clip sets the frame.
WAN 2.7 Video Edit with a length override
This one is an editor first. It re-renders your clip from a prompt and defaults the output to the source length, but supplying an output duration lets you run past the original. It charges 2.5 credits per second at 720p and 3 at 1080p, and it can keep the source audio instead of generating new audio.
Only one of them bills just the new seconds
Grok Imagine Video Extend charges for the extension and nothing else. WAN 2.7 and Flux 3 Video charge for the entire finished clip, so the seconds you already had get paid for twice. Seedance goes further and bills the source length on top of the output, because the model reads the whole input clip back in.
Auto length costs more up front
Seedance lets you leave the output length to the model. When you do, the job reserves credits at the 30 second maximum, because the finished length is unknowable at submit time, and refunds the difference once it settles.
Source length is a hard gate, not a suggestion
WAN 2.7 rejects a continuation outside 2 to 10 seconds before any credits move, and it rejects an output that is not longer than the source. Grok rejects anything outside 2 to 15 seconds. Both checks run before the charge, so a bad clip length costs nothing.
Adding 6 seconds to a 5 second clip: 12 credits on Grok Imagine Video Extend, 27.5 credits on WAN 2.7 at 720p. Same footage, different billing rules.
Why Choose Pixel Dojo for Video Extension
Professional-quality results with cutting-edge AI technology
12 credits for the common case
Six more seconds on the end of a clip, the default setting on Grok Imagine Video Extend, is 12 credits.
One key for all five
Each is a POST to /api/v1/models/{model}/run with the same API key and the same request shape, so switching between them is a string change.
Extend backwards as well as forwards
Seedance 2.5 in extend mode takes a direction, so new footage can go in front of the clip to build a lead in.
How It Works
How to pick one and price the job:
Measure the clip you already have
Length decides the shortlist. Under 10 seconds opens all five. Between 10 and 15 seconds rules out WAN 2.7 continuation. Over 15 seconds leaves Seedance 2.5, which accepts up to 30.
Decide whether you want new footage or a new version
Grok Imagine Video Extend, WAN 2.7 continuation and Seedance extend mode keep the original and add to it. Flux 3 Video and WAN 2.7 Video Edit re-render the whole clip from your prompt.
Multiply the rate by the seconds that model bills
Grok charges the added seconds. WAN 2.7 and Flux 3 Video charge the finished length. Seedance charges the source plus the output. Doing that sum first is what stops a 10 second extension costing more than you expected.
Loved by creators on PixelDojo
Real feedback from people using PixelDojo, pulled from our in-product surveys.
All the tools, readily available and easy to understand
You guys, the LORA training is the BEST in the space right now, great job with that!
Tons of variety. Ease of use. Tools for beginners and advance users.
easiest workflow to get best output image or video
awesome functionality, UX, constant flood of improvements, discord interactivity
Awesome site with so many features
Explore more AI tools on PixelDojo
AI Tools
Compare & Switch
- Best AI Image Generators
- Best AI Video Generators
- Midjourney Alternatives
- Civitai Alternatives
- Runway Alternatives
- Leonardo Alternatives
- Pika Alternatives
- Luma Alternatives
- Magnific Alternatives
- Veo Alternatives
- Flux Alternatives
- Freepik Alternatives
- Seedance Alternatives
- Seedream Alternatives
- Pixverse Alternatives
- GPT Image Alternatives
- Synthesia Alternatives
- Playground Alternatives
- NightCafe Alternatives
- Canva AI Alternatives
- ElevenLabs Alternatives
- ComfyUI Alternatives
- Fal Alternatives
- Replicate Alternatives
Common Questions
Everything you need to know about Video Extension
Which AI model is cheapest for extending a video?
Grok Imagine Video Extend, at 2 credits per second of new footage. Its 6 second default costs 12 credits and its 10 second maximum costs 20, and the length of the clip you feed it does not change the bill.
How long can the source clip be?
It depends on the model. Grok Imagine Video Extend takes 2 to 15 seconds up to 100 MB. WAN 2.7 continuation takes 2 to 10 seconds. Seedance 2.5 takes 4 to 30 seconds up to 200 MB. Flux 3 Video caps the finished clip at 15 seconds when a video is supplied.
Do I pay for the footage I already had?
On four of the five, yes. WAN 2.7, WAN 2.7 Video Edit and Flux 3 Video bill the whole output length, and Seedance bills the source length plus the output length. Grok Imagine Video Extend is the exception and bills only the seconds it adds.
Can new footage be added to the start of a clip?
Seedance 2.5 Video Edit in extend mode takes a direction setting, so the new footage can go before the source instead of after it. The other four continue forwards from the last frame.
What happens if I let the model choose the length?
Seedance reserves credits at its 30 second maximum, then refunds the difference when the job settles. Reserving high is deliberate, because a job that under reserves cannot claw the money back afterwards.
Can I do this from the API?
Yes. All five are POST calls to /api/v1/models/{model}/run with the clip passed as a public URL. WAN 2.7 needs the mode set to video continuation, Seedance needs the mode set to extend, and Flux 3 Video switches by itself when a video field is present.