Step by step
How to make AI videos, shown one real step at a time
Start with a still, animate it, give it a voice, extend it, caption it. We made the whole video below on models that are unlimited on any Pixel Dojo plan.
Most guides list tools. We made one short video and show every step as it came out, each one built from the step before. Then we made the same idea again on Seedance 2.5 with an extension and burned in captions, so you can see what credits add.
7.5 credits ($0.47) per 5 s clip at 720p. Included in every plan, from $10/month.
Already a member? Sign in
By Brian Lovett · Updated
More videos made the same way on MiniMax H3 Turbo
We wrote these four prompts straight to video on MiniMax H3 Turbo, with sound. On any plan, clips like these spend no credits.
Circus: a flyer in emerald crosses between two catchers under a striped big top.
Wide low angle inside a striped red and white big top at night. A trapeze artist in her thirties in an emerald sequined costume lets go of the bar at the top of the swing, turns once in the air and is caught by the wrists by her partner hanging from the opposite bar. Spotlights sweep through haze. Sound: the creak of the rigging, the crowd gasps, then cheers.
Monsoon evening: a wide rooftop at dusk with water tanks, laundry lines and warm windows behind.
Rooftop of an old apartment block over a dense city at dusk. A man in his sixties in a white shirt steps out as the first monsoon downpour hits, opens his arms and laughs, rain bouncing off the water tanks and laundry lines around him. Handheld medium shot, warm lights in the windows behind. Sound: roaring rain, distant thunder, his laugh.
Night dive: a manta ray glowing red above a diver as the camera tilts up.
Night dive in deep blue water. A diver in her thirties holds a red flare as a giant manta ray glides slowly over her, its pale belly lit red, plankton sparkling in the light. Slow upward camera tilt. Sound: muffled bubbles and a deep underwater hum.
Shadow theater: a painted bird takes off across a lit paper screen.
Behind a glowing paper screen in a dark theater, a shadow puppeteer in her fifties shapes her hands into a bird, and the bird shadow flaps its wings and flies across the screen. Side view shows her silhouette and the warm lamp behind her. Sound: soft flute music, the rustle of paper, a hush from the audience.
The six steps to make an AI video
- 1
Write the idea as one shot
Who, doing what, where, in what light. Ours: a neon sign maker bending a glowing red tube over a blue flame in a dark brick workshop.
- 2
Make the still first
Generate the opening frame as an image. Fixing a face or a set is cheap on an image and slow on a video.
- 3
Animate from that frame
Send the still to an image to video model as the start frame and describe only the motion, the camera and the sound.
- 4
Add a line or sound
Put the dialogue in quotes and name the sounds. Our models wrote the voice, the burner hiss and the hum into the clip.
- 5
Extend the shot
Start a new clip on the last frame of the one before, or use a video extender, and describe what happens next.
- 6
Caption it
Run auto captions so the line reads on a muted feed. The words come from the speech in the clip.
Walkthrough one: every step on unlimited models
The still is from Qwen Image 2.1 Turbo and every clip is MiniMax H3 Turbo at 720p. Both are unlimited on any paid plan. The tiles show what each step would cost through the API.
Qwen Image 2.1 Turbo made the opening frame.
Cinematic film still, 35mm. A neon sign maker in her forties with cropped grey hair and a leather apron bends a glowing tube of red neon over a blue gas flame in a dark brick workshop at night. Finished abstract neon shapes glow on the wall behind her: a pink wave, a cyan crescent moon, a yellow star. Reflections of light on her safety glasses, haze in the air, shallow depth of field, warm and cool contrast.
MiniMax H3 Turbo animated the still, with the burner hiss and an electric hum.
<Picture 1> She slowly bends the glowing red tube into a curve over the blue flame, turning it with both hands, the flame flickering against the glass. Slow push in toward her hands. Sound: the hiss of the gas burner, a low electric hum, a faint buzz from the signs.
Same start frame, a new prompt with the line in quotes. The voice is in the clip.
<Picture 1> She lifts her eyes from the flame, looks straight into the camera and says calmly: "Every curve takes one breath." Then she looks back down at the glowing tube. Static medium shot. Sound: the hiss of the gas burner, a low electric hum, her voice close and clear.
A new MiniMax H3 Turbo clip that starts on the last frame of step 4.
<Picture 1> She bends the glowing red tube into a smooth curve over the blue flame, then lifts the finished curve away and holds it up to the light, the glass glowing brighter. The camera drifts slowly to the left. Sound: the hiss of the gas burner, a low electric hum.
- Her face, glasses and the three wall signs stayed the same in all three clips because every clip started from the same picture.
- MiniMax H3 Turbo spoke the quoted line word for word; we checked the audio.
- Starting a clip on the last frame of the one before carried the shot on with no cut in the scene.
- This whole chain spends no credits on any paid plan.
Walkthrough two: the same idea on credit models
We made the same idea again with Nano Banana 2.1 for the still, Seedance 2.5 for the talking clip, the Grok video extender for four more seconds and auto captions on top.
Nano Banana 2.1 at 2K, 1.5 credits.
Cinematic film still, 35mm. A neon sign maker in her forties with cropped grey hair and a leather apron bends a glowing tube of red neon over a blue gas flame in a dark brick workshop at night. Finished abstract neon shapes glow on the wall behind her: a pink wave, a cyan crescent moon, a yellow star. Reflections of light on her safety glasses, haze in the air, shallow depth of field, warm and cool contrast.
Seedance 2.5, 5 seconds at 720p with sound, 30 credits.
She looks up from the flame, straight into the camera, and says calmly: "Every curve takes one breath." Then she goes back to bending the glowing red tube over the blue flame. Slow push in. Sound: the hiss of the gas burner, a low electric hum, her voice close and clear.
The Grok video extender added 4 seconds to the clip for 13.5 credits.
She lifts the finished red curve away from the flame and holds it up to the light, the glass glowing brighter, as the camera drifts slowly to the left.
Auto captions, 5 credits.
- Seedance 2.5 gave the richest set: smoke, a full workbench and a second wall of signs.
- The extender carried her motion on without a cut and kept the voice from the first five seconds.
- Auto captions transcribed her line exactly and highlighted each word as she said it.
How to make AI videos for free
You can make the first walkthrough without spending credits. On any paid Pixel Dojo plan, MiniMax H3 Turbo is unlimited for clips up to 5 seconds at 720p, and Qwen Image 2.1 Turbo, FLUX 3 Image Klein and P-H3-Image are unlimited for 1K stills.
With no plan at all, you get 3 free effects a day from our one click video and photo effects. Free effect videos carry a watermark.
Seedance, Kling, Veo and the other top models run on credits. Each model page shows the price before you press generate.
Habits that make AI videos look made on purpose
One shot per clip
Five seconds holds one action. Ask for two and you get a rushed version of both.
Quote every spoken word
Write the line in double quotes and say who says it. Short lines land best.
Name the sounds
A burner hiss, rain, a crowd. Models that make audio follow a sound list closely.
Reuse your best frame
Keep one great still and start every clip from it. That is how the face and the set stay the same.
Draft unlimited, finish on credits
Find the motion on MiniMax H3 Turbo, then run the final take on Seedance 2.5 or Kling 4.0 when the shot needs more.
Let Sensei run the chain
Describe the video and Sensei plans the steps, prices them, waits for your approval and checks every take.
What else you get on Pixel Dojo
Unlimited on every plan
On every paid plan, Relax runs Qwen Image 2.1, FLUX 3 Image Klein and MiniMax H3 Turbo, plus every video and photo effect, unlimited with no credits spent.
Sensei, the AI director
Describe a project and Sensei plans it, prices it, waits for your approval and reviews every take.
Video Studio
Generate, edit and extend a clip in one place, on whichever top video engine suits the shot.
One click video effects
300+ one-click video effects turn a photo into a trend clip, from AI Hug to AI Inflate, with no prompt to write, and 3 are free every day.
Film Studio
An idea becomes a finished film: script, cast, storyboard, clips and an editor pass that trims every shot, with each step priced before it runs.
Start from a genre
Pick a format such as a UGC ad, a micro drama, a movie trailer or an explainer, and the matching studio opens with the settings and a sample idea filled in.
Credits that roll over
Your credits roll over while you stay subscribed, so nothing you pay for disappears at the end of the month.
Every top model, plus an API
More than 80 image and 60 video models in one account, with a public API and a hosted MCP server for your own apps and AI assistants.
What the steps cost
Plans are Lite $10 a month for 160 credits, Pro $25 for 500 credits, Pro Max $50 for 1,100 credits. Every plan includes the unlimited models from walkthrough one, and every credit model from walkthrough two. Credits roll over while you stay subscribed.
In walkthrough two the talking clip was 30 credits, the extension 13.5 credits and the captions 5 credits. Each tile shows its own price.
Under our refund policy, a first purchase can be refunded within 7 days, in full if under 20% of the credits are used and pro-rated after that.
How to Make AI Videos questions
How do I make an AI video step by step?
Write the idea as one shot, make the opening still, animate it from that frame, add dialogue or sound in the prompt, extend it from the last frame, then add captions. This page shows a real run of every step.
Can I make AI videos for free?
On any paid Pixel Dojo plan, MiniMax H3 Turbo videos are unlimited, up to 5 seconds at 720p, with no credits spent. Without a plan you get 3 free effects a day, with a watermark.
How do I make an AI video talk?
Put the line in double quotes in the prompt and say who speaks it. MiniMax H3 Turbo and Seedance 2.5 both said our line word for word, with the voice built into the clip.
How do I make an AI video longer?
Start a new clip on the last frame of the one before, as in walkthrough one, or use the video extender, as in walkthrough two. Both continue the same shot.
Do I need editing skills to make AI videos?
No. Each step here is a prompt. For a longer project, Sensei or Film Studio plans the shots and edits them for you.
Video Effects New
Your first video is free. One photo, one click.
310 ready-made effects, from AI Hug and Earth Zoom In to product spins and listing walkthroughs. Upload a photo and press Generate. No prompt to write, and a free account gets 3 a day.
Make your first MiniMax H3 Turbo clip
Write a prompt above or start from any example on this page. One plan covers every top model, six studios and Sensei, and your credits roll over.
Start with MiniMax H3 Turbo