Image to video
Your image becomes the first frame and the model takes it from there. It is the fastest route to a product clip, a moving portrait or an animated illustration, because the look is already decided — you are only directing the motion.
Made with this model
Every clip below was rendered here, on the settings shown under it. The caption is the exact prompt that produced it.
Seedance 2.0 · 720p · 16:9 · 4s
MiniMax H3 · 2K · 16:9 · 5s
How image to video works here
Upload the first frame
JPEG, PNG or WebP. Whatever is in the picture — a face, a bottle, a landscape — is what the model has to keep consistent.
Describe the motion only
Not the scene. “Steam rises and the camera pushes in” works; re-describing the photo fights the image you just uploaded.
Render
Seedance 2.0 honours the aspect ratio you pick; MiniMax H3 follows the shape of your image. Both hand back an MP4.
Specs at a glance
| Model | Duration | Resolution | Native audio | Reference images | From |
|---|---|---|---|---|---|
| Seedance 2.0 | 4–15 seconds | 480p · 720p · 1080p · 4K | Yes | Not used | 38 credits for 5 seconds |
| MiniMax H3 | 5–15 seconds | 2K | No | Not used | 130 credits for 5 seconds |
Getting motion that holds up
Use a clean, well-lit source. Motion amplifies compression artefacts and blur you would never notice in a still.
Ask for less than you think. A slow push and drifting steam reads as real; a full camera orbit rarely does.
Leave headroom in the frame. Anything cropped tight against an edge has nowhere to move into.
For products, name the material. “Condensation on cold glass” tells the model what the light should be doing.