Vidu Q3 Pro First-Last Frames interpolates a smooth, cinematic transition between two key images — your start frame and end frame — guided by a text prompt. Perfect for transformation reveals, scene transitions, product morphs, and storytelling beats that need a clean, controlled arc from A to B.
About this model
Vidu Q3 Pro First-Last Frames interpolates a smooth, cinematic transition between two key images — your start frame (image_url) and end frame (last_image) — guided by a text prompt. Perfect for transformation reveals, scene transitions, product morphs, and storytelling beats that need a clean, controlled arc from A to B.
Cost analysis
| Provider | Cost | Notes |
|---|---|---|
| muapiapp | $0.15 per second at 720p ($0.75 per 5s clip) | Per-second pricing — 360p $0.05/s, 540p $0.07/s, 720p $0.15/s, 1080p $0.16/s. |
| Fal.ai | Not available | Vidu Q3 Pro first-last frames is not listed on Fal.ai at this time. |
| Replicate | Not available | Vidu Q3 Pro first-last frames is not listed on Replicate at this time. |
Per-second pricing — 360p $0.05/s, 540p $0.07/s, 720p $0.15/s, 1080p $0.16/s.
Vidu Q3 Pro first-last frames is not listed on Fal.ai at this time.
Vidu Q3 Pro first-last frames is not listed on Replicate at this time.
** Competitor pricing is estimated based on similar model architectures and usage tiers.
Configuration schema
| Parameter | Type | Description | Default |
|---|---|---|---|
| Prompt | string | Text prompt describing the transition. | The frozen bird begins cracking from within as glowing orange light shines through the ice. Steam bursts outward while flames ignite across the wings. The sculpture violently shatters apart and transforms into a blazing phoenix that launches upward through fire and smoke. |
| First Image URL | string | URL of the starting (first) frame image. | https://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/vidu-q3-pro-first-last-frames-1.jpg |
| Last Image URL | string | URL of the ending (last) frame image. | https://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/vidu-q3-pro-first-last-frames-2.jpg |
| Resolution | Enum (4 options) | The resolution of the generated video. | 720p |
| Aspect Ratio | Enum (5 options) | Aspect ratio of the output video. | 16:9 |
| Duration | int | The duration of the generated video in seconds. | 5 |
| Audio | boolean | Whether to generate audio for the video. | false |
Text prompt describing the transition.
The frozen bird begins cracking from within as glowing orange light shines through the ice. Steam bursts outward while flames ignite across the wings. The sculpture violently shatters apart and transforms into a blazing phoenix that launches upward through fire and smoke.URL of the starting (first) frame image.
https://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/vidu-q3-pro-first-last-frames-1.jpgURL of the ending (last) frame image.
https://d3adwkbyhxyrtq.cloudfront.net/webassets/videomodels/vidu-q3-pro-first-last-frames-2.jpgThe resolution of the generated video.
720pAspect ratio of the output video.
16:9The duration of the generated video in seconds.
5Whether to generate audio for the video.
falseDeveloper documentation
Provide both keyframes: Pass image_url (the first frame) and last_image (the final frame). The model will interpolate between them.
Write a transition prompt: Describe what happens between the two frames — the motion, the transformation, the camera move.
Pick resolution and aspect ratio: 1080p for finals, 720p for day-to-day work. Match the aspect ratio to your source images.
Set duration: Choose 1–16 seconds. Longer durations give more room for the transition to breathe.
Submit: Async endpoint — poll /api/v1/predictions/{request_id}/result or use a webhook_url.
Frequently asked
For the cleanest results, yes — match both images and the requested `aspect_ratio` to avoid cropping or letterboxing.
Q3 Pro is tuned for smooth, prompt-guided interpolation. The longer the duration you pick, the more time the model has to ease between the two states.
No — first-last frames requires both. If you only have a starting image, use the Image-to-Video endpoint instead.