Choose a category
3 models available
4 models available
3 models available
2 models available
4 models available
8 models available
Seedance 2.5 First to Last Frame
Pin both ends and let it plan the move between them.
Frames
Both are required here. Only have a start? Use Image to Video instead.
Creative Vision
Describe how it gets from the first frame to the last one.
Video Parameters
Priced per second — 30 seconds costs 7.5× what 4 does.
720p costs 2.25× what 480p does. Draft at 480p.
Adaptive keeps your first frame's shape — usually what you want.
AI audio synced to the video. Does not change the price.
Cinema Preview
Generation History0 records
Seedance 2.5 First to Last Frame
Give Seedance 2.5 the image it starts on and the image it ends on, and it plans the whole move in between — up to 30 seconds, at 480p or 720p, with audio generated in the same pass. Pinning both ends is the most reliable way to get a long clip that arrives somewhere instead of wandering.
Both frames are required here — if you only have a start, use the Image to Video page
Why Choose First to Last Frame
An open-ended generation decides where it is going in the first second and commits. Supplying the destination changes the problem from "invent something" to "get from here to there", which is a far easier thing to get right.
Both Ends Are Fixed
Upload a start image and an end image. The model interpolates a path between them rather than choosing a direction on its own — the difference between a shot that drifts and a shot that lands.
Up to 30 Seconds Between Them
4 to 30 seconds, one second at a time. The same two frames can produce a fast cut or a slow transformation depending purely on the length you pick, and the price scales with it linearly.
Adaptive Keeps Your Framing
The ratio defaults to Adaptive, so the output matches the shape of the frames you uploaded. Use images with the same dimensions at both ends and the result stays consistent throughout.
Audio Included, Free
Sound is produced with the picture in one pass and does not change the price. It is on by default; turn it off only when you specifically want a silent clip.
Reference Video and Audio Still Apply
Under Advanced you can attach a clip of up to 30 seconds to lend its motion, or an audio track of up to 30 seconds to drive speech and rhythm. A reference video also lowers the price by about 40%.
Failed Runs Are Refunded
Credits are held at submission and returned automatically if the generation fails or the provider never answers. Long clips are the expensive ones, which is exactly why this is not optional.
For Reveals, Transformations and Transitions
Anything that has to hit a specific final composition — a product turning to face camera, a room changing season, a logo assembling — is a start-and-end problem, not a prompt problem.
How to Use First to Last Frame
The two images define where the shot begins and ends. The prompt defines the route — and over 30 seconds, the route is most of the work.
Upload the First Frame
JPEG, PNG or WebP up to 30MB. This is the opening composition and it sets the output's shape, so use the framing you actually want rather than a crop of it.
Upload the Last Frame
Same formats and limits, and required on this page. Frames that share a subject and a rough camera position give the cleanest interpolation; two unrelated images will produce a transition rather than a move.
Describe the Route Between Them
"The camera pushes in slowly as the light shifts from dusk to night" tells the model how to spend the seconds. Without it, it will choose a pace for you — usually a duller one. The box holds 20,000 characters.
Pick Length, Then Resolution
Try the move at 480p and 4 seconds for 30 credits, since length changes the whole feel of an interpolation. Once the pace is right, re-run it longer and at 720p for delivery.
Build Longer Sequences
Use the end frame of one generation as the start frame of the next and the whole sequence stays continuous — 30 seconds at a time, with each transition under your control.
Seedance 2.5 First to Last Frame
FAQ
Frame requirements, pacing, pricing and how this page differs from plain image to video.
You supply the image the clip starts on and the image it ends on, and the model generates the movement between them. On this page that is ByteDance's Seedance 2.5, which can spread that movement over as much as 30 seconds. FlowVeo3 is an independent platform and is not affiliated with ByteDance.
Yes, on this page. With only a start frame this would be the Image to Video page, which is a separate mode — same model, same price, open ending. Requiring both here is what makes the two pages do different things.
They do not have to, but the closer they are in subject and framing, the more the result reads as one continuous shot. Two unrelated images still work — you just get a transition between them rather than a camera move within a scene.
Between 30 and 430 credits, set by resolution and length: 30 credits for 4 seconds at 480p, 75 for 5 seconds at 720p, and 430 for a full 30 seconds at 720p. A reference video attached under Advanced lowers all of those by about 40%. The exact cost is shown before you submit.
A great deal. The same pair of frames interpolated over 4 seconds is a snap; over 25 seconds it is a slow reveal with room for the model to invent intermediate action. Length is a directing decision here, not just a duration.
Your frames', by default — the ratio control is set to Adaptive. Upload both images at the same dimensions for the cleanest result, and override the setting only when a delivery format requires a specific shape.
JPEG, PNG and WebP, up to 30MB per frame. Output resolution is set separately by the 480p or 720p control, so a large source image does not by itself produce a larger video.
You are not charged. Credits are held when you submit and released back to your balance if the model fails or the provider never returns a result. Generated videos stay available for seven days, so download anything you want to keep.
Try First to Last Frame
Upload where it starts, upload where it ends, and let the model handle the thirty seconds in between.