Flow Veo 3 Apps
Choose a category
Model
Grok Imagine
2 models available
Wan
2 models available
2.7 Text to Video
2.7 Image to Video
Hailuo
2 models available
Wan 2.7 Image to Video
First and last frame control, plus audio-driven motion.
Frames
First frame is required. Add a last frame to control where the shot ends.
Creative Vision
Video Parameters
Cinema Preview
Generation History0 records
Wan 2.7 Image to Video Generator
Wan 2.7 image to video takes a first frame and, if you want, the last one too, then generates the motion in between at up to 1080p. Add a driving audio track for lip sync or beat-matched movement, or hand it an existing clip and have it continue from where that clip ended.
Best for controlled shots: a defined start, a defined end, and motion that has to land on a specific beat
The Most Controllable Image to Video Model Here
Four separate inputs shape the result: first frame, last frame, driving audio and a starting clip. Only the first frame is required, and every other one narrows down what the model is allowed to invent.
First Frame Control
Your image becomes frame one, up to 30MB in JPEG, PNG or WebP. Composition, colour grade and subject identity all carry over from the still you supply.
First and Last Frame
Add a closing image and the model works out the path between the two. This is how you get a shot that arrives exactly where you need it instead of drifting somewhere plausible.
Audio-Driven Motion
Upload a track up to 50MB and movement follows it: mouth shapes for dialogue, body motion for a dance, cuts and gestures that land on the beat.
Continue an Existing Clip
Supply a starting clip and generation picks up from its final frame, which is how you build a longer sequence out of several runs instead of one impossible take.
720p or Native 1080p
Draft at 720p from 8 credits, then re-run the settings that worked at 1080p. Duration is a per-second slider from 2 to 15 seconds either way.
Seeds and Negative Prompts
The same controls as the text-to-video model: a fixed seed for repeatable runs and 500 characters of negative prompt to rule out artefacts you keep seeing.
Define the Start, Define the End
Two stills and one sentence give the model far less room to guess than a prompt alone. That is the whole reason to use image to video rather than text to video.
Four Ways to Drive a Wan 2.7 Shot
The same page covers four different jobs, depending on which optional inputs you fill in. Pick the lightest one that gets you the shot.
First Frame Only
One still plus a prompt. The cheapest and fastest route, and the right choice when you care how the shot starts but not exactly how it ends.
First and Last Frame
Two stills bracket the motion. Ideal for product turns, before-and-after reveals, and any transition that has to end on an exact frame.
Audio-Driven
A portrait plus a voice track for talking-head clips, or a character plus music for dance. Clean, single-speaker audio gives noticeably better sync than a busy mix.
Clip Continuation
Feed in a video up to 30MB and generate what happens next. Chain two or three runs together when a single 15-second take is not long enough.
Why There Is No Aspect Ratio Menu
The output shape is taken from your frames, so cropping the first frame is how you choose the framing. A ratio selector here would be a control that does nothing.
Wan 2.7 Image to Video
FAQ
Frames, driving audio, clip continuation, resolutions and credit costs for Wan 2.7 image to video.
Try Wan 2.7 Image to Video
Upload one frame and run 2 seconds at 720p for 8 credits to see how the model reads your image.