Generate video
Generate a clip from a prompt, a first frame or references, and pick the video model that fits the job.
You generate a new clip in the generate bar. Light can set up the same shots for you if you'd rather describe the video in plain words. See Ask Light.
Generate a video
Open + at the top of the left toolbar and choose Generate Video. An empty video shot appears on the board and the generate bar opens in the side panel.
- Pick a video model. On a free plan, the generate bar locks video models.
- Write the prompt. Describe the action and the camera, like "slow turn of a silver ring on black velvet, soft window light, macro, the camera pushes in".
- Set the frame and length. Choose the aspect ratio, like 9:16 for a reel or 16:9 for a brand film, then the resolution, like 720p or 1080p, and the duration. Each model has its own list of durations.
- Turn sound on or off with the speaker button, on models that make their own audio.
- Add a first frame, an end frame or references if the model takes them. The clip starts on the first frame and ends on the end frame.
- Check the cost on the run button, which shows the credits next to an arrow, and press it.
The shot shows Generating… while it works, then the clip takes its place. Video takes longer than a still.
Start from a still
Video on an image menu starts a video shot with that image as the first frame. Pick a model and run it, and the still of your silver ring becomes the opening frame of the reel.
Start from references
A reference video keeps the same character, product, style or scene across clips. Ask Light for it, or turn on Omni in the generate bar and add one or more references. Seedance, Happy Horse 1.0, Veo 3.1, Kling O3 Pro and Gemini Omni Flash are some of the models that take references.
Pick a video model
- For a clip with dialogue, ambience or music in the same take, use Veo 3.1, Seedance, Kling V3 Pro or Grok Imagine Video 1.5. On Kling V3 Pro, sound stays off until you turn it on with the speaker button.
- To keep the same character or product across clips, use a model that takes references, like Seedance, Happy Horse 1.0, Veo 3.1 or Gemini Omni Flash.
- For a fast clip, use Grok Imagine Video 1.5.
- For a face that speaks a voice track, use Chromatic Lipsync or Kling V2 Avatar Pro. See Extract frames and audio.
Last-frame and duration controls appear when the model supports them. Seedance 2.0, for example, takes a first and last frame and a duration from 4 to 15 seconds, and Seedance 2.5 goes up to 30 seconds. Veo offers a shorter list of 4, 6 or 8 seconds. The full list is in Models.
What it costs
Video models need a paid plan. Longer and higher resolution clips cost more, and the run button shows the price before you start. If a clip fails, the credits go back to your wallet. See Credits and When a run fails.