Generate video on SomaLabs
On plans that include video, SomaLabs creates short clips two ways: text-to-video from a written prompt, and image-to-video that animates a still you already made. A scene builder breaks the clip into start frame, motion, and end frame so you can describe how the shot opens, what moves, and how it settles.
Video lives in the same studio as your images, so you can move from a still into motion without a second tool. Video is available on eligible plans, where your plan includes it. This guide covers the two ways to create a clip - from text and from an existing image - and how the scene builder structures a shot.
Two ways to make a clip
- Text-to-video: describe the shot in words and the model generates a clip from scratch.
- Image-to-video: start from a finished still and animate it into a short motion clip.
Animate a still you already made
Image-to-video is often the easiest path to a good clip: you already have a frame you like, so you are only describing the motion. Take a finished still from your gallery, use it as the starting frame, and write what should move - a slow turn, a gentle camera push, a shift in pose. Because the clip starts from an image you approved, you spend your effort on the movement rather than rebuilding the whole shot.
Use the scene builder
- Start frame: describe the opening moment - what the shot sees as the clip begins.
- Motion: describe what changes - the action, the movement, and any camera move.
- End frame: describe how the clip settles into its final pose.
- Set the duration, confirm the credit cost, and generate.
Video results can still vary, and motion is harder to control than a still, so plan to iterate on the Motion description in particular. Keep the movement simple for your first clips - one clear action reads better than several competing ones - and build up from there.
See which plans include video