A studio for short generative clips. Describe a scene — or upload a single frame — and VXLABS renders a watermark-free MP4 you can download and post. Two engines behind one queue: one for text, one for animating stills.
Read this before a long render. These are hard limits of the engines, not settings we hide.
One continuous shot of up to 81 frames at 16 fps — five seconds. The model can technically do longer, the free GPU cannot do it in a sane time, so it is capped. Need more? Render two clips and cut them together.
Renders run on a single free GPU, so a 5s clip takes 15–30 minutes, up to ~45 at peak, one at a time. If nothing is warm the box boots first — another 2–6 min. Keep the tab open; this page retries on its own.
One still only — no video in, no multiple references, JPG/PNG/WEBP under 8 MB, sent straight from your browser. The first frame stays fixed and motion grows from it, so faces and hands can still warp.
Subject, action, camera, light — in that order. It is the same model in both modes, so a prompt works for a bare scene and for an uploaded frame alike. On-screen text and logos rarely survive; over the cap the request is rejected, not truncated.
Clips come back with no audio track at all. Add music or voice in an editor afterwards; there is no lip-sync and no voice cloning here.
The box draws on a fixed weekly GPU allowance, and shuts down after 12 idle minutes. One job at a time: a second request waits, it is not queued twice. When the allowance is gone, the studio says so instead of hanging.
| Clip | Typical wait | Peak wait | Output |
|---|---|---|---|
| Box asleep | +2–6 min | +10 min | boot, then it stays warm |
| Text → video | 15–30 min | ~45 min | MP4 · 480p · 16 fps |
| Image → video | 15–30 min | ~45 min | MP4 · 480p · 16 fps |
| Queue | 1 job | serial | second request waits |