Choose your first frame
Upload your own image or start from one of six original stills. The opening frame is the single biggest predictor of a usable clip.
The studio · free · no account
Pick a still, describe the shot, choose a camera move and export a clip. Pavo AI Video renders every frame on your device — there is no queue, no credit balance and no upload. The result is a real WebM file you can download.
01 · Source still
Sample still · Product
02 · Shot prompt
03 · Render settings
Everything above runs on your device. Your image and prompt are never uploaded, and no account or credits are involved. How rendering works →
Three steps
No prompt engineering marathon. One decision per control, and every setting can be changed and re-rendered in seconds.
Upload your own image or start from one of six original stills. The opening frame is the single biggest predictor of a usable clip.
Write a short shot prompt. The studio reads it for motion words and suggests a camera path, which you can accept or override.
Pick 16:9, 9:16 or 1:1 and a 3, 5 or 8 second duration. Generate, preview, then save the WebM to your device.
Plain speaking
We would rather be precise than call a local motion renderer a large video model.
| It does | It does not | What that means for you |
|---|---|---|
| Animate a still with six camera paths and a light sweep | Invent new objects, scenes or talking characters | Geometry, faces and hands stay exactly as supplied — no generative distortion |
| Record 30 fps WebM locally at up to 1280×1280 | Encode MP4 in the browser | WebM plays in modern browsers and editors; transcode to MP4 if a platform demands it |
| Keep your image and prompt on your device | Send anything to a model API or a render queue | No waiting, no quotas, no third-party data handling |
| Export a clean plate with an optional caption bar | Bake logos or generated lettering into the frame | You keep full control of titles and grading in your editor |
Need a starting point? Load one of the 12 shot presets or write a structured brief with the six-line prompt builder. Full technical detail is in How rendering works.
Privacy by construction
The studio uses the Canvas 2D API and MediaRecorder. There is no upload endpoint, no account database and no analytics script watching your prompt. See the privacy policy for what a normal web server still logs when you load a page.