furybot

Steps

1. Decide whether you are starting from a photo. With no photo, the clip is generated from your description alone. Give it a start photo and the clip continues that exact picture — same person, same room, same light. If you care what the subject looks like, start from a photo.
2. Describe motion, not just the scene. A still-image prompt describes what is in frame. A video prompt has to say what changes: who moves, which way the camera goes, what happens over the seconds. "A woman on a balcony" gives you a photograph that trembles; "a woman on a balcony turns to look back into the room, camera slowly pushing in" gives you a shot.
3. Pick a length. Between two and ten seconds, depending on your plan. Render time scales roughly in proportion, so use two seconds while you are still finding the shot.
4. Add narration if you want it. Type a line and it is spoken over the clip by a voice running on the same machine — nothing is sent to a speech service. You can upload your own audio track instead.
5. Choose MP4 or GIF, and download. MP4 is better for sharing and keeps the audio. GIF loops anywhere but is silent and much larger. Download it: video results are swept on the same short schedule as everything else.

Tips

• Iterate at two seconds, then re-run the prompt you settled on at full length. Ten-second drafts are how a weekly allowance disappears.
• One clear action per clip. Two things happening at once is where the frames start disagreeing with each other.
• Hands, text and fast motion are still the weak points across every video model, ours included. Compose around them rather than hoping.
• Photo-to-video is the reliable route to a consistent character across several clips: generate one still you like, then animate it more than once.

Limits worth knowing before you start

• Two to ten seconds, by plan. This is not a tool for a finished film — it is a tool for a shot.
• Expect minutes, not seconds. The queue is real and video sits at the expensive end of it.
• Narration is generated on the same infrastructure as everything else. It is not a voice clone of a specific person, and imagery of real people carries an AI-generated label.
• Nothing is archival. Uploads are deleted as soon as the job finishes, generated files are swept automatically after about a day, and the prompt text is blanked on the same schedule. Download what you want to keep — this is a workshop, not an archive.
• Adult content is absent from this product rather than switched off. There is no setting, plan, administrator action or support request that introduces it, on the site or on the API.
• Four things are refused on every plan, each one traceable to a law rather than to our taste. They are listed in full, with the statute behind each, on the safety page.

Questions

How long can a clip be?

Between two and ten seconds, depending on your plan. Longer clips take proportionally longer to render.

MP4 or GIF?

Both are offered. MP4 is better for sharing and keeps audio; GIF loops but is silent and much larger.

Can I animate a photo I took?

Yes. Supply it as the start frame and the clip continues that picture rather than inventing a new subject.

Account, billing and retention questions are on the FAQ.