a barman at closing sets down the glass, looks up and says, "last one"
Prompt onlyVideo that arrives already finishedSound included
Six ways in. Words, a frame, two frames, a face, a voice.
Describe a scene, or hand it a face and a voice. What comes back is a finished shot — the performance, the room, and the sound of both — not a silent clip you still have to score.
First four seconds free
Nine reference images. Three audio clips. The same face and the same voice, take after take.
Six ways to start. Pick whichever you already have.
frame 1 — yours
frame 1 — yourslast frame — yoursthe model fills this9 images · 3 clips
1 photograph+1 voice clipup to 3 voice clips
Nothing is locked behind a plan.
Change any of them per render. What you pay follows the size and the seconds, and nothing else.
- How long
- Four to thirty seconds a shot
- How it looks
- Up to 1344×768 at 24fps, any shape you need
- How it sounds
- Speech, score and room tone, already mixed
- What you get
- An MP4, the moment it finishes
Why people come back.
- 01
Get the same take back
Reuse a seed and the same take comes back frame for frame. Raise the step count on the one shot that has to hold up, and leave the rest cheap.
- 02
Watch it happen
Position, elapsed time and the credit cost while it runs. The file is downloadable the moment it lands, at the size you asked for.
- 03
It arrives with sound
Dialogue, score and room tone are written with the picture, in one stereo mix. There is no second model and nothing to line up afterwards.
- 04
Pay for seconds, not seats
Charged when a render is queued, at the rate for the size you picked. No plan, no seats.
- 05
A failure costs you nothing
A failed render returns every credit it took, in full, without a ticket.
The first four seconds are free. The failed ones are free too.
Sign in with Google and the account exists. Credits come in packs and are spent by the GPU second, at the rate for the size you picked.