Turn your words into film.
Say what happens, in the chat. The Director turns it into a Prompt you can change, and the film is made when you press Generate.
An old fisherman mends a net on the dock at dawn. Gulls call overhead, and the camera drifts in until his hands fill the frame.
Three examples
A character
An old fisherman mends a net on the dock at dawn. Gulls call overhead, and the camera drifts in until his hands fill the frame.
The words The film The whole ask
An old fisherman mends a net on the dock at dawn. Gulls call overhead, and the camera drifts in until his hands fill the frame.
A product
A slow turn around a perfume bottle on black stone. Light runs along the glass, and a drop slides down its side.
The words The film The whole ask
A slow turn around a perfume bottle on black stone. Light runs along the glass, and a drop slides down its side.
A landscape
Fog rolls through a pine forest at sunrise. The camera glides low between the trunks as the light breaks through.
The words The film The whole ask
Fog rolls through a pine forest at sunrise. The camera glides low between the trunks as the light breaks through.
Three steps, in one chat.
Say what happens
Describe the scene in the chat, in your own words: who's in it, where it is, and how the camera moves.
Read the Prompt
The Director writes the Prompt from your words, and on Auto picks the model and says why. Every part of it can be changed before anything runs.
Generate
The card shows the model, the length and the price on Generate. No render is charged until you press it.
31 models can make your scene.
Six of them are here, side by side. They differ in how long the film runs, whether it has sound, and what it costs.
| Model | Lengths | Sound | Starts at |
|---|---|---|---|
| Kling 3.0 Pro | 3–15 s | No | 68 creditsa 5-second take at 1080p |
| Seedance 2.0 | 4–15 s | Yes | 154 creditsa 5-second take at 720p |
| Veo 3.1 | 4, 6, 8 s | Yes, with speech | 27 creditsa 4-second Lite take at 1080p |
| H3 | 4–15 s | Yes | 30 creditsa 5-second take at 768P |
| Seedance 2.5 | 4–30 s | Yes | 105 creditsa 5-second take at 480p |
| Wan 3.0 | 2–30 s | Yes | 30 creditsa 5-second take at 480p |
Every one of these six can also start from a picture of yours, as Image to Video does. Each starting price is a first take from your words.
The practical details
AI Video Generator, in practice
- What it takes
- Your words, in the chat. Every one of the six above can also start from a picture of yours, as the film's first frame.
- How long
- From 2 to 30 seconds on the six above, depending on the model.
- Its shape
- 16:9 or 9:16 on every one of the six above, and 1:1, 4:3, 3:4 or 21:9 on some. On Auto, Thisbe picks the frame from your words.
- Sound
- Five of the six above make sound with the film, and Veo 3.1 speaks a line you write in quotes.
- What it costs
- Priced by the second or by the clip, depending on the model. The card shows the price on Generate before anything runs.