No renders yetDescribe a topic on the left and generate your first video.
Settings
The model writes the narration and decides what gets drawn. Groq is fast and free to start — create a key at console.groq.com/keys and paste it below.
When an image API is configured, the model may place photographs on the board.
It writes a short visual query — “honeybee on a lavender flower” — and the
engine resolves it here, caches the result, and reuses it on later renders.
The model never sees your key.
If your endpoint returns the image bytes directly, leave the result path empty.
Narration can come from your machine's own synthesiser — free, offline, no
key — or from a cloud provider. Fish Audio's S2.1 Pro free tier
is free under fair use; get a key at fish.audio/go-api.
Browse voices
Most of this library is community-uploaded, including clones of real
people and characters. Check the rights before publishing commercially —
“rights-cleared” narrows it to voices Fish has secured, currently a small,
mostly Japanese set. Choosing no voice uses Fish's own default.
Upload a picture of a hand holding a pen (PNG or WebP with a transparent
background), then click the image to mark exactly where the pen tip is.
That point is what gets placed on the ink, so the hand holds the line instead of
floating beside it.
Upload a hand imagePNG or WebP with alpha, up to 12 MB
No custom hands yet — the built-in vector hand is used.
The duster
What sweeps the board clean between scenes. Turn it off and the board still
clears — there is simply nothing shown doing it.
Upload a duster imagePNG or WebP with alpha — click it afterwards to mark the face that touches the board
No uploaded dusters — the built-in one is used.
Images available to the board. Anything you upload here can be requested by name
or keyword; anything fetched from your image API is cached here automatically.
Add imagesPNG, JPEG or WebP — keywords come from the filename