Video2Storyboard reads any reference video and returns a production-ready board: every shot, its keyframes, and a prompt describing character, action, camera and light — ready to remake with your own characters.
Automatic shot cuts (PySceneDetect, with a learned detector for fast cuts) split the video into shots with timecodes.
First, middle and last frame of every shot — the visual truth of the original.
A local vision model (Qwen3-VL) writes a structured spec per shot — no cloud, no uploads.
ComfyUI regenerates each shot with your character (face-locked), keeping composition and performance.
Approved frames go through Motion Lab pose replay to become the finished remake.
The board imports straight into ToonFlow for scripting, voices and publishing.
{
"shot": 7,
"timecode": "00:12.4-00:15.1",
"character": "woman, 30s, black blazer",
"action": "turns from window, walks to desk",
"camera": "medium shot, slow dolly-in",
"light": "warm practical, left side",
"keyframes": ["s07_a.jpg", "s07_b.jpg", "s07_c.jpg"]
}
Video2Storyboard ships inside the Motion Lab Workflow Pack and is included free with every Full remake order. Join the waitlist on the Motion Lab page, or order a remake and ask for the board.
Join the waitlist