What it is Mise is an AI cinematography tool for BookTok and YouTube creators. Speak or type a book review (up to 50 books per video), and Mise directs the video inside a persistent 3D library: it glides the camera across your bookshelves, pulls each named book out into a format-aware Hero Zone facing the viewer at a clean 3D angle, holds dedicated Negative Space for your talking-head overlay (9:16 BookTok or 16:9 YouTube), opens the 3D book while you talk to flip through your Star Rating, Personal Review, and Highlighted Quote before closing back to the front cover, and surrounds the book with high-DPI 3D effects and living pop-up dioramas that frame the outer border without ever covering the title or author's name. Then it watches its own rendered WebGL frames, critiques them against the plan, and automatically patches and re-renders any shot that missed. Built at the Gemini 4 Argon SF Hackathon (Cerebral Valley x Google DeepMind), October 25, 2026, in the More Dimensions track. Built with Google Antigravity and Gemini 4 Argon (2,097,152-token context window). The problem BookTok and YouTube creators who read on e-readers have no physical shelf to film in front of. Creators with shelves get a visual identity for free. E-reader readers do not. The audience already exists; the demand is latent. They do not know they can have a library that performs while they just read and talk. Who it is for BookTok and BookTube creators making review videos, especially e-reader readers with no physical shelf. The library in the video persists across videos like a real room, so viewers can audit the shelf. That continuity is the trust mechanism. What was built today versus what was pre-existing Pre-existing (not in this repo): The 3D library room itself, built earlier with Claude Code. It is the stage, not the demo: simple, persistent, already made.
Team
