# go-ahead

- **Event:** [fal x Sequoia 72-Hour Video Hackathon](https://cerebralvalley.ai/e/fal-x-sequoia-virtual-hackathon)
- **When:** Jul 17 at 12:00 PM – Jul 19 at 8:00 PM (PDT)
- **Where:** Online
- **Team:** [Christopher Pietsch](https://cerebralvalley.ai/u/chrispie)
- **GitHub:** https://github.com/cpietsch/fal-hackathon-submission
- **Demo video:** https://youtu.be/WR3kKBcK3Ro
- **Gallery:** https://cerebralvalley.ai/e/fal-x-sequoia-virtual-hackathon/hackathon/gallery
- **Page:** https://cerebralvalley.ai/e/fal-x-sequoia-virtual-hackathon/hackathon/gallery/86

Text is a terrible language for camera work. Nobody says "dolly in 1.4 meters with a soft ease and subtle handheld sway" — but every director can walk that move. Blocking makes your phone the camera and hands your real movement over to the AI model.

How it works: A single gray cube stands on a 3D stage. Click it and tell it — by voice or text — what it is ("a rusty 1950s jukebox with glowing neon rims"). The description attaches to the main prompt like an image in a chat composer. Then pair your Android phone via QR: the moment you hit START, the desktop switches into the film camera, counts down 3-2-1, and your phone's 6DoF pose (WebXR, streamed at 60 Hz over WebSocket) is the camera. Walk a dolly, arc around the cube, crane up — amplified 5× so a living room becomes a film set.

The discovery: the performed take is translated into cinematographer language deterministically — dolly/truck/pan/orbit distances, easing and handheld shake are measured from the trajectory in code, phrased as one DP sentence, and injected into the prompt for Kling on fal. The model receives your actual camera intent, not a guess. Takes land as browser-style tabs; an After-Effects-style curve panel lets you smooth the move or drag bezier handles at its peaks before generating. Object + motion + reference images + atmosphere text compose into one prompt — one send, one shot, previz and result side by side.

---

Markdown version of https://cerebralvalley.ai/e/fal-x-sequoia-virtual-hackathon/hackathon/gallery/86. Site index for agents: https://cerebralvalley.ai/llms.txt · full text: https://cerebralvalley.ai/llms-full.txt