Shot Planner — storyboard first, render second
Plan every AI shot before spending a cent. Each shot = a start frame, an optional end frame, and one sentence of direction. The planner prices the whole board and writes the exact commands.
The primitive this wraps: Gemini Omni Flash (
gemini-omni-flash-preview, $0.10/s) accepts one image (video starts on that frame) or two images (video travels from frame A to frame B). Below is the proof — two real stills from Jordan's welcome shoot, and the camera move the model built between them.
Proof — two stills in, one camera move out
Start frame · welcome shoot @ 4s
End frame · welcome shoot @ 28s
The generated move · 8s · $0.80 · identity held throughout
The board
Edit anything — costs and commands update live. Saved in your browser automatically.
Render commands — copy and run
Batch file for Google's official generate_video.py (from google-gemini/gemini-skills → gemini-omni-flash-api). Save as jobs.json, then: python video/generate_video.py --batch jobs.json --concurrency 2