Product

The script-to-shot operating system.

One continuous pipeline. Story enters as text. It leaves as approved frames and controllable AI video — with a reusable world behind every shot.

01
Script upload
02
LLM scene extraction
03
World generation
04
Spatial layout
05
Virtual camera
06
Frame approval
07
AI video generation
08
Export packages
STEP 01
Script upload

Drop a screenplay, treatment, or beat sheet. Multiple formats supported.

Capabilities
  • FDX · Fountain · PDF · plain text
  • Auto-detected scene boundaries
STEP 02
LLM scene extraction

Scenes, beats, entities, costumes, locations, lighting, and implied camera intent.

Capabilities
  • Entity graph
  • Wardrobe + props inventory
  • Intent annotations
STEP 03
World generation

A coherent visual world is generated — environment, palette, mood, and key dressing.

Capabilities
  • Reference-conditioned
  • Style-locked across scenes
  • Editable plates
STEP 04
Spatial layout

Scenes resolve into navigable 3D space. Move through them. Place talent. Block action.

Capabilities
  • Three.js viewport
  • Gaussian splat support
  • Imported assets
STEP 05
Virtual camera

Real lens math. Real blocking. Choose focal length, height, aperture, and movement.

Capabilities
  • Lens library
  • Dolly · crane · handheld
  • Continuity checks
STEP 06
Frame approval

Lock the frames that define the shot. Approved frames become the generative contract.

Capabilities
  • Side-by-side compare
  • Approve / reject
  • Versioned takes
STEP 07
AI video generation

Video models conditioned on approved frames, blocking, and motion intent.

Capabilities
  • Per-shot generation
  • Continuity-aware
  • Re-run with edits, not re-prompts
STEP 08
Export packages

Ship per-shot exports — frames, plates, camera data, EDLs, world references.

Capabilities
  • Per-shot ZIP
  • Editorial-ready
  • Reusable world links
Request access

See it on a real script.

We're previewing the full pipeline with selected studios and creators.