Blocking (3D)

3D blocking (previz) is where you stage characters, props, and cameras in a 3D space before shooting. What you place here does not stay a sketch — figure positions, poses and outfit colors, props and practical lights, environment/fog/weather, and even who is riding what are all written into the AI video generation prompt automatically: the blocking becomes the prompt.

1. 1Place figures (man/woman/child), props (54 types — cars, weapons, furniture…) and lights from the left asset panel by clicking or dragging into the viewport. Press D to duplicate the selection; star an asset to pin it as a favorite.
2. 2Load one of 10 scene templates (two-person OTS dialogue, corridor chase, elevator, crowd, car ride, truck cab, motorcycle ride, police car chase…) by clicking its thumbnail. Hover a thumbnail to play its preview clip.
3. 3Open the AI situation composer and type something like "rainy subway platform, late night" — set dressing, practical lights, environment, fog and rain/snow are composed at once. Type @ to mention characters and places as highlighted chips; pick "start from a scene" to fill the prompt with a scenario scene's title, location, cast and first line. Click a chip to swap or remove it, and unknown names can be generated as 3D objects on the spot.
4. 4Select a figure and choose a vehicle (car, truck, motorcycle, bicycle, horse, boat) under Ride in the inspector — the figure sits on the seat (motorcycles/bicycles grip the handlebars). From then on, move only the vehicle and the rider follows as one body: a chase scene is just drawing the car's path. Dismount puts them down beside the seat.
5. 5In shooting mode, Auto coverage reads your staging and lays down standard coverage in one click (two-shot master + OTS×2 + CU×2 for two people; 3 cameras for one), and duo motions (meet/pass/chase) are one click each.
6. 6Build multi-camera hard-cut sequences with the camera cutlist, and pose figures with presets, joint editing, or AI pose prompts ("kneel and plead").
7. 7Save a whole staged scene (props, cast, cameras, mood) as My Prefab for reuse, and keep versioned snapshots on disk with Save & Versions (v1, v2, …).
8. 8When ready, render a control clip or capture a still and hand it to AI video generation. Sheet images of characters placed in the scene are attached automatically as identity references ("@ImageN is {name}") to keep faces consistent.

Scene composition and pose generation use Gemini (BYOK); reference video generation uses the Seedance family; text-to-3D objects use Tripo3D. 3D staging, templates, coverage, and riding are local deterministic features and work without any keys.