3D Workflow Guide
Scene Blocking and 3D Workflow Guide
Scene Blocking is for shots where spatial decisions are the bottleneck: camera side, subject placement, scale, depth, eyelines, or the final frame the clip must land on. Use it to remove ambiguity before image and video generation, not as a required step for every shot.
When 3D Helps
Use 3D when language alone is too vague. A prompt can say "the hero stands behind the table," but a blocking snapshot can show the exact camera side, crop, distance, foreground/background spacing, and relative subject scale. That makes it useful for wide shots, two-shots, group staging, entrances, reveals, over-the-shoulder setups, product placement, and shots that must cut cleanly into another shot.
Skip 3D when the shot is simple, close, and already well constrained by a strong character or location reference. In those cases, the faster path is usually a precise start-frame prompt and one clean motion beat.
Two Scene Blocking Surfaces
Creator Flow shot blocking
Use the shot-level Scene Blocking modal when the snapshot should be saved on the shot. It works from the shot's location or zone, visible cast, character variants, camera state, placement controls, saved views, and output snapshot.
Standalone 3D Scene Blocking
Use the standalone 3D tool for reusable model or location-view references. It loads ready 3D assets from the library, captures snapshots, and gives you editable binding text that can be copied or exported as JSON.
Recommended Workflow
Step 1
Prepare the 3D assets
Generate or upload the character, variant, object, ability, and location assets that actually need spatial planning. For shot-specific Creator Flow blocking, the important dependencies are a ready location or zone and ready character or variant models.
Step 2
Block the camera and cast
Open Scene Blocking from the shot when the blocked frame should be saved on that shot. Choose the location or zone, select the visible characters, place them in frame, adjust scale, save useful camera views, and capture a snapshot.
Step 3
Use the snapshot for the right frame
Use a start-frame blocking snapshot to guide the opening image composition. Use end-frame blocking when the video should land on a specific final pose, camera side, depth relationship, or screen position.
Step 4
Keep appearance references separate
Treat the 3D snapshot as composition authority. Character identity, wardrobe, texture, materials, and final lighting should still come from the normal character, location, gallery, or asset references.
Start Frames
In Creator Flow, the shot image is the opening frame anchor for video generation. Use scene blocking before generating or regenerating that image when the composition needs to be exact. The blocking snapshot should tell the image model where the camera is, where the characters sit on screen, how deep the space is, and what staging relationship matters.
- Use the blocking snapshot for composition, camera side, crop, depth, and relative scale.
- Use character and location references for identity, materials, textures, and final visual detail.
- Regenerate the start frame when the saved blocking state changes enough to make the old snapshot stale.
End Frames
End frames are destination anchors. Use the End Frame panel in Shot Production when a clip needs a specific final beat. You can upload a frame, choose from the gallery, use the next shot's start frame for continuity, add references, create an end frame from a prompt, or open end-frame Scene Blocking.
End-frame blocking is most useful when the motion should finish on a clear destination: a character lands in a doorway, a product reaches hero position, two characters end in a precise over-the-shoulder geometry, or the next cut needs a matching screen direction. The start frame remains the continuity anchor; the end frame tells the video where to land.
World Generator and Standalone Video
World Generator video mode can use start frames from upload, gallery, asset library, or recent uploads, and it can use an end frame when the selected model supports it. Use the same rule there: start frames define where motion begins, end frames define where supported motion should land, and 3D snapshots are composition references rather than appearance replacements.
Asset Readiness
- Use Image to 3D for character, variant, object, and other model-style assets.
- Use Location Image to 3D for locations or zones that need spatial camera planning.
- Use 3D model upload/export when you already have compatible model artifacts to inspect or reuse.
- If a row is missing, active, or failed, generate or repair that 3D asset before relying on the blocking snapshot.
Common Mistakes
- Using 3D proxy materials as style references. Use them for staging, not final look.
- Changing the camera after capture, then generating from an old snapshot.
- Adding too many references that disagree with the start frame or destination frame.
- Using an end frame for every clip instead of only clips that need a controlled landing point.
- Trying to solve identity drift with 3D blocking instead of stronger character references.
Use these guides with scene blocking when you need stronger prompts, world-aware generation, or full Creator Flow production.