Assets & reconstruction
Physics-enabled 3D assets, PBR data, video-to-3D reconstruction and parametric CAD across diverse environments.
World Compiler / 05-stage pipeline
We turn playable worlds into aligned assets, cameras, behavior, controls, engine signals and task semantics.
Assets, behavior and game worlds
Game and 3D data bring together high-density environments, behavior capture and production-ready assets for teams building scalable visual intelligence.
Discuss a data programPhysics-enabled 3D assets, PBR data, video-to-3D reconstruction and parametric CAD across diverse environments.
Gameplay capture with multiple calibrated views, action trajectories, controls and synchronized visual context.
G-buffer and engine-level outputs for visual learning, reconstruction and world-model work.
3D model asset library
A broad model library supports world building, simulation, visual pretraining and game-data production with consistent, reusable source assets.





Visual reconstruction data
Reference footage is reconstructed into structured Blender scenes with matched composition, camera motion, character blocking and action trajectories—not just another rendered clip.
A cinematic reference is rebuilt as a clean white-model environment, preserving spatial layout, camera language, character positions and movement paths.
Scene composition, action timing, camera movement and key visual effects are reproduced as editable 3D data aligned to the original video.
Parametric CAD reconstruction
Dimensioned drawings and reference geometry are reverse-engineered into manufacturing-grade STEP models and editable build123d code, with geometry validation built into delivery.
Synchronized multiview game data
A single interactive task can be delivered with a first-person action camera, five calibrated fixed views, depth, trajectory and frame-level control signals.







Representative samples
Collection and production
Our library combines cleaned 3D assets with game-world visual data and engine-level outputs. It is designed for teams that need scale, diversity and structure rather than a collection of isolated clips.
Need a tailored dataset, schema or sample package?
contact.vanso@gmail.com