Training data for world models — gameplay video paired with the ground truth behind it: 4D geometry, tick-level world state, and human-written visual reasoning.
The four product lines below are live, browsable samples packaged in the format we ship in production. Together they cover the loop a world model has to learn: watch the world move (4D game capture), read the state that produced every frame (engine-native logs), reason about what it sees (chain-of-thought visual and STEM data), and build the objects its worlds are made of (parametric 3D assets). Nothing is estimated by an off-the-shelf perception model — depth, pose, state and events are read straight from the engine, and reasoning chains are written by people.
tidelaccess
Gameplay video from a proprietary capture pipeline over AAA titles, paired with the ground truth a world model actually needs — measured from the engine, not estimated by a perception model.
Three browsable categories — A · Action + Camera, B · 4D + Depth, C · Multi-Camera — every frame with real telemetry on one timeline.
Title / genre mixes, camera rigs, action distributions and episode lengths captured to order on the same pipeline — engine as generator and evaluator for your model's predictions.
Every video frame ships with the exact world state the simulation computed at that instant — entities, components, events. Nothing is detected; everything is read. One 60 Hz clock across all modalities.
Episode ep001 in a live inspector: video with SEG / DEPTH / BBOX overlays, pixel→state decoding, and a per-tick world-state bar.
The same engine data replayed as causal chains — shooter, target, health drop, impact, elimination — assembled step by step over the video.
The seeing skills frontier models still fail — on the BabyVision benchmark (arXiv:2601.06521) adults score 94.1% while the best model scores 49.7% — and the reasoning-first data that trains them.
Counting, find-the-difference, mazes, shadow matching, pattern logic — every sample with a human-written, step-checkable chain of thought.
High-school Math / Physics / Biology diagram problems, LaTeX-typeset, with verifiable answers and worked solutions.
Simulation-ready objects generated by build scripts: change a dimension, rerun, and the mesh, UVs, PBR textures and renders all regenerate consistently. Real-world scale, named movable parts, and the editable Blender source with every piece.
Forty objects across furniture, porcelain, lighting, appliances, props, vehicles and full rooms: rotate the live GLB in the browser, play the turntable, operate the movable parts.
Categories, parameter ranges, variant counts and material sets built to order on the same pipeline — the same six elements on every object.