SIM
PHOTOREAL
Gravity invertedtraj corr 0.9999
Gravity points up: g = +9.81 m/s². The released ball falls to the ceiling.
InfiniteWorld — Parallel-Physics Worlds
A VideoQA benchmark that tests whether video models represent physical laws as editable structure rather than an entangled realism prior. The videos come from worlds whose laws have been systematically altered — gravity inverted, friction removed, restitution super-elastic — each rendered with exact per-frame ground truth from the physics engine that produced it.
aMethod
Every clip is built in three layers. Engine truth is the source of record; neural translation makes it photorealistic; a tracking-based certifier confirms the altered law survived the translation pixel-to-trajectory.
LAYER 1
Engine truth
PyBullet simulates each world at 240 Hz with a dialed-in law. Exact object trajectory, camera matrices, and depth are stored per frame.
R² ≥ 0.999999 fit
→
LAYER 2
Neural sim→real translation
Cosmos-Transfer1-7B (depth-conditioned) re-renders the 832×480 sim into a photorealistic 1280×704 clip, keeping motion locked to the engine.
~953 s / clip on A100
→
LAYER 3
SAM2 coherence certification
SAM2 tracks the object in the translated video; its path is correlated against the projected engine trajectory. Only clips whose physics survived pass.
gate: tracked ≥ 0.80 · corr > 0.95
bCertified sim ↔ real pairs
Five worlds, one per certified law family. Left is the engine sim; right is the photorealistic Cosmos translation. Each pair passed coherence certification — the object's tracked pixel path in the photorealistic clip matches the projected engine trajectory, so the altered law survived rendering.
SIM
PHOTOREAL
Restitution supertraj corr 0.9990
Super-elastic bounces: e = 1.30 (clamp 8 m/s). Each apex climbs higher than the last.
SIM
PHOTOREAL
Friction scaledtraj corr 1.0000
Friction removed: μ = 0 (Earth ref μ = 0.35). The pushed box slides without stopping.
SIM
PHOTOREAL
Dissipation airtraj corr 1.0000
Thick air drag: k = 0.4 /s velocity decay. The flight arc dies mid-air.
SIM
PHOTOREAL
Gravity scaledtraj corr 0.9993
Gravity at 10%: g = −0.981 m/s². Slow, floaty drop with low bounces.
Physics-preserved = SAM2-tracked pixel trajectory in the photorealistic clip correlates with the projected engine ground truth on the dominant motion axis. Gravity-inverted and restitution-super pairs are from the certified pilot (22/24 survived); friction, dissipation, and gravity-scaled are live-run survivors from the corpus certification.
dBy the numbers
3000
simulated clips
(1200 anomaly · 1200 control · 600 combo)
(1200 anomaly · 1200 control · 600 combo)
≥0.999999
min engine-fit R² across all 3000
15,910
QA items · 5 task types
PASS
blind-LLM gate
(gpt-5-mini + deepseek-v3)
(gpt-5-mini + deepseek-v3)
eThe five QA tasks
Every question is designed to be unanswerable without watching the video — confirmed by the blind-LLM audit below. Real examples from the training split; answers are hidden.
law_identification · 3,600 items
In this video, which one of the following aspects of physics behaves differently from normal Earth physics?
- The direction of gravity
- Bounciness (restitution) of collisions
- None — the physics matches normal Earth physics
- The strength of gravity
Show answer
The direction of gravity
parameter_regression · 3,600 items
Estimate the effective vertical gravitational acceleration acting on the object in this video, in m/s². Use a signed value: negative means directed downward, positive means upward.
free-response numeric · unit m/s² · graded within ± toleranceShow answer
−0.981 m/s² (full credit ±0.6, half credit ±1.2)
forward_prediction · 1,540 items
Consider the position of the moving object at t = 1.38 s in the video. At t = 2.50 s, where is the object relative to that position?
- Noticeably lower
- At about the same height
- It is out of view (off-screen)
- Noticeably higher
Show answer
Noticeably lower
cross_world_counterfactual · 1,170 items
If this exact scene were re-run with normal Earth physics (same object, same initial push), would the object slide a shorter total distance, a longer total distance, or about the same distance as it does in the video?
- A longer distance than in the video
- About the same distance
- A shorter distance than in the video
Show answer
About the same distance
law_induction · 6,000 items
The first two videos come from the same world, whose physics differs from normal Earth physics in a specific, consistent way. Does the third video come from a world with the same physics — the same law(s) altered in the same direction and by the same amount?
- No
- Yes
Show answer
Yes
fBlind-LLM gate
Two strong text-only models answer question text + options with no video, no IDs, at temperature 0 (audit of 2026-07-23, 300 stratified items). A task passes only if both stay below chance + 8 points — i.e. the video is genuinely required. All four task types pass.
| Task type | chance | gate | gpt-5-mini | deepseek-v3 | verdict |
|---|---|---|---|---|---|
| law_identification | 25.0 | 33.0 | 18.3 | 25.0 | PASS |
| forward_prediction | 25.0 | 33.0 | 25.0 | 18.8 | PASS |
| cross_world_counterfactual | 33.3 | 41.3 | 34.0 | 37.0 | PASS |
| law_induction | 50.0 | 58.0 | 51.7 | 51.7 | PASS |
Accuracies in %. parameter_regression is free-response numeric and is not part of the MCQ blind gate.
Every released row carries blind_answerable = false.
gLaw families & dials
Six law families, each dialed across five magnitudes plus a matched Earth-physics control. C-tier = neurally translated and coherence-certified for the sim↔real pairs above.
| Family | Altered law | dial | C-tier certification |
|---|---|---|---|
| gravity_inverted | Direction of gravity flips upward | g = +9.81 m/s² | pilot 11/12 |
| gravity_scaled | Strength of gravity scaled down | g_scale = 0.1–1.0 | run in progress · 11/12 of certified so far |
| friction_scaled | Surface friction scaled to zero | μ = 0 (ref 0.35) | run in progress · 12/12 of certified so far |
| restitution_super | Super-elastic collisions | e = 1.3 (clamp 8 m/s) | pilot 11/12 |
| dissipation_air | Strong in-flight air drag | k = 0.4 /s | run in progress · 10/12 of certified so far |
| buoyancy_inverted | Inverted buoyancy in liquid | F = 1.8·m·g up | C-tier coming in v1.1 — water-occlusion re-render |
Corpus generation is complete for all 3000 clips (six families × five magnitudes + controls, all fit at R² ≥ 0.999999). C-tier certification of the full corpus is a live run: 91 / 1200 translated clips certified so far, 62 passing. Per-family "of certified so far" counts are from that in-progress run.