Principia benchmark: top video models score below 0.42 on Newtonian physics despite ~0.8 on VBench
anand_bhattad · x · 2026-09-06
Researchers from IISc and Johns Hopkins released Principia, a benchmark that tests video generators' Newtonian physics consistency using calibration-independent relational constraints between paired objects.
- Covers 8 phenomena: gravity, restitution, friction, rotational inertia, projectile motion, momentum, pendulum, and mass–spring oscillation, using real scenes recorded under controlled protocols.
- Across thousands of generations from 5-6 SOTA video generators and 4 VLMs: no video model exceeds 0.42 physical consistency on Principia, despite all scoring 0.8 on VBench—visual realism doesn't imply physical understanding.
- Best VLM detects physics violations at only 67% accuracy; most perform near chance.
Takeaway: video models are still far from physically consistent generation.
Related event: Principia Benchmark Exposes Video Models' Poor Physics Understanding(4 posts)→
More from Multimodal
- Kling 4.0 tip: reuse a clear frame as reference to move characters across scenes — testingcatalog · 2026-10-23
- Kling AI officially launches Kling 4.0: 15 references per task, 30s clips, 10 keyframes — testingcatalog · 2026-10-23
- Kling 4.0 hands-on: multi-shot cyberpunk robot demo shows strong character consistency — testingcatalog · 2026-10-23
- Run Lola Run director Tom Tykwer uses Seedance in hybrid AI filmmaking experiment — Uncanny_Harry · 2026-10-23
- "Still Worth Something" S01E01: a Pixar-style fully AI-animated comedy short — LPGroves · 2026-10-23
- Music publisher Reservoir sues Suno and Anthropic over AI music training — TinfoilTricorn · 2026-10-23