PVBench scores physical plausibility in generated video
The benchmark uses 600 prompts and 46 dimensions, with object-level measurements instead of only visual-language judgments.
2 stories, newest first · one card per story
The benchmark uses 600 prompts and 46 dimensions, with object-level measurements instead of only visual-language judgments.
Expanding-domain neural deflation recovers the expected local invariants, then their successive breakdown.
Today 1.6K readers · 2.5K views · 779 reads · all time 189K views