A row of identical glass bottles with clearly different water levels is struck one by one, from nearly empty to nearly full; instead of fuller bottles producing lower pitches, the tones become progressively higher as the water level increases.
Seedance 2.0ByteDance · proprietary
Kling 3.0 OmniKuaishou · proprietary
Veo 3.1Google · proprietary
LTX-2.3Lightricks · open
Ovi 1.1character.ai · open
JavisDiT++JavisDiT · open
MagiHumanGAIR-NLP · open
Per-prompt rubric
Basic standards
Visual presence
- glass bottles
- water
Event: a row of glass bottles with different water levels is struck one by one from nearly empty to nearly full
Audio presence
- glass bottles
- water
Sound: bottle tones
Key standards
Visual physical commonsense (V-PC)
- A row of identical glass bottles with clearly different water levels is visible, and each bottle is visibly struck one by one in continuous physically plausible motion.
- The bottles are struck in sequence from nearly empty to nearly full, with the increasing water level visible across the row.
Audio physical commonsense (A-PC)
- Each strike produces a clear sustained tone with the bright glass-on-water timbre of a real struck filled bottle.
- The tones become progressively higher in pitch from the first strike on the nearly empty bottle to the last strike on the nearly full bottle.
Cross-modal physical commonsense (AV-PC)
- Each bottle tone onset aligns in time with a visible strike of the corresponding bottle.
- The higher-pitched tones correspond to the bottles with visibly higher water levels, and the lower-pitched tones correspond to the bottles with lower water levels.