A cymbal is crashed loudly with a long shimmering ring. Then a hand grabs the cymbal edge and the sound stops instantly.
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
- a cymbal
- a hand
Event: a hand grabs a crashed cymbal
Audio presence
- a cymbal
Sound: a cymbal crash
Key standards
Visual physical commonsense (V-PC)
- A cymbal is visibly crashed once with a stick during the clip.
- A hand visibly grabs the cymbal edge while the cymbal is still ringing.
Audio physical commonsense (A-PC)
- The cymbal produces a long shimmering ring after the crash.
- The ring decays to silence almost instantly once the hand grabs the cymbal edge, with no residual ringing.
Cross-modal physical commonsense (AV-PC)
- The crash transient in the audio aligns in time with the visible stick strike.
- The abrupt end of the ring coincides in time with the visible moment the hand contacts the cymbal edge.