A person speaks outdoors during a thunderstorm. A loud clap of thunder arrives, and even though they keep speaking, their voice is completely covered for several seconds.
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 person
- a thunderstorm sky
Event: a person speaks outdoors during a thunderstorm
Audio presence
- a person
- thunder
Sound: ['human speech', 'thunder clap']
Key standards
Visual physical commonsense (V-PC)
- The thunderstorm is shown with rain and a dark sky or visible lightning, and the person keeps visibly speaking throughout the thunder.
Audio physical commonsense (A-PC)
- A thunderclap is substantially louder than the voice and completely covers the voice for several seconds while the person is visibly speaking.
Cross-modal physical commonsense (AV-PC)
- During the visible speaking that coincides with the thunderclap, no voice is intelligibly audible above the thunder.