A person speaks at normal volume inside a car. The windows are rolled down on a highway, and even though they keep speaking, the wind noise completely covers their voice.
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 car interior
- car windows
Event: a person speaks inside a car with the windows rolled down on a highway
Audio presence
- a person
- wind
Sound: ['human speech', 'wind noise']
Key standards
Visual physical commonsense (V-PC)
- The car interior is visibly plausible with the windows rolled down, and the person is visibly speaking at normal volume.
Audio physical commonsense (A-PC)
- The wind noise is substantially louder than the voice and completely covers the voice.
Cross-modal physical commonsense (AV-PC)
- During the visible speaking while the windows are down, no voice is intelligibly audible above the wind noise.