A door opens just a few centimeters while someone speaks inside a room; the sound suddenly becomes fully clear and loud, as if the door were completely open.
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 door
- a room
Event: a closed door opens by a few centimeters while someone speaks inside the room
Audio presence
- a person
Sound: a person speaking
Key standards
Visual physical commonsense (V-PC)
- A door is visible in the fully closed position at the start of the clip while a person speaks from inside the room behind it.
- The door visibly opens by only a few centimeters during the clip and remains in this slightly cracked position rather than opening further.
Audio physical commonsense (A-PC)
- The voice from inside the room is muffled and quiet while the door is fully closed.
- The voice becomes fully clear and loud once the door cracks open by a few centimeters, with brightness and loudness matching a fully opened door rather than a small crack.
Cross-modal physical commonsense (AV-PC)
- The transition from muffled quiet voice to fully clear loud voice aligns in time with the moment the door visibly opens by a few centimeters.