← all videos

A pianist sits on the left side of a wide stage and plays a deep chord. Then a violinist on the right side answers with a high singing note.

C1-2-09 Steady State · source_anchoring principle: Multi-Source Spatial Separation (Wave)
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 pianist
  • a violinist
  • a stage

Event: a pianist and a violinist perform on a stage

Audio presence

  • a piano
  • a violin

Sound: ['piano chord', 'violin note']

Key standards

Visual physical commonsense (V-PC)

  • A visible pianist holding a piano stays at a stable position on the left side of the stage.
  • A visible violinist holding a violin stays at a stable position on the right side of the stage.

Audio physical commonsense (A-PC)

  • The piano chord carries the deep low-register resonant chord timbre of a real piano.
  • The violin note carries the high sustained bowed-string singing-note timbre of a real violin.
  • The piano chord sounds first as a discrete event, and the violin note answers as a second discrete event.

Cross-modal physical commonsense (AV-PC)

  • The piano chord is localized to the left side of the audio field, matching the pianist's visible left-side position.
  • The violin note is localized to the right side of the audio field, matching the violinist's visible right-side position.