WorldAlign: Decoupled 4D Reward for World-Consistent Video Generation

Key points

• Geometry-aware post-training offers a promising way to improve world consistency.1
• To address these limitations, we introduce WorldAlign, a decoupled 4D reward framework that semantically separates static regions and dynamic subjects and provides feedback by aligning each with a world prior suited to its assumptions.1
• For static regions, WorldAlign aligns static geometry with a geometric world prior through semantically guided masked reprojection, enabling more reliable static-consistency evaluation; an auxiliary camera-motion reward discourages nearly static solutions.1
• These results support decoupled world-prior alignment for more faithful visual world simulation.1

Sources (1)

• [1]WorldAlign: Decoupled 4D Reward for World-Consistent Video Generation
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 05:35 PM

  Faithful visual world simulation requires generated videos to maintain 4D world consistency, encompassing both static and dynamic consistency.

  Geometry-aware post-training offers a promising way to improve world consistency.

Extractive summary: sentences quoted from the sources.