Orbia: Do Generated Videos Form a Consistent 3D World?
We introduce Orbia, the first benchmark dedicated to a systematic and comprehensive evaluation of 3D consistency in interactive world models.
ProofPaper ↗
Key points
- Video generators serve as interactive worlds for exploration and simulation, yet realistic and visually appealing frames do not necessarily depict a consistent 3D world.
- Existing benchmarks assess selected geometric properties or use vision-language models to judge memory and consistency, leaving systematic 3D evaluation underexplored.
- For 3D self-consistency, cross-view depth and feature correspondence, regional scale agreement, and held-out Gaussian rendering test whether generated views describe the same 3D scene.
- Our Orbia benchmark provides a systematic basis for diagnosing 3D inconsistency and developing more persistent video world models.
Sources (1)
- [1]Orbia: Do Generated Videos Form a Consistent 3D World?Hugging Face Daily Papers · Oct 14, 12:00 AM
We introduce Orbia, the first benchmark dedicated to a systematic and comprehensive evaluation of 3D consistency in interactive world models.
Video generators serve as interactive worlds for exploration and simulation, yet realistic and visually appealing frames do not necessarily depict a consistent 3D world.
Extractive summary: sentences quoted from the sources.
Before this
- Oct 13, 2026Decoupled Drift-Diffusion World Models for Consistent Future Frame Prediction
- Oct 13, 2026Action Jets: Learning the Local Shape of the Action-conditioned Data Manifold
- Oct 13, 2026Gemini 4 Argon for Robotics
- Oct 13, 2026OccWAM: Occupancy World-Action Models with Joint Scene-to-Ego Flow Matching
- Oct 13, 2026Fish Audio Drama 3 Technical Report
- Oct 13, 2026WorldMemBench Evaluates Static and Dynamic Memory in Video World Models