aiqbench analysis
This research introduces EgoGenEval, a new benchmark addressing the critical gap in visual generators' ability to maintain physical consistency during ego-motion, which is essential for embodied AI and spatial reasoning. The study reveals that current models struggle to simultaneously execute camera motion and preserve scene state, suggesting that standard pairwise supervision methods are insufficient for solving this coupled challenge. Consequently, the authors propose a trajectory-centric paradigm that integrates explicit pose and visibility supervision to overcome these limitations.
Beyond Visual Quality: Evaluating Physical Consistency under Ego-Motion with EgoGenEval
arXiv:2609.11172v1 Announce Type: cross Abstract: Recent visual generators produce high-fidelity images yet often violate physical consistency under ego-motion, limiting their use for spatial reasoning and embodied planning. Existing benchmarks largely focus on isolated images or single-step quality, leaving this challenge underexplored. We introduce EgoGenEval, a geometry-grounded, pose-free benchmark designed to evaluate the physical consistency of visual generators under ego-motion, and organize our study into two parts. (1) EgoGenEval contains 1,400 cases and 2,360 target views spanning single-step and multi-step ego-motion. It separately measures Camera Motion Grounding (CMG) and Scene State Preservation (SSP), with both metrics validated against blinded human judgments. Evaluating 16 pose-free generators together with two pose-conditioned references reveals that current models struggle to execute camera motion while maintaining scene state, and that no system performs well on both axes at once. (2) To examine whether benchmark-derived data can improve these capabilities, we build EgoGen-Train from the same geometry-grounded pipeline and run controlled SFT studies. These show that pairwise supervision does not reliably improve camera-motion grounding and scene-state preservation together: even at the full training pool and the longest budget, scene preservation gains a fraction of what camera motion does. This points to the pairwise teacher-forced objective itself as the binding constraint, motivating a trajectory-centric paradigm that couples self-conditioned rollouts with explicit pose and visibility supervision.
Abstract:Recent visual generators produce high-fidelity images yet often violate physical consistency under ego-motion, limiting their use for spatial reasoning and embodied planning. Existing benchmarks largely focus on isolated images or single-step quality, leaving this challenge underexplored. We introduce EgoGenEval, a geometry-grounded, pose-free benchmark designed to evaluate the physical consistency of visual generators under ego-motion, and organize our study into two parts. (1) EgoGenEval contains 1,400 cases and 2,360 target views spanning single-step and multi-step ego-motion. It separately measures Camera Motion Grounding (CMG) and Scene State Preservation (SSP), with both metrics validated against blinded human judgments. Evaluating 16 pose-free generators together with two pose-conditioned references reveals that current models struggle to execute camera motion while maintaining scene state, and that no system performs well on both axes at once. (2) To examine whether benchmark-derived data can improve these capabilities, we build EgoGen-Train from the same geometry-grounded pipeline and run controlled SFT studies. These show that pairwise supervision does not reliably improve camera-motion grounding and scene-state preservation together: even at the full training pool and the longest budget, scene preservation gains a fraction of what camera motion does. This points to the pairwise teacher-forced objective itself as the binding constraint, motivating a trajectory-centric paradigm that couples self-conditioned rollouts with explicit pose and visibility supervision.
39 pages, 10 figures, and 18 tables. Code: this https URL
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2609.11172 [cs.CV]
(or arXiv:2609.11172v1 [cs.CV] for this version)
https://doi.org/10.48550/arXiv.2609.11172 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 323 words
Related coverage
[AINews] Opus 5.5 is good at explainer videos
Latent Space · Sep 29, 2026
Runway’s WorldPrompt and the Engineering of Real-Time Worlds
Latent Space · Sep 25, 2026
TriWorldBench: A Tri-View Consistency Perspective on Embodied World Models
arXiv · Sep 24, 2026
Dyna-2: General Purpose Physical Agents with Hour-Long Autonomy — Jason Ma, Dyna Robotics
Latent Space · Oct 13, 2026