Accurate evaluation of spatial simulation for camera‑controllable generation
Develop rigorous and unambiguous evaluation methodologies and benchmarks to accurately assess spatial simulation in camera‑controllable image generation, addressing the ambiguity of calibration‑based evaluators when generated images exhibit subtle geometric differences so that geometric consistency can be measured precisely.
References
Accurately evaluating spatial simulation thus remains an open challenge and is crucial for advancing camera-controllable generation.
Simulation and UR5e results support the control utility of this representation within the evaluated adaptation settings, while independent verification of hidden-scene geometry remains open.
EgoGenEval covers static indoor scenes, four camera-action families, and rollouts of up to three steps; dynamic scenes, wider motions, and longer horizons remain open.