PhysEditBench - Roughness: leaderboard

Roughness maps from image editors prompted to turn indoor photographs into physical maps (PhysEditBench, 2026) with a direct image-to-map prompt on OpenRooms-FF: root mean squared error on the 0-1 scale. Lower is better.

Metric: Roughness RMSE on the 0-1 scale against the reference roughness map, direct RGB-to-map prompt, OpenRooms-FF (1,125 images); prompted image editors on the paper main-axis split (indoor OpenRooms-FF and InteriorVerse scenes; official access settings of Appendix Table 19); task-specific specialist models are reference rows, not published; lower is better. Source: arxiv.org. Saturation forecast: Rough model projection: around 2026. 4 models tracked. Updated 7 Oct 2026.

Top models

#ModelScoreOverall rank
1GPT Image 20.323
2doubao-seedream-5-0-2601280.364
3GPT Image 1.50.375
4Qwen-Image-2.00.384

Interactive version: theaggregate.ai/benchmark?slug=physeditbench-roughness · How It Works · Data refreshed daily, snapshot 2026-11-21.