A 1,000-pair real-world HDR benchmark with two new affine-invariant error scores shows the best image-editing models reproduce the relative structure of real light transport but degrade in dim regions, and that VLMs fail at pixel-level light checks.
In: SIGGRAPH Asia 2022 Confer- ence Papers
3 Pith papers cite this work. Polarity classification is still indexing.
3
Pith papers citing it
fields
cs.CV 3representative citing papers
PhysEditBench is a protocol-conditioned benchmark evaluating image editors on dense prediction of depth, normal, albedo, roughness, and metallic maps from RGB images using curated data and fixed scoring rules.
GaNI combines NeuS geometry reconstruction with a light-position-aware inverse neural radiosity stage that adds implicit near-field modeling, surface angle loss, and roughness smoothness priors to recover reflectance parameters from co-located light-camera captures.
citing papers explorer
-
Do Image Editing Models Understand Lighting?
A 1,000-pair real-world HDR benchmark with two new affine-invariant error scores shows the best image-editing models reproduce the relative structure of real light transport but degrade in dim regions, and that VLMs fail at pixel-level light checks.
-
PhysEditBench: A Protocol-Conditioned Benchmark for Dense Physical-Map Prediction with Image Editors
PhysEditBench is a protocol-conditioned benchmark evaluating image editors on dense prediction of depth, normal, albedo, roughness, and metallic maps from RGB images using curated data and fixed scoring rules.
-
GaNI: Global and Near Field Illumination Aware Neural Inverse Rendering
GaNI combines NeuS geometry reconstruction with a light-position-aware inverse neural radiosity stage that adds implicit near-field modeling, surface angle loss, and roughness smoothness priors to recover reflectance parameters from co-located light-camera captures.