arXiv Computer Vision By Ruiyang Chen, Feiran Li, Heng Guo, Zhanyu Ma

FlashNormal: Detailed Surface Normal Estimation from Flash and No-Flash Images

Read the original on arXiv Computer Vision →

FlashNormal is a diffusion-based method that estimates detailed surface normals from flash/no-flash image pairs, leveraging flash-induced shading variations and a curvature-guided detail enhancement strategy to improve surface detail recovery and reduce shape‑reflectance ambiguity. The approach is designed for practical use on modern smartphones and is evaluated on EvalFlash, a new real‑world dataset of 20 objects with ground‑truth normals. Experiments show FlashNormal outperforms existing single‑image methods and surpasses prior flash/no‑flash normal estimation techniques on EvalFlash.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

Hugging Face Trending Papers
3d ago

Casual Flash Lighting for Gaussian Splat Inverse Rendering

The paper introduces a method that combines casual indoor photographs taken with and without flash to recover geometry, materials, and lighting. By using flash residuals to constrain albedo and BRDF and static lighting to capture specular highlights, the approach employs a hash‑encoded MLP anchored to a 2DGS depth map for view‑consistent material decomposition. Experiments on synthetic and real scenes show improved diffuse color, albedo, roughness, and relighting performance, achieving a 4.17 dB PSNR gain over the best baseline.