SuperNormal: Neural Surface Reconstruction via Multi-View Normal Integration
Xu Cao, Takafumi Taketomi
Abstract
We present SuperNormal, a fast, high-fidelity approach to multi-view 3D reconstruction using surface normal maps. With a few minutes, SuperNormal produces detailed surfaces on par with 3D scanners. We harness volume ren-dering to optimize a neural signed distance function (SDF) powered by multi-resolution hash encoding. To accelerate training, we propose directional finite difference and patch-based ray marching to approximate the SDF gradients nu-merically. While not compromising reconstruction quality, this strategy is nearly twice as efficient as analytical gra-dients and about three times faster than axis-aligned finite difference. Experiments on the benchmark dataset demon-strate the superiority of SuperNormal in efficiency and ac-curacy compared to existing multi-view photometric stereo methods. On our captured objects, SuperNormal produces more fine-grained geometry than recent neural 3D reconstruction methods. Our code is available at https://github.com/CyberAgentAILab/SuperNormal.