Abstract:NeRF synthesizes novel views of a scene with unprecedented quality by fitting a neural radiance field to RGB images. However, NeRF requires querying a deep Multi-Layer Perceptron (MLP) millions of times, leading to slow rendering times, even on modern GPUs. In this paper, we demonstrate that real-time rendering is possible by utilizing thousands of tiny MLPs instead of one single large MLP. In our setting, each individual MLP only needs to represent parts of the scene, thus smaller and faster-to-evaluate MLPs can be used. By combining this divide-and-conquer strategy with further optimizations, rendering is accelerated by three orders of magnitude compared to the original NeRF model without incurring high storage costs. Further, using teacher-student distillation for training, we show that this speed-up can be achieved without sacrificing visual quality.
| Comments: | ICCV 2021. Code, pretrained models and an interactive viewer are available at this https URL |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV) |
| Cite as: | arXiv:2103.13744 [cs.CV] |
| (or arXiv:2103.13744v2 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2103.13744 arXiv-issued DOI via DataCite |
Submission history
From: Christian Reiser [view email]
[v1]
Thu, 25 Mar 2021 10:53:05 UTC (9,201 KB)
[v2]
Mon, 2 Aug 2021 15:58:25 UTC (11,360 KB)