MERF: Memory-Efficient Radiance Fields for Real-time View Synthesis in Unbounded Scenes
Authors:
- Christian ReiserCRChristian Reiser
- Richard SzeliskiRSRichard Szeliski
- Dor VerbinDVDor Verbin
- Pratul P. SrinivasanPSPratul P. Srinivasan
- Ben MildenhallBMBen Mildenhall
- Andreas GeigerAGAndreas Geiger
- Jonathan T. BarronJBJonathan T. Barron
- Peter HedmanPHPeter Hedman
Preprint
Abstract
Neural radiance fields enable state-of-the-art photorealistic view synthesis. However, existing radiance field representations are either too compute-intensive for real-time rendering or require too much memory to scale to large scenes. We present a Memory-Efficient Radiance Field (MERF) representation that achieves real-time rendering of large-scale scenes in a browser. MERF reduces the memory consumption of prior sparse volumetric radiance fields using a combination of a sparse feature grid and high-resolution 2D feature planes. To support large-scale unbounded scenes, we introduce a novel contraction function that maps scene coordinates into a bounded volume while still allowing for efficient ray-box intersection. We design a lossless procedure for baking the parameterization used during training into a model that achieves real-time rendering while still preserving the photorealistic view synthesis quality of a volumetric radiance field.Keywords
All Related Versions
This publication has 35 references indexed in Scilit:
- Digging into Radiance Grid for Real-Time View Synthesis with Detail Preservation2022
- AdaNeRF: Adaptive Sampling for Real-Time Rendering of Neural Radiance Fields2022
- R2L: Distilling Neural Radiance Field to Neural Light Field for Efficient Novel View Synthesis2022
- TermiNeRF: Ray Termination Prediction for Efficient Neural Rendering2021
- Convolutional Occupancy Networks2020
- MatryODShka: Real-time 6DoF Video View Synthesis Using Multi-sphere Images2020
- DeepView: View Synthesis With Learned Gradient Descent2019
- The Unreasonable Effectiveness of Deep Features as a Perceptual Metric2018
- Efficient View-Dependent Image-Based Rendering with Projective Texture-Mapping1998
- Optical models for direct volume renderingIEEE Transactions on Visualization and Computer Graphics, 1995