This repository contains the code needed to visualize trained Mega-NeRF models.
The codebase for training Mega-NeRF models and extract sparse voxel octrees can be found here.
Note: This is a preliminary release and there may still be outstanding bugs.
mkdir build && cd build
cmake ..
make -j12
You will need a recent version of cmake, LibTorch, and glfw3. The full list of dependencies can be found in the CMakeLists file.
The codebase has been mainly tested against CUDA >= 11.1 and 32GB V100 GPUs.
./mega-nerf-viewer $OCTREE_PATH --model_path $MODEL_PATH
The M key toggles dynamic octree refinement. The R key toggles guided ray sampling.
Large parts of this codebase are based on the Plenoctree renderer.