A pytorch implementation of the HashGrid Encoder from instant-ngp, as described in Instant Neural Graphics Primitives with a Multiresolution Hash Encoding.
Note: This repo only tries to implement the hash grid encoder for now, and is far from instant (especially for NeRF experiments).
- HashGrid Encoder
- basic pytorch CUDA extension
- fp16 support
- Experiments
- SDF
- better SDF calculation (especially for non-watertight meshes)
- NeRF (cannot converge in 5s of course, but 50s is enough...)
- better ray marching strategy.
- SDF
We use the same data format as instant-ngp, e.g., armadillo and fox.
Please download the data from instant-ngp and put them under ./data
.
# SDF experiment
bash scripts/run_sdf.sh
# NeRF experiment
bash scripts/run_nerf.sh