instant_ngp

refs:

https://github.com/MaximeVandegar/Papers-in-100-Lines-of-Code/tree/main/Instant_Neural_Graphics_Primitives_with_a_Multiresolution_Hash_Encoding

https://pytorch.org/tutorials/advanced/cpp_custom_ops.html

https://pytorch.org/tutorials/advanced/cpp_extension.html

https://github.com/pytorch/extension-cpp/

https://pytorch.org/docs/stable/notes/extending.html

autograd

https://pytorch.org/tutorials/intermediate/custom_function_double_backward_tutorial.html

https://pytorch.org/docs/stable/notes/extending.html

ds: https://drive.google.com/drive/folders/1eO7DXFhWWpauC-9LDhOimtIKxY3yRCIm
in the nerf paper they used spherical harmonics for color instead

TODO

• dataset class
• tcnn hashing
• tcnn forward
• occupancy grid updates
• cuda ray marching
• test kernel
• full test pipeline
• implement a viewer for training

setup

conda install -c conda-forge cuda=12.1 gxx python=3.11.8
pip3 install torch torchvision torchaudio #torch 2.4.1 cu12.1
pip install git+https://github.com/NVlabs/tiny-cuda-nn/#subdirectory=bindings/torch
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.4.1+cu121.html
pip install einops kornia matplotlib opencv-python lpips imageio imageio-ffmpeg scipy pymcubes trimesh dearpygui lightning