To get started, the requirements file contains everything that must be installed after cloning the repo. Simply run the following command to create it, we recommend installing within a conda environment you create:
Then, create a conda environment and install the requirements within it.
conda env create myEnviroName
conda activate myEnviroName
pip install -r requirements.txt
The program main functionalities are in the jupyter .ipynb files:
- processing_training_data - Use this file to process training data you generated from blender or omniverse code, so that it can be used in Nerf-reconstruction
- Nerf-reconstruction - Program to train a NeRF given rays (can be used for unzipped data you already have)
- Testing - Use this to test PSNR in rendering novel views
- MeshExtraction - Use this to generate a 3D mesh from your trained model
- visualize - Use this to visualize pose and ray data for one or many poses
To run with an existing dataset, simply unzip it in the datasets folder.
