MattPort

Setup the environment

# Clone the repo
git clone --recurse-submodules git@github.com:ethanweber/mattport.git

# Create the python environment
conda create --name mattport python=3.8.13
conda activate mattport
pip install -r environment/requirements.txt

# Install helper library
cd external/goat
python setup.py develop
# then go back, i.e., `cd /path/to/mattport`

# Install mattport as a library
python setup.py develop

# Running the test cases
pytest tests

Getting the data

Download the original NeRF dataset and put it in the following format.

- data/
    - blender/
        - fern/
        - lego/

Training a model

# Run with the default config
python scripts/run_train.py
# Run with config changes
python scripts/run_train.py machine_config.num_gpus=1

Setting up Jupyter

python -m jupyter lab build
bash environments/run_jupyter.sh