Merge pull request #5 from mgrankin/conda

Add conda environment spec

README.md

@@ -61,13 +61,23 @@ Optimizing a NeRF takes between a few hours and a day or two (depending on resol
 
 ## How to train your NeRF super-quickly!
 
-To train a "full" NeRF model (i.e., using 3D coordinates as well as ray directions, and the hierarchical sampling procedure), first setup dependencies. In a new `conda` or `virtualenv` environment, run
+To train a "full" NeRF model (i.e., using 3D coordinates as well as ray directions, and the hierarchical sampling procedure), first setup dependencies. 
+
+### Using pip
+In a new `conda` or `virtualenv` environment, run 
+using pip
 ```bash
 pip install -r requirements.txt
 ```
-
 **Importantly**, install [torchsearchsorted](https://github.com/aliutkus/torchsearchsorted) by following instructions from their `README`.
 
+### Alternatively, using conda
+```bash
+cd nerf-pytorch 
+conda env create
+conda activate nerf
+```
+
 Once everything is setup, to run experiments, first edit `config/lego.yml` to specify your own parameters.
 
 The training script can be invoked by running

environment.yml

@@ -0,0 +1,21 @@
+name: nerf
+channels:
+  - pytorch
+  - conda-forge
+  - defaults
+dependencies:
+  - python>=3.7
+  - pip
+  - pytorch>=1.3
+  - cuda100
+  - cudatoolkit>=10.1
+  - cudatoolkit-dev
+  - imageio
+  - numpy
+  - pyyaml
+  - tensorboard
+  - torchvision
+  - tqdm
+  - pip:
+    - opencv-python-headless
+    - "git+https://github.com/aliutkus/torchsearchsorted#egg=torchsearchsorted"