Sand Atlas

Documentation here, or compile it yourself following the details below.

Installation

First, install ffmpeg and blender on your system. Both of these need to be available in your system path (so that running ffmpeg and blender in the command line works).

This package can then be installed via pip install sand-atlas. If installing from github, try cloning and then running:

pip install -e .

If you make any changes to the source code, re-run those two lines to have your changes reflected in your installed package.

Usage

If you are interested in reproducing a dataset that is available in the Sand Atlas, you can use the installed script sand_atlas_process, like so:

sand_atlas_process <path_to_json_file> --label <path_to_labelled_image> --raw <path_to_raw_image>

If you supply only the raw image, the script will attempt to label the image for you. If you supply the labelled image, the script will use that to generate the dataset. The json file contains the metadata for the dataset you are interested in. The json files for the datasets in the Sand Atlas are available here.

Documentation

We use sphinx to manage the docs. Update documentation with:

cd docs
make html

Once these are built, you can commit and push the changes to github to have them refreshed on github pages. You can also view them locally.