lxmpicturelab

Compare and experiment with different picture formation algorithms using various image sources.

The results are published to https://mrlixm.github.io/picture-lab-lxm.

⚠️ This repository is currently partially public, the source image assets are not version-controlled yet.

what

These are the tasks this repository defines:

1. stores photographic and cgi image source assets retrieved from external web sources.
   • generate consistent metadata for them
   • uniformize them (size, file format, encoding, ...)
2. combine all those assets to a single "set" image (a mosaic).
3. build a picture formation algorithm to be usable as a "renderer"
4. render the images generated in 2. with one or multiple renderers of 3. to generate a set of "comparison" images.
5. run and publish the results to a static website

All those tasks match an executable python script:

1. asset-generate.py
2. sets-generate.py
3. renderer-build.py
4. comparisons-generate.py
5. site-build.py

scripts usage

pre-requisites

• uv for python project management
• git for doc publish / contributing to this repo

preparation:

• download this repository anywhere on your system
• cd path/to/repo
• uv sync : create the python virtual environment

To run a script you can use the following template:

uv run script/path/scriptname.py

asset-generate.py

Ingest a pre-asset as "asset" by optimizing and conforming it.

Use uv run asset-generate.py --help to display its documentation.

⚠️ This script requires you for now to manually download each image source from its url listed the json metadata file.

You can then check the README in assets/ to better understand the asset workflow.

ingest-assets-int.py

Batch run asset-generate.py on all the known assets of this repository.

sets-generate.py

This scripts requires the assets in assets/ to exists.

The current configuration will combine all the existing assets found in the directory.

You can create additional variants by editing the global variable SET_VARIANTS and adding it a new SetVariant instance. You can restrict which assets are combine by setting the SetVariant.asset_filter field.

The output of this script is found at sets/ (currently not version-controlled).

renderer-build.py

Use uv run renderer-build.py --help to display its documentation.

Will build the necessary resources to use a picture formation algorithm specified from online sources (requires internet connection).

comparisons-generate.py

Use uv run comparisons-generate.py --help to display its documentation.

This script depends on:

• the image file generated by previous scripts which are provided by giving their asset identifier (manual execution needed).
• the renderers generated by renderer-build.py (automatically executed).

By default, the comparison results and the renderers are stored in an .workbench/ directory at root. You can change those with the command line interface.

site-build.py

Creates a static html website with a specific set of assets, that are run through comparisons-generate.py.

The html is manually authored, with the help of Jinja2 templating.

The comparisons results will be stored in a different directory but only need to be generated once. You can then quickly rebuild the html website without waiting for the long comparisong process.