The concept of silkscreen is to use simulation based inference combined with artpop to infer properties of galaxies directly from their images. The main initial goal is the measure the distances of galaxies using only their broadband cutouts, but the possibilities are endless! Many more details about the current implementation are contained in the paper on the arXiv
Unfortunately the code is not particularly use friendly at the current moment. The best place to start is the read through the paper, which goes into much more detail, and look at the docs/basic_workflow.py for a basic practical implementation. We plan on improving this in the future but if you have questions or issues about using the code now then please contact us either by raising an issue here or directly by email!