Your account of walking through the SF MOMA and being mesmerized by the possibility of turning the art into algorithms exactly mirrors my experience. I too wondered if there was a way to develop Chuck Close's mosaics with some image manipulation, but for small project I instead focused on interpreting Sol LeWitt's wall drawings as code first.

Here's a shameless plug of a very rudimentary React/WebCanvas project I've been working on: http://artgorithms.s3-website-us-west-2.amazonaws.com/

Great work! I made a whole interface for easily generating this kind of art in the browser

http://blog.zmitri.com/studio/

open source: https://github.com/dmitric/studio

If you want to add some "shaders" would love to collaborate in some way.