Just to clarify -- the images on that site are all generated from LLM-written code? It's very impressive if so. I would never have guessed these weren't human-made from the visuals alone.

Yeah it's entirely LLM driven! If you look at the repo, I explain more about how to use it. Basically, you give an agentic coding harness capable of tool usage (Opus 5.5, Qwen 3.8-27b, etc) the skill which explains the drawing system: simple things like how to set the colors, position rectangles, draw polygons, etc. and it uses p5.js to render them.

It has three different modes:

• You can simply describe a picture, and it'll automatically generate an image using procedural code.

https://kq-styles.specr.net/#/scene/custom/elk-valley

• You can ask it to paint over an existing image. That's what those examples show, like me creating EGA-style equivalents of various scenes from the original Myst.

https://kq-styles.specr.net/#/scene/myst/r01-c00

• The last way is to give it a rough pencil sketch. It's the mode I use the most since it's very valuable when building an adventure game to position objects in a room.

https://kq-styles.specr.net/#/scene/sketches/12-crossroads