Thanks! Mostly I start with either photographs I’ve taken or just random scribbles I’ve drawn in paint programs, arrange them on a fixed grid, randomize their order and do some minor filtering on them in Processing, run them through some select modules in TD (sometimes I skip this part, but displacing elements and merging them together in real-time can do wonders), and then I do most of the heavy lifting with Filter Forge (oftentimes programming a filter, resampling, and going back to the nodes multiple times). By the end, I send them through Nik collection to try to give them a washed-out, lomo type of feel, but even this last part gets skipped a lot because it doesn’t need any more interest (the ones without vignettes are a dead giveaway of those which are skipped, because I go crazy in that department most of the time).
I definitely want to expand this with Pillow and more P3 libraries when I get more knowledgeable and comfortable using them for different things, but FF is pretty damn cool for generative visual programming