AI generated images are everywhere and while i can think anything about authorship or originality, i have to admit that playing with it is real fun and results are sometimes astonishing. i am not interested in generating images from word description. But using fractal structure as a base - covering existing structures with various surfaces or "augmenting" it with industrial/natural/scifi feel - could be a lot more interesting.
i have almost zero experience with ai web interfaces so i took inspiration from gallery of https://www.deviantart.com/isider (thank you very much, dear Isa!) just because i really like style and results.
Link to ZOO ui (used models: depth, normal): https://zoo.replicate.dev/controlnet
base fractal images: jpg, 1920x1080 (yes, could be png but it is just fun),
i got 3x2x6 images (3 inputs, depth/normal model, 6 variations for each), took the best
results up-scaled twice with Waifu2x with no additional post process.
455z2c3d.jpg and results (2x depth model/2x