playing around with an LLM a bit to generate images to see if my description contains enough details before I bother an actual artist with them. The results are (unsurprisingly) awful but at least the "thinking" text is funny