I turned an LLM into a Cinematic Visual Prompt Architect — Sharing the Framework

Been testing this for a while and decided to share. I used to think better AI images were mostly about finding the right keywords and artist