Generates a still-life detailer from a single image.

An ollama LLM can be added to further automate the process, but it is risky since the wan2.1vace model reacts strongly to verbs and any verbs will potentially ruin the still-life nature of the output.