@estherschindler Cool. Archivists train for this so for TMI try searching for "descriptive metadata" and "images/video/audio" and for optional bonus points "controlled vocabulary" to learn more tips and tricks.
Yet another link:
https://daily.jstor.org/metadata-for-image-search-and-discovery/
@estherschindler As an aside, I've found the llava-v1.5-7b-q4 LLM is shockingly competent at describing images, especially since it can run on my laptop. The results are often good enough to use as grist for a professional entry.
https://huggingface.co/liuhaotian/llava-v1.5-7b