site banner

Small-Scale Question Sunday for November 6, 2022

Do you have a dumb question that you're kind of embarrassed to ask in the main thread? Is there something you're just not sure about?

This is your opportunity to ask questions. No question too simple or too silly.

Culture war topics are accepted, and proposals for a better intro post are appreciated.

4
Jump in the discussion.

No email address required.

Is there a good starter guide for AI image generation in terms of how to use tags, and which? For example, if im trying to get an image of a nuthatch wearing a navy captain's hat, using Stable Diffusion (niche, I know, but I figured a niche example would be best). I seem to really be struggling in how to set up my tags.

The Stable Diffusion subreddit has a decent links post with lots of guides. However, at this point, the ability of the model to generate highly specific images like that is very limited, and it'd likely take less time to generate a picture of nuthatch and then use inpainting to generate a navy captain's hat on its head than to find the right prompt-setting combination to get the image you want.

Baen just recently published a book on the subject, though I haven't read it myself.

https://www.baen.com/an-illustrated-guide-to-ai-prompt-mastery.html

The best (only?) way I know of is to check and imitate the tagging of the training sources, like stock image sites and danbooru. There are amusing anecdotes about the ai having an odd understanding of "balls" due to conflicting tagging.