Feel free to try Joycaption for captioning : https://lemmy.world/post/30096816

I’ve implemented that kind of system in my generator: https://perchance.org/fusion-ai-image-generator

I invoke it by letting the user write , for example ,

“photo of a #adjective# #name#”

and then prior before sending the prompt to the generator I run

“photo of a #adjective# #name#”.replaceText(“#adjective#”, adjectives.selectOne).replaceText(“#names#”, name.selectOne)

to get the randomized string

//----//

In Stable Diffusion the “prompt” is not an “instruction” as in ai-chat.

The Stable Diffusion model never refers to the prompt when generating the image.

Instead, it converts the string to numbers and runs them through layers and layers of matrices.

Like a ball tumbling down aling the pins in a pachinko machine.

SD Prompting is very “psuedo-science” since nobody really knows what this kind of grid can and can’t do with the “ball” you throw into it.

Since I’m building a randomizer around this , I’ve covered a lot of topics on this in the fusion discord: https://discord.gg/dbkbhf8H2E

//—//

Also;

NSFW topics are better asked at https://lemmy.world/c/casual_perchance.

Questions asked at this forum are better if written in a SFW context.