I'm gonna hope this is the new coomer booba thread so just in case the anon is still here and lurking
>>58022116From my limited knowledge of AI sloppa genning here's a basic explanation of what types of files to what and how to gen your own
1. Download any stable diffusion GUI program. On /g/ the /ldg/ and /sdg/ threads have links to recommended programs. There's many options like EasyDiffusion, SwarmUI, Forge, ComfyUI, etc. I haven't tested them all and some are more user/beginner friendly than others. I'd imagine most of them probably install all the required things for you so it should be easy to get the program up and running. I use SwarmUI
2. The most important thing to have is an AI "model" or I believe the more appropriate term is "checkpoint". The model most of these programs originally used was called Stable Diffusion so there's a lot of misnaming thinking all AI is stable diffusion like calling all tissue paper kleenex or all gelatin jello. Many private corporations like NovelAI, Meta/Facebook, Google, Bing, etc all have their own models that they don't share with consumers cause they want you to specifically use their services. Basically the model or "checkpoint" is the brain of the AI and is what limits what kind of things the program can create. Like I believe the original model didn't know what the fuck Will Smith looked like so it just shat out garbage when you would prompt it. Later models were specifically trained on anime art so that they can put out pics of Pokegirls and porn better since it actually knows what it is. Think of it like if you told an 80 year old boomer to draw a Pokemon they wouldn't know what the fuck to do compared to if you told a zoomer/millennial to do so. Some of the more popular checkpoints/models are PonyDiffusion, Illustrious, etc.
cont.
