craiyon seems to understand my intent better (esp with multiple things interacting in the scene) with prompts but SD does way better drawings. time to try img2img like this i guess
Ethan Gonzalez
how the hell do i get rid of the anime-style eyes? should i use the huggingface model? i wonder if theres a way to reduce the waifu eye training
Any poorfags here using Colab? How do you upload large files there? I want to try out lewd danbooru model, but when I'm trying to upload it, it just shows me red circle with no progress.
>have 2 tags for small breasts >still get huge melon results most of the time >get one result with normal breasts put it into img2img >all of the results are of huge breasts again
jesus just how many images of huge tits are in this thing
GOG FUCKGING DAMN IF I HVE TO REPEAT MYSELF ANOTHER TIME making a successful mememorph is 80% prompt and 20% fiddling with denoise until something comes out photo of everyday objects Steps: 30, Sampler: Euler a, CFG scale: 7, Seed: 4250397121, Size: 512x512, Model hash: 7460a6fa, Batch size: 3, Batch pos: 1, Denoising strength: 0.79
maybe we should make a guide and pin it in OP? the same OP that is at the character limit
Does anyone understand what the Var. Strength / Var. Seed options in the X/Y Grid UI do? There's no tooltips, the Var.seed does absolutely nothing for me, and the Var.strength does something like it's changing the seed? I don't get it.