AI 558 🏴‍☠️🏴‍☠️ (Patreon)

https://civitai.com/user/FilthyDegenerate

Spent the last two hours testing different upscalers for img2img and also many many different settings to see how far I can push it, I want some sort of 'universal' settings like a merged model so I don't have to fiddle with it every 10 seconds.

Sadly using weeb LoRA's on realistic works tends to degrade the quality of the render, it smooths the skin, I tried to install OneTrainer but failed, asked the 'big brains' at reddit but got nothing, the SD sub is dogshit and people there don't know shit (just like SD youtubers), some clown did insist that I was not properly giving the correct path to my LoRA folder when I specified multiple times I was, this is the problem with using Python shit or stuff aimed at Linux/English systems, it never works properly on Windows.

Plus it does not help I'm using a fucking PT-BR OS which complicates things even more, can't even install Koyha because full of errors and I won't use NMKD either because it takes an hour to train basic shit, I was planning on making realistic LoRA's and post then here and Civitai, but that ain't happening.