Sup people. I was away for some time due to personal stuff yet i came back. I did try to repeat things i did on tensor but achieved very little succes. I am not very smart in technical things, so i don't really know or want to dip down in all of it which is ironic, because people call Ai "not art" cause it is to easy to make good images with it. So yea, I wasn't able to repeat what i did with same kind of setting and models, they do have something out there to make every image look much better i supose as well as some lora which was essential to my type of "artwork". I did try Pony and SDXL for few months, it was slow and painful, prompt adherense is abysmal, loras are pretty shitty if you want to make imposible thing in realistic style. I guess i didn't try ALL the things, and i assume you can get better results with impainting yet i'm just a silly gooner like you guys and i simply want to generate a thing, do my thing and move along, not to sit hours upon hours inpainting penises. The right decission ocured to me not very long ago and i can say I'm even kinda proud with result I'm getting with Z-image turbo. It was much faster to setup (partly cause i was already familliar with Comfy UI at this point) , checkpoint themselves are better in terms of giving good results without 10 Loras (with Z-image i use like 2 or 3) and the fucking machine hears my fucking prompt... I just wanted to say what i'm doing something good and will be sharing some of it in the future. It won't be in that crazy ammount i was trying to keep up with (100 images a week is a lot, belive me) , but i hope it will be enough for you guys. PEACE.
