⚠️Experimental LoRa⚠️
I recommend a typical weight of 1 to get the closest style to ratatatat74, but since this used a pretty good dataset of mine, it should be resilient even at 0.7 weight. As the name suggests, this is Illustrious-only LoRa, I used 100 images to train this, upscaled using ESRGAN anime 4xplus_a to ensure the lines are preserved even once the bucket cropping is applied. Hiresfix is also a great addition to your generations, as it adds new details to your gens while also upscaling it to the max 2k res.
Update(v2, NoobAI-base): I made major changes to my approach, and learned a lot since I first made this model.
