All preview images were made before pruning, but it should not reduce quality of generated images!Form for <a target="_blank" rel="ugc" href="https://forms.gle/dxxZx3sf1jhxYj7XA">requests</a>.I also sometimes do polls on my <a target="_blank" rel="ugc" href="https://www.patreon.com/ChameleonAI">Patreon</a><a target="_blank" rel="ugc" href="https://civarchive.com/models/18031/patreon.com/ChameleonAI"> </a>for the next LORA (no payment required).With this mix I wanted to have a model that is mainly focused on anime style images. I wanted to have a Model that stays close to my other mix in terms of style while improving it a bit on the rpg site of things. It still struggles quite a bit for non humanoids, but I've managed to get some really pretty stuff out of it :) My settings:Using Hires.fix is very recommended!!!Sampler: Euler a, Steps: 40, CFG scale: 5 / DPM++ 2M Karras, Steps: 30, CFG scale: 5Negative prompt: (worst quality, low quality:1.4), EasyNegative , bad_prompt_version2 Hires.fix: Laten(nearest-exact)/4x-UltraSharp, Hires steps same as Sampling steps, Denoising strength ~0.5, Upscale by 1.5Used embeddings for sample pictures:<a target="_blank" rel="ugc" href="https://civarchive.com/models/7808/easynegative">EasyNegative</a><a target="_blank" rel="ugc" href="https://huggingface.co/datasets/Nerfgun3/bad_prompt">bad_prompt_version2</a> Used Lora for the three Watame pictures:<a target="_blank" rel="ugc" href="https://civarchive.com/models/9430/tsunomaki-watame-4-outfits-or-hololive">Tsunomaki Watame</a>If you have any feedback or questions just write in the comments or hit me up on <a target="_blank" rel="ugc" href="https://www.reddit.com/user/ChameleonAI">reddit</a> or discord.

ChameleonAI "RPG" Mix

chameleonaiRPGMix_v10.safetensors

v1.0

With this mix I wanted to have a model that is mainly focused on anime style images. I wanted to have a Model that stays close to my other mix in terms of style while improving it a bit on the rpg site of things. It still struggles quite a bit for non humanoids, but I've managed to get some really pretty stuff out of it :) My settings:Using Hires.fix is very recommended!!!Sampler: Euler a, Steps: 40, CFG scale: 5 / DPM++ 2M Karras, Steps: 30, CFG scale: 5Negative prompt: (worst quality, low quality:1.4), EasyNegative , bad_prompt_version2 Hires.fix: Laten(nearest-exact)/4x-UltraSharp, Hires steps same as Sampling steps, Denoising strength ~0.5, Upscale by 1.5

&lt;p&gt;pruned, baked in vae version&lt;/p&gt;

Files with SHA256: ca0e3359f710824b2aa3141e1f000361bc05f3b1c52d2420055f35eec422ee36