If you want to support me, bidding buzz on the weekly <a target="_blank" rel="ugc" href="https://civarchive.com/auctions">auctions page</a> for the model helps me keep it up for on-site generation.<h2 id="recommended-settings-:">Recommended settings :</h2><ul><li>Positive prompt : masterpiece, best quality</li><li>Negative prompt : worst quality, low quality, bad anatomy, watermark, censored</li><li>Sampler : Euler a</li><li>CFG Scale : 5-7</li></ul>I highly recommend using the <a target="_blank" rel="ugc" href="https://github.com/Bing-su/adetailer">ADetailer</a> extension for better faces and eyes.For v2, I merged v1 with some NoobAI models, @GarejiSama's <a target="_blank" rel="ugc" href="https://civarchive.com/models/1649312?modelVersionId=1866796">AdAstra</a> and @Manityro's <a target="_blank" rel="ugc" href="https://civarchive.com/models/1732765/vermilion">Vermilion</a>, making v2 much more compatible with NoobAI based LoRAs. The baked in VAE is Anzhc's, which you can get <a target="_blank" rel="ugc" href="https://huggingface.co/Anzhc/Anzhcs-VAEs/blob/main/SDXL%20Anime%20VAE%20Dec-only%20B3.safetensors">here</a>, gives a much sharper look to images, reduced noise, and makes colors less bland overall. LustrousMix v2 generations are more colorful than v1's, with more contrast and sharpness. There is a slight art style change as well, with more mature faces. <img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/babdaa02-2197-4a29-9b37-13d2467db4e2/width=525/babdaa02-2197-4a29-9b37-13d2467db4e2.jpeg" /><h3 id="hope-you-enjoy!">Hope you enjoy!</h3>

Files with SHA256: 5c0e3ed489766001c68f1fb983375179a4c9dc614bfe1cf29db59f7fa8268f1e