The Illustrious version of a realistic merge designed for generating good results from NSFW concept LoRAs and transformations. I personally use this model to test out my IL LoRAs. I haven't been able to put much time into the merges yet - I'm mostly posting this just so my LoRA generations have a base model to show.<h2 id="generating-upead36t4">Generating</h2>I found the best results with forge to be:<h3 id="txt2img:-fncgty0wz">txt2img: </h3><ul><li>864x1152</li><li>cfg: 8</li><li>sampler: euler a</li><li>scheduler: beta</li><li>steps: 30</li></ul><h3 id="img2img:-640oqcjsp">img2img: </h3><ul><li>sampler: DPM++ SDE</li><li>scheduler: beta</li><li>steps: 30</li><li>resize by 1.5x</li><li>denoise 0.15</li><li>cfg 6</li><li>Script: SD Upscale -&gt; ESRGAN 4x+ @ scale factor 2</li></ul>With this flow, img2img is absolutely necessary to get a good result. Do not skip it!<h3 id="tips-zblrou8qc">Tips</h3>Using "anthro", "furry", or "scalie" will make the results less human.Specifying a hair color, facial features, or expressions (like "ahegao" or "smiling") will make the results more human.

Files with SHA256: dd28b276c2dbd43e330c5787802c91fed995c385e58c775de24d5a9152e7633d