Hi everyone,This is a LoRA that supports NSFW content generation, however, tweaking with weights can also result in SFW contents.**This version only works on SDXL 1.0 models**The results are generally high quality if not overloaded with irrelevant and unecessary prompts. The model which I suggest would be:<ul><li>Helloworld1.0</li><li>you can also try out with other SDXL 1.0 models</li></ul>Include terms like Korean, Cinematic, Octane render, realistic, ray traced, etc for the best resultsThe recommended weight lies anywhere between 0.4 to 0.8, more than this would result in ignorance of other prompts.<ul><li>CFG scale could be as high as 11 or even 12</li><li>Scheduler and samplers: DPMPP 2M SDE Karras, Euler ancestral, DPMPP 2M (SDE)</li><li>Vae should be default</li><li>steps: 35</li><li>Upscaler: ERSGAN 4x</li><li>Always choose a higher resolution else the faces will be messed up. Best is 768 x 1024 for details</li></ul>For more such LoRAs or Merges you can chat with me and follow. Thank you!

Files with SHA256: 6a6b80a7bd68cd093c1ce5ee5b6f35127a0847568c36d81d076ce2115fa76113