Improvements are currently being explored.<ul><li>Sampling: DPM++ 2M or DPM++ 2M SDE</li><li>VAE: sdxl_vae.safetensors</li><li>ScheduleType: Beta</li><li>Step: 20</li><li>CFGScale:4</li><li>Upscaler: R-ESRGAN 4x+ or Remacri<ul><li>Step:16</li></ul></li></ul>Please use ADetailer if necessary.Background rendering may feel slightly weak, and output stability can vary depending on settings and prompts.In some cases, better results can be achieved without using an upscaler.This model is designed with an emphasis on dynamic, powerful posing.v1.1 Changed to look like anime. If there is a model that is good at drawing backgrounds, I would like to merge it.If the image is distorted in v1.2, it is likely that the upscaler is causing the problem.

Files with SHA256: f271dcd6737a35b6a9eca8ad3dab4a0dd9750f92430a37d6639d53779ed54fb5