独创的三步生图法，目的是解决lora、画师tag等对模型生图能力固化的现象，通过多步生成赋予最大的生成操作空间。
默认步数与放大系数在RTX3060上用时450秒~570秒左右，可以据此调整适合您的步数与放大系数。
默认模型仅供测试使用。您可以使用任何您喜欢的模型。
部分设置参数参考了Legendaer的工作流。
The unique three-step image generation method aims to address the phenomenon where models are overly constrained in their image generation capabilities due to factors such as LORA and artist tags. By employing multiple steps of generation, it provides the greatest operational space for the generation process.
This workflow is simple and easy to understand, including LoRa, amplification, facial restoration, and one-stop photo generation. The default number of steps and amplification coefficient take around 450 to 570 seconds on the RTX 3060. You can adjust the number of steps and amplification coefficient according to your needs.
The default model is for testing purposes only. You can use any model you prefer.
Some of the parameters were set based on the workflow of Legendaer.
我的推特:@ekll01
My Twitter: @ekll01
面部检测模型下载地址Anzhc Face seg 640 v4 y11n.pt · Anzhc/Anzhcs_YOLOs at main
眼部检测模型下载地址Anzhc Eyes -seg-hd.pt · Anzhc/Anzhcs_YOLOs at main
4x-UltraSharpV2_Lite模型下载地址https://huggingface.co/Kim2091/UltraSharpV2/tree/main
This workflow logic is relatively simple, but it includes lora, amplification, and repair.
The default model is only for testing purposes. You can use any model you like.
Face detection model download address Anzhc Face seg 640 v4 y11n.pt · Anzhc/Anzhcs_YOLOs at main
Eye detection model download address Anzhc Eyes -seg-hd.pt · Anzhc/Anzhcs_YOLOs at main
4 x - UltraSharpV2_Lite model download address https://huggingface.co/Kim2091/UltraSharpV2/tree/main
Description
Comprehensive optimizations have been made, including rebalancing partitions, reducing the number of load operations, and adding NSFW fixes.
全面进行了优化，包括重新调整分区、减少了加载次数、加入了NSFW修复。