<a target="_blank" rel="ugc" href="https://ko-fi.com/zyd232">Buy me a coffee ☕</a><h3 id="features-ut6ydj0rr">Features</h3><ol><li>Hires Fix</li><li>Face Detailer - Generate clear face when the face is too small.</li><li>Detailed Skin - Generate pores and skin details.</li><li>All base model compatible. (SD1.5, SDXL...)</li></ol><h3 id="requirement-ouud8kvdc">Requirement</h3><ul><li>Models<ol><li>Upscale: <a target="_blank" rel="ugc" href="https://huggingface.co/gemasai/4x_NMKD-Siax_200k/blob/main/4x_NMKD-Siax_200k.pth">4x_NMKD-Siax_200k.pth</a> (..\ComfyUI\models\upscale_models\)</li><li>Ultralytics: <a target="_blank" rel="ugc" href="https://huggingface.co/Bingsu/adetailer/blob/main/face_yolov8m.pt">face_yolov8m.pt</a> (..\ComfyUI\models\ultralytics\bbox\)</li><li>Sam: <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/Gourieff/ReActor/blob/main/models/sams/sam_vit_b_01ec64.pth">sam_vit_b_01ec64.pth</a> (..\ComfyUI\models\sams\)</li><li>Controlnet: <a target="_blank" rel="ugc" href="https://huggingface.co/xinsir/controlnet-tile-sdxl-1.0/blob/main/diffusion_pytorch_model.safetensors">controlnet-tile-sdxl-1.0..safetensors</a> (download and rename the file, put it into the path: ..\ComfyUI\models\controlnet\SDXL)</li></ol></li><li>Custom Nodes<ul><li>After importing the workflow for the first time, please use the "Install Missing Custom Nodes" button in <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Manager">ComfyUI Manager</a> menu to get all the nodes needed.</li></ul></li></ul><h3 id="tips-pa50y1pdg">Tips</h3><ol><li>Change the Checkpoint and LoRAs in the Group "Origin" and "RefineSkin" to your own models.</li><li>For better result of skin details, the checkpoint in Group "RefineSkin" should be a model which is good at generating sharp and clear details. <a target="_blank" rel="ugc" href="https://civarchive.com/models/133005?modelVersionId=456194">Juggernaut_X_RunDiffusion</a> is recommended.</li><li>Change value of cfg, steps and sampler, depending on used models.</li><li>Change the latent width and height to your favorite, in Node "Efficient Loader", Group "Origin".</li><li>If you want to change the prompts about the charater's face, edit the prompts in the Group "Origin" and "FaceDetailer" at the same time.</li></ol>__________________________________________________________________<h3 id="-i8omlgal0">特点</h3><ol><li>高清修复</li><li>面部修复 - 当脸部太小时，生成清晰的脸部。</li><li>皮肤细节 - 生成毛孔和皮肤细节。</li><li>兼容所有模型 (SD1.5, SDXL...)。</li></ol><h3 id="-wigpgtic4">需求</h3><ul><li>模型<ol><li>Upscale: <a target="_blank" rel="ugc" href="https://huggingface.co/gemasai/4x_NMKD-Siax_200k/blob/main/4x_NMKD-Siax_200k.pth">4x_NMKD-Siax_200k.pth</a> (..\ComfyUI\models\upscale_models\)</li><li>Ultralytics: <a target="_blank" rel="ugc" href="https://huggingface.co/Bingsu/adetailer/blob/main/face_yolov8m.pt">face_yolov8m.pt</a> (..\ComfyUI\models\ultralytics\bbox\)</li><li>Sam: <a target="_blank" rel="ugc" href="https://huggingface.co/datasets/Gourieff/ReActor/blob/main/models/sams/sam_vit_b_01ec64.pth">sam_vit_b_01ec64.pth</a> (..\ComfyUI\models\sams\)</li><li>Controlnet: <a target="_blank" rel="ugc" href="https://huggingface.co/xinsir/controlnet-tile-sdxl-1.0/blob/main/diffusion_pytorch_model.safetensors">controlnet-tile-sdxl-1.0..safetensors</a> (download and rename the file, put it into the path: ..\ComfyUI\models\controlnet\SDXL)</li></ol></li><li>自定义节点<ul><li>首次导入工作流程后，请使用 <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Manager">ComfyUI Manager</a> 菜单中的 “Install Missing Custom Nodes” 按钮获取所需的所有节点。</li></ul></li></ul><h3 id="-1mamhvwat">提示</h3><ol><li>将 “Origin” 和 “RefineSkin” 组中的 checkpoint 和 LoRA 更改为您自己的模型。</li><li>为了获得更好的皮肤细节效果，“RefineSkin” 组中的 checkpoint 需要是能够生成清晰锐利细节的模型。</li><li>根据所使用的模型，更改 cfg、步数和采样器的值。</li><li>在“Origin”组的“Efficient Loader”节点中，将潜空间图像的宽度和高度改为您需要的值。</li><li>如果要更改有关人物面部的提示词，请同步编辑 “Origin” 组和 “FaceDetailer” 组中的提示词。</li></ol>

zyd232's Workflow - Hires Fix + Face Detailer + Detailed Skin

All base models are compatible (SD1.5, SDXL...), depending on your choice.

Files with SHA256: 80468d60c505a132f141d55cf54d664ce82be2195da77076d9c22349a877c49c