(check v1.0 page for more images)This workflow generates a person twice, on the same background at the same pose, and concatenates pictures together. Possible applications:<ul><li>Show person clothed/nude</li><li>Compare different outfits</li><li>Compare characters</li><li>Test LoRAs</li></ul><h2 id="heading-3506">Installation and dependencies</h2><ol><li>Install <a target="_blank" rel="ugc" href="https://github.com/ltdrdata/ComfyUI-Impact-Pack">Impact pack</a> custom nodes;</li><li>Install <a target="_blank" rel="ugc" href="https://github.com/Fannovel16/comfyui_controlnet_aux">ControlNet Auxiliary Preprocessors</a> custom nodes;</li><li>Download <a target="_blank" rel="ugc" href="https://dl.fbaipublicfiles.com/segment_anything/sam_vit_h_4b8939.pth">ViT-H SAM model</a> and place it in "\ComfyUI\ComfyUI\models\sams\";</li><li>Download <a target="_blank" rel="ugc" href="https://huggingface.co/lllyasviel/ControlNet-v1-1/tree/main">ControlNet Openpose</a> model (both .pth and .yaml files), and put it into "\comfy\ComfyUI\models\controlnet";</li><li>Download and open this workflow.</li></ol><h2 id="heading-3507">Resources</h2>Any model, any VAE, any LoRAs. If you choise SDXL model, make sure to load appropriate <a target="_blank" rel="ugc" href="https://huggingface.co/lllyasviel/sd_control_collection/tree/main">SDXL ControlNet model</a>.<h2 id="heading-2899">How to use</h2><ol><li>Write prompt and negative for background. Keep everything not related to person there, including tags like "masterpiece" or "raw photo";</li><li>Write prompt and negative for left person;</li><li>Write prompt and negative for right person;</li><li>Select model, tune settings if you want;</li><li>Run it!</li></ol><h2 id="heading-2900">OnOff_Upscaler</h2>In archive, there is the companion workflow called "OnOff_Upscaler". It splits your OnOff picture in two, upscales both, and conjoin them back. It will load prompts automatically.<ol><li>Install <a target="_blank" rel="ugc" href="https://github.com/WASasquatch/was-node-suite-comfyui">WAS Node Suite</a> custom nodes;</li><li>(optional) Install <a target="_blank" rel="ugc" href="https://github.com/chrisgoringe/cg-prompt-info">prompt-info</a> custom nodes;</li></ol><h2 id="heading-2901">Notes</h2><ul><li>In the workflow you can find the variable called "Background insertion step". Increase it if background feels inconsistent, and decrease if persons look copy-pasted;<img src="https://image.civitai.com/xG1nkqKTMzGDvpLrqFT7WA/d343fa47-4714-47db-ac2a-8b9a2b256f2e/width=525/d343fa47-4714-47db-ac2a-8b9a2b256f2e.jpeg" /></li><li>OpenPose ControlNet struggles with anime, latex and obscured bodies. If clothes is skintight, you can use Depth ControlNet instead, it will work great;</li><li>Left image is generated first. If you want to swap them at the end, just do it on the second to last node called "ImageCompositeMasked". Useful with LoRAs forcing specific pose.</li><li>If you only want to join two different pictures together, without control, look for simplistic <a target="_blank" rel="ugc" href="https://civarchive.com/models/298159?modelVersionId=334920">BeforeAfter workflow</a> instead.</li></ul>Sometimes, there is a bug with loading values in primitives. Default values are:OnOff workflow:<ul><li>Steps: 25</li><li>Background insertion step: 11</li><li>CFG Scale: 6.6</li><li>Sampler: dpmpp_sde_gpu</li><li>Scheduler: karras</li><li>Width: 512</li><li>Height: 768</li></ul>OnOff_Upscaler workflow:<ul><li>Upscale method: bilinear</li><li>Steps: 27</li><li>CFG Scale: 8.0</li><li>Sampler: dpmpp_2m</li><li>Scheduler: karras</li><li>Denoise: 0.41</li></ul><a target="_blank" rel="ugc" href="https://civarchive.com/user/Postpos/models">Check out my other workflows</a>

OnOff | ComfyUI workflow

Automatic prompt read for OnOff_Upscaler. 22.02.2024

Files with SHA256: 2875ef03778264e49688ce1549fb85e0aadbd449a44f5528b3207487a31ad35b