<h3 id="comfyui-workflow-for-single-image-generation-eijhsoh07">ComfyUI Workflow for Single Image Generation</h3><ul><li>Needed Custom Node: <a target="_blank" rel="ugc" href="https://github.com/r-vage/ComfyUI-RvTools_v2">RvTools v2 (Updated!!)</a></li><li>it's mainly designed for flux but if you bypass pulid and redux it's possible to use any checkpoint with this workflow</li></ul>

FLUXinator (STD/GGUF), PuLID, Redux, img2prompt, controlnet, F1depth/canny, Detailer, SUPIR

<ul><li>v2.01</li><li>for some tests i've changed the supir model to v0F but sometimes you can see big color stripes on the image, v0Q is the prefered model to use.</li><li>forget to update the SUPIR group,<ul><li>added lora stack and with the current setup it will kind of keep the skin details</li><li>removed the empty prompt option, enter the prompt into the SUPIR conditioner or remove it for empty prompt.</li></ul></li><li>changed the save behavior. first save node saves the json file now. images don't contain the workflow anymore.</li><li>v2.0</li><li>Reworked both Image to Prompt groups, it allows to remove certain things from the prompt (description of bg, mood, image)</li><li>added sage attention &amp; torch compile</li><li>checkpoint loader v4 and gguf quadruple clip loader</li><li>changed the location of the sampler settings, you can find it in the sampler group</li><li>replaced all vram purges with the "clean vram used" node wich allows to force a clean by sending the image or whatever through this node.</li></ul>

Files with SHA256: a5571542494347c6a9bda84373f4c2878f8c1a936fd9b75ab817300232520b86