
This is a ComfyUI workflow that includes loading a checkpoint, loading LORAs, diffusion with either straight or additionally the SDXL Lightning LORA for fast/test pics, and the optional selection of 5 different kinds of upscalers; pixel space, iterative, Ultimate SD, SUPIR, and SUPIR plus an additional Ultimate SD step to get up to 8x. In prompting you can toggle on or off the addition of Pony tags to your positive prompt.
You will need to load in extra custom nodes including rgthree, KJN, SUPIR, Impact, UltimateSD, Primere off the top of my head. I recommend for new starters to install ComfyUI Manager and rgthree from git, then fire up ComfyUI, load this workflow, and use ComfyUI Manager to install the other missing nodes and models. From the command line you can also use the __update_comfy__.py script in the rgthree folder to keep everything up to date with git, or use ComfyUI Manager from the gui. The script will update ComfyUI and all your custom nodes.
I have a RTX3090 and so the tile encoder/decoder sizes are set larger as there's stacks of vram in that card - you might want to drop it down if your hardware is lower spec.
Many thanks to rgthree and the other people involved in the community creating and maintaining custom nodes
描述:
Initial public release
训练词语:
名称: comfyuiSDXLWorkflowWith5_v10.zip
大小 (KB): 11
类型: Archive
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success