jhj SDXL Workflows — v1.0

jhj SDXL Workflows is for txt2img, img2img, inpainting, ControlNet, and IPAdapter image workflows; it uses text or loaded-image inputs and lists SaveImage for image output.

The available variants are txt2img, img2img, inpaint, txt2img ControlNet, img2img ControlNet, and IPAdapter.

The listed processing nodes include CLIPTextEncode, KSampler, VAEEncode, VAEDecode, LoadImage, SaveImage, ControlNetApplyAdvanced, ControlNetLoader, IPAdapterModelLoader, and UltimateSDUpscale.

Use a text prompt with a txt2img variant or load an image for an image-based variant; SaveImage is listed for saving the image result.

Suggestion · not verified

This v1.0 workflow set is published by Jhj0517 and based on SDXL 1.0.

Named model files include 4x_foolhardy_Remacri.pth, sd_xl_base_1.0.safetensors, sdxl-vae.safetensors, xinsircontrolnet-openpose-sdxl-1.0.safetensors, ip-adapter-plus-face_sdxl_vit-h.safetensors, and open_clip_model.safetensors.

Named components include AIO_Preprocessor, CheckpointLoaderSimple, CLIPSetLastLayer, CLIPTextEncode, CLIPVisionLoader, ControlNetApplyAdvanced, ControlNetLoader, CR Apply LoRA Stack, CR Aspect Ratio, EmptyLatentImage, GetImageSizeAndCount, InpaintModelConditioning, and IPAdapter.

Other named components include easy loraStack, IPAdapterModelLoader, KSampler, LoadImage, PreviewImage, SaveImage, UltimateSDUpscale, UpscaleModelLoader, VAEDecode, VAEEncode, and VAELoader.

To use upscaling, enable the additional upscale nodes; they are disabled by default.

Suggestion · not verified

For SD1.5, switch checkpoints; Pony or Illutrious is listed as SDXL fine-tuned and working with this workflow.

Suggestion · not verified

The workflow version is controlled in https://github.com/jhj0517/ComfyUI-workflows/tree/master.

Before choosing a variant, verify that the named model files and components are available in your setup.

Suggestion · not verified