Simple Inpainting Workflow for any model (SDXL, Flux, etc) - Version 2

5.0

1 reviews

8.6K

5.1K

Description

This is the second workflow for this YouTube tutorial: https://youtu.be/mkitinPOlY4

It's my go to, easy to use, Inpainting logic.

You input an image, create a mask in the Mask Editor, and prompt for what you want to change.

The only parameters you'd want to look at and fix to your likings, in order of importance, are:

- ControlNet Strength

- Grow Mask With Blur (Growth, Blur)

Models:

- Xinsir Depth (SDXL): https://huggingface.co/xinsir/controlnet-depth-sdxl-1.0

- ControlNet Union Pro (Flux): https://huggingface.co/Shakker-Labs/FLUX.1-dev-ControlNet-Union-Pro/tree/main

- Flux Q4_O GGUF quant: https://huggingface.co/city96/FLUX.1-dev-gguf/blob/main/flux1-dev-Q4_0.gguf

Node Diagram

To show the workflow graph full screen.

To toggle the lock state of the workflow graph. In the locked state, you can pan and zoom the graph. In the unlocked state, you can select, move and modify nodes. The graph is locked by default.

Discussion

siamese_pesky_724 months ago

Thank you for sharing this impressive workflow! As a beginner, I've learned a lot from studying it.

Quick question about the image scaling: While the input image is properly scaled to dimensions divisible by 8, the cropped section is only scaled to 1 megapixel before going into the inpainting module. Shouldn't this section also be scaled to ensure dimensions divisible by 8? Or am I missing something here?

👍1

learning_in_20234 months ago

The preview image in the Flux Group gives perfect inpaint but it is cropped. The preview image in the Flux above it always has shades. How can we get the full picture in the Flux group or fix the shades in flux group above it?

(Edited)

Author

andrea baioni

123.2K

1.9K

365.4K

Reviews

ZiYao Hu

4 months ago

Versions (1)

- latest (5 months ago)

Node Details

Primitive Nodes (10)

DepthAnythingV2Preprocessor (2)

FluxGuidance (1)

Note (5)

SetUnionControlNetType (1)

UnetLoaderGGUF (1)

Custom Nodes (61)

ComfyUI

- CLIPTextEncode (4)
- ControlNetLoader (2)
- DifferentialDiffusion (2)
- KSampler (2)
- CheckpointLoaderSimple (1)
- InpaintModelConditioning (2)
- PreviewImage (10)
- DualCLIPLoader (1)
- VAELoader (1)
- VAEDecode (2)
- ImageToMask (4)
- MaskToImage (2)
- ControlNetApplyAdvanced (2)
- LoadImage (1)

ComfyUI Essentials

- SimpleMath+ (2)
- GetImageSize+ (5)
- ImageResize+ (7)

comfyui-art-venture

- ImageScaleToMegapixels (3)

KJNodes for ComfyUI

- GrowMaskWithBlur (2)

WAS Node Suite

- Bounded Image Crop with Mask (4)
- Bounded Image Blend with Mask (2)

Model Details

Checkpoints (1)

Juggernaut-X-RunDiffusion-NSFW.safetensors

LoRAs (0)

OpenArt

Workflows

Active Sessions