ComfyUI-Flux-Inpainting

★ 45

图像修复外拓扩展低 VRAMFlux Fill

将 NF4 Flux Fill 模型封装为 ComfyUI 节点，支持图像 inpainting 与 outpainting，在更低 VRAM 条件下也能高效修复与扩展画面。

💡 在资源受限的设备上对图片缺损区域进行修补与画面扩展。

🍴 2 Forks💻 Python🔄 2025-05-14

🔗 GitHub 原文

📦

网盘下载

复制链接后前往夸克网盘下载

https://pan.quark.cn/s/a9fb3a59e10c

📦 requirements.txt

numpy
diffusers>=0.32.0
transformers>=4.45.0
bitsandbytes>=0.44.0
accelerate>=1.1.1
gguf>=0.10.0

📄 README

ComfyUI-Flux-Inpainting

Introduction

This repository wraps the flux fill model as ComfyUI nodes. Use NF4 flux fill model, support for inpainting and outpainting image. Compared to the flux fill dev model, these nodes can use the flux fill model to perform inpainting and outpainting work under lower VRM conditions

News

Dec 26, 2024: Add FluxGuffInpainting node to support loading gguf models

Dec 11, 2024: Add FluxSimpleInpainting, FluxTransformerInpainting, FluxVAELoader nodes can be used to split pipelines and comfyui clip and vae loader can be used instead

Installation

Method 1:

Navigate to the node directory, ComfyUI/custom_nodes/

git clone https://github.com/rubi-du/ComfyUI-Flux-Inpainting.git

cd ComfyUI-Flux-Inpainting

pip install -r requirements.txt

Restart ComfyUI

Method 2:

Directly download the node source code package, unzip it into the custom_nodes directory, and then restart ComfyUI.

Method 3:

Install via ComfyUI-Manager by searching for “ComfyUI-Flux-Inpainting”.

Usage

Example workflows are placed in ComfyUI-Flux-Inpainting/workflow.

FluxInpainting node workflow: Workflow Address

FluxSimpleInpainting node workflow: Workflow Address

FluxTransformerInpainting node workflow: Workflow Address

FluxGuffInpainting node workflow: Workflow Address

Nodes

FluxInpainting

FluxSimpleInpainting

FluxTransformerInpainting

FluxVAELoader

FluxGuffInpainting

`FluxGuffInpainting` special Instruction

The FluxGuffInpainting node only need to load the guff model and put them into the ComfyUI/models/unet folder and don’t need load other submodels from FLUX.1-Fil-dev

vae model download links: https://hf-mirror.com/YarvixPA/FLUX.1-Fill-dev-gguf/tree/main

`FluxTransformerInpainting` special instruction

The FluxTransformerInpainting node only need to load the transformer submodel from the FLUX.1-Fil-dev-nf4 model and don’t need load other submodels from FLUX.1-Fil-dev

`FluxVAELoader` special instruction

The FluxVAELoader node need to load the flux fill vae model from the ComfyUI/models/vae directory, so you need to download the model and place it in the ComfyUI/models/vae directory

vae model download links: https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev/tree/main/vae

Models

The node needs to load the transformer and text_decoder_2 submodels from the FLUX.1-Fil-dev-nf4 model and load other submodels from FLUX.1-Fil-dev, so these two models need to be placed in the models folder

Model download links:

FLUX.1-Fil-dev: https://huggingface.co/black-forest-labs/FLUX.1-Fill-dev

FLUX.1-Fil-dev-nf4: https://huggingface.co/sayakpaul/FLUX.1-Fill-dev-nf4

FLUX.1-Fill-dev-gguf: https://hf-mirror.com/YarvixPA/FLUX.1-Fill-dev-gguf/tree/main

The directory structure is as follows:

ComfyUI/models/
└── FLUX.1-Fill-dev
    ├── vae
    └── scheduler
    └── text_encoder
    └── ...
└── FLUX.1-Fill-dev-nf4
    ├── transformer
    └── text_decoder_2