ComfyUI nodes for optimized inpainting workflows
Top 46.3% on sourcepulse
This repository provides ComfyUI nodes for efficient inpainting by cropping the masked area, processing it, and then stitching the result back into the original image. It targets users of ComfyUI who want to speed up inpainting workflows, improve prompt accuracy, and control resolution for detailed generation or artifact avoidance.
How It Works
The core functionality is split between two nodes: 'Inpaint Crop' and 'Inpaint Stitch'. 'Inpaint Crop' intelligently extracts the masked region, optionally expanding it for context, resizing it to a target resolution (including downscaling to prevent artifacts or upscaling for detail), and handling mask adjustments like filling holes or expanding boundaries. The 'Inpaint Stitch' node then seamlessly integrates the processed inpainted region back into the original image, preserving unmasked areas and automatically blending the seams. This approach optimizes processing by focusing only on the relevant area, unlike full-image sampling.
Quick Start & Requirements
git clone https://github.com/lquesada/ComfyUI-Inpaint-CropAndStitch.git
into custom_nodes/
.Highlighted Details
Maintenance & Community
Licensing & Compatibility
Limitations & Caveats
The project relies on ComfyUI and its ecosystem. While the nodes are designed for precision, complex mask manipulations or extreme resizing factors might still require careful parameter tuning.
2 months ago
1 day