ComfyUI-denoise-mask-scheduler

ComfyUI-denoise-mask-scheduler
★ 6

图像修补去噪掩码逐步调度实验性功能
该节点在图像修补的扩散采样过程中，按步选择性应用并逐步调整去噪掩码（通过 modelpatcher 的 denoise_mask_function），以减少明显的掩码边界并允许对非掩码区做轻微修改，当前主要支持跳过策略。
💡 在图像修补中降低掩码边界可见性并微调未掩盖区域。
🍴 1 Forks💻 Python🔄 2024-11-07
📦
网盘下载
复制链接后前往夸克网盘下载
https://pan.quark.cn/s/e58c8376a81b
image
image
image
image
image
📄 README

ComfyUI-denoise-mask-scheduler

ComfyUI-denoise-mask-scheduler experimental approach involves selectively applying a denoise mask at each step during the inpainting inference process in diffusion models.

Introduction

In inpainting tasks, the non-mask area should remain unchanged. However, this often results in visible mask boundaries. If minor alterations to parts of the non-mask area are acceptable, this approach can be quite effective.

At each sampling step, the denoise mask is progressively modified according to predefined rules using the denoise_mask_function of comfyui modelpatcher options. I refer to this process as the *denoise mask schedule*. This schedule can include transformations like mask skipping, random masking, and morphology adjustments.

However, only the skip option is currently implemented. New schedules will be continuously added as effective ones are identified.

Install

Download the repository and unpack into the custom_nodes folder in the ComfyUI installation directory.

Or clone via GIT, starting from ComfyUI installation directory:

cd custom_nodes
git clone https://github.com/MiddleKD/ComfyUI-denoise-mask-scheduler.git

Nodes

Skip mask and Apply alpha

  • using sampling step
  • using sigma value(noise strength)

    • Workflows

    Skip mask and Apply alpha

  • basic inpainting

    • *input/basic/denoise-mask-scheduler*

  • inpainting controlnet

    • *input/controlnet/controlnet+denoise-mask-scheduler (look at the forehead)*