My first nodes has been released (Patreon)

After the beta release and multiple iterations, I am delighted to launch my first node pack: NO8D-control. It greatly optimizes the user experience based on the beta version, adds support for Nodes2.0, and makes your experience of using LORAs far simpler and more convenient.

This node pack brings tremendous convenience for you to call LORAs. It enables you to apply multiple LORAs to a single image simultaneously and specify target areas for their effects. Thanks to its robust support for slider-LORA, you can linearly adjust imaging results within designated regions. Its specific structure consists of three parts:

NO8D-LoRA stack

This is a multi-LoRA loading manager that supports the stacked use of multiple LoRAs/slider-LoRAs.

Click Settings to adjust the weight range of this LoRA. Configure it once for repeated reuse, eliminating the need to memorize the recommended weight for each LoRA.

You can use the up and down arrow keys on your keyboard to adjust values in the number input box: the up/down keys adjust by ±0.1, while Shift + up/down keys adjust by ±0.5.

Click the ⸬ in front of a LoRA to drag and reorder it.

Tap the 👁️ icon to disable the LoRA, which will no longer take effect on image generation.

Click the button inside the red box to invert the disabled LoRA selection.

These features greatly facilitate you in combining various LoRA sets.

NO8D-Inpainting

I have rewritten this node, making it more intuitive and user-friendly to operate.

Strictly speaking, this is a node integrating Ksample and Inpainting.

Here you can view the standard options for Ksample. I have simply streamlined how seed values are modified; now it only has two states: Random and Locked, controlled via the 🔓 icon.

The row below is for mask drawing, which functions essentially the same as Inpainting. It offers two mask drawing tools: a brush and a lasso. It is worth noting that when this section is not activated, the denoising strength is set to 1.0, allowing generated images to retain sufficient variation. Once you start drawing a mask, the denoising strength automatically switches to 0.75 — this value was determined as optimal after extensive testing. It maintains a reasonable degree of variation while preventing disjointed transitions between masked and unmasked areas. Naturally, you may adjust this value manually at your discretion; all settings are fully customizable.

Two icons in the bottom-left corner of the preview window: ◐ ⟳

◐: Invert Mask

⟳: Clear Mask

The bottom row displays historically generated images. After multiple generations, you can pick your preferred ones. Note that a maximum of 8 historical images are stored. You can draw masks on any historical image, and the node will regenerate new visuals based on that selected historical image.

NO8D-A/B preview

Generated images are displayed here with A/B comparison support. A stands for the newly generated image, and B refers to the image generated in the previous round, making it easy for you to compare the differences between two generation outputs.

If you want to compare the current image with images from previous rounds, the node also provides historical images, with a maximum support of 8 images.

Alright, those are all the functions of this node group. Thanks to Wylmquest and Seraph91P [COS] for their suggestions, and we hope you are satisfied with this version. This node group has been released on GitHub and can be accessed via the corresponding link. It is extremely lightweight and requires no dependencies to install.

https://github.com/no8d/ComfyUI-NO8D-controls

In addition, I have provided a simple workflow, which you can access below

Click here

Do you think this set of node groups is helpful for your work? Or are there any new features you would like to add? Feel free to leave a message and share your thoughts with me.