Node packVerifiedExternal

wlsh_nodes

WLSH ComfyUI Nodes

Sourcegithub.com/wallish77/wlsh_nodes

Maintainer@wallish77Cgrade

Package profile

Status: Verified
Publisher: @wallish77
Source: GitHub
Stars: 146

README

WLSH ComfyUI Nodes

A set of custom nodes for ComfyUI created for personal use to solve minor annoyances or implement various features. Includes nodes to read or write metadata to saved images in a similar way to Automatic1111 and nodes to quickly generate latent images at resolutions by pixel count and aspect ratio.

Node descriptions

Loaders

Node Name	Description
Checkpoint Loader with Name	The regular checkpoint loader with an output that provides the name of the loaded model as a string for use in saving filenames

Samplers

Node Name	Description
KSamplerAdvanced (WLSH)	Modified version of the advanced KSampler to return the `denoising` option for experimentation. Also provides the INFO output that provides sampler settings for use in other nodes.

Conditioning

Node Name	Description
CLIP Positive-Negative	Simple CLIP box containing positive and negative prompt boxes. Mostly a space saver. Also available as an SDXL version.
CLIP Postive-Negative w/Text	Same as the above, but with two output ndoes to provide the positive and negative inputs to other nodes. Also available as an SDXL version
CLIP +/- w/Text Unified (WLSH)	Combined prompt/conditioning that lets you toggle between SD1.5 and SDXL (just make sure to change your inputs). Sacrifices some features of SDXL, namely by assuming POS_G and POS_L are the same (as well as the negative), assuming zero crop, and assuming target size is 2x input size. Width and height inputs are option, will assume 1024x1024.

Latent

Node Name	Description
Empty Latent by Pixels	Quickly generate an empty latent by selecting an aspect ratio, an orientation (landscape or portrait) and the number of megapixels (1MP = 1024 * 1024 in this case). Also provides the height and width as integers for plugging in elsewhere.
Empty Latent by Ratio	Quickly generate an empty latent by selecting an aspect ratio, an orientation (landscape or portrait) and the length of the 'shorter' side. Also provides the height and width as integers for plugging in elsewhere.
Empty Latent by Size	Nearly identical to the default empty latent node, but also provides the width and height as integer outputs for plugging in elsewhere.
SDXL Quick Empty Latent	Quickly generate an empty latent of a size from a list of ideal SDXL image sizes. Also provides the height and width as integers for plugging in elsewhere.

Image

Node Name	Description
Image Load with Metadata	Extract image metadata from Auto1111-sourced images or from images saved with the nodes in this pack. Can then connect these outputs to other nodes. Requires comfyUI commit d7b3b0f8c11c6261d0d8b859ea98f2d818b7e67d or newer to show preview of selected image
Grayscale Image	Takes an input image and returns the image in grayscale. Useful for some controlnet actions

Inpainting

Node Name	Description
Generate Border Mask	Used to generate a 'stripe' of mask with blurred edges that overlaps an original iamge with the outpainted new space. Use on a second pass of an image to 'redo' the masked stripe region and help smooth out the edges of an outpainted image. Pass output to a `Convert Image to Mask` node using the `green` channel.
Outpaint to Image	Extends an image in a selected direction by a number of pixels and outputs the expanded image and a mask of the outpainted region with some blurred border padding.
VAE Encode for Inpaint Padding	A combined node that takes an image and mask and encodes for the sampler. Can apply some padding/blur to improve inpainting behavior.