These files are Custom Nodes for ComfyUI
ComfyUI is a super powerful node-based, modular, interface for Stable Diffusion. I have a brief overview of what it is and does here. And full tutorial content coming soon on my Patreon.
In this model card I will be posting some of the custom Nodes I create. Let me know if you have any ideas, or if there's any feature you'd specifically like to see added as a Node!
Please consider joining my Patreon! Advanced SD tutorials, settings explanations, adult-art, from a female content creator (me!) patreon.com/theally
Description
Drop in your ComfyUI_windows_portable\ComfyUI\custom_nodesfolder and select the Node from the Image Processing Node list. Allows for image sharpening (or un-sharpening) to be applied prior to the save step.
FAQ
Comments (29)
Make a node that takes part of the latent noise and stamps it somewhere else with options to reflect vertically and/or horizontally and to resize it
This was actually something I was thinking about - like WebUI's Latent Mirroring! Thanks!
I just found, there's already a latent mirroring Node which can flip and rotate the latent noise image :o
@theally I see. I got some others for you. hehe
InvokeAI is also designing node based ui panel. controlNet now is added into diffuser, so InVokeAI will get that too.
Oh, good to know! Thanks! It's the future - being able to chain nodes together is mind-blowing!
As I checked today, invokeAI is almosted finished their node process. They are merging UI API now. Which basically is the last part of this job for verion1.0
So it's like chaiNNer and Cascade fused together with Stable Diffusion stuff? Sounds neat. :D
@theally I made a HR fix workspace you're welcome to share in your pack if you'd like: https://civitai.com/models/20215/comfyui-hr-fix-workspace
I think I messed up putting in right category anyway. Hard to find.
I was working on a hires fix flow! Thanks! This is interesting. I have been running out of vram trying to execute this one though, not sure why!
@WAS Yes, when I get to the VAEEncode step I bomb out, out of VRAM. I have 10GB. How much do you have?
@theally I have 24gb on the 4090. Occasionally I OOM with just 2mb over, but only when I leave ComfyUI open and walk away for awhile. Not sure what that's about.
I did release a v2 which has a downscale after AI upscale. First version just upscales with upscale model, which is often 4x these days, so the final image trying to be encoded and diffused is 2048x3072 :O Didn't think of how that would impact people lol
@theally Also not entirely sure, but I think ComfyUI may just be loading up all these models with no management, where as WebUI will switch to them, and cache out non-diffusion models out to CPU (RAM), like BLIP model, upscale models, etc.
Additionally, if ComfyUI adds a upscale model scale factor, that would cut out the need for a post resolution target, as can just upscale 2x instead of 4x. Really anything, like 1.5x which I've used on 8gb cards.
find this ui, it is amazing, more customizable, mo fun, expect have more feature. Follow~~~
Alpha-based denoising (allow me to give it some fancy name 😂).
A denoising node that denoises different parts of the image at different strengths based on a grayscale image mask provided to it (The Alpha Map).
We should be able to set the maximum and minimum denoising strengths that are going to be mapped to the brightness level data of the pixels of the mask (like 0xffffff being denoising strength x and 0x00000 denoising strength of y)
We should also have the ability to check an invert option for the mask.
Difference map generator. takes two or more images as input and generates a grayscale image with the parts most different in the images being the brightest
Hi, I haven't found a better place to ask this, so I'll just ask here. Is there other way to gain access to your tutorials? Paypal won't let me subscribe on patreon, and I'm very interested in your contents. THX.
Hey! I'm afraid not at the moment. I am exploring other platforms though, so hopefully in future I'll have a better place to present my content! Thanks! xo
Gradient node.
A node that generates complex gradients. these can then be fed to Controlnet to control the lighting of a scene
Blending node. (this is so basic maybe there is one. not that I know of)
A node allowing us to combine images with different blending modes such as lighten and multiply to name a few.
Image queue (or Image sequence) node.
A node that allows us to select a folder with multiple images to add as an image input. the difference with the default image input is that we can select based on what factors the next image should be provided and when
Take a look at ainodes-engine, which lets you create the entire SD process in nodes. Maybe you're able to join forces :)
the future is coming!
ComfyUI is promising! Extra node are welcome ! Thanks !
matte!