Node packVerifiedExternal

RES4LYF

Sourcegithub.com/ClownsharkBatwing/RES4LYF

Maintainer@ClownsharkBatwingCgrade

Package profile

Status: Verified
Publisher: @ClownsharkBatwing
Source: GitHub
Stars: 1.3K

README

SUPERIOR SAMPLING WITH RES4LYF: THE POWER OF BONGMATH

RES_3M vs. Uni-PC (WAN). Typically only 20 steps are needed with RES samplers. Far more are needed with Uni-PC and other common samplers, and they never reach the same level of quality.

res_3m_vs_unipc_1 res_3m_vs_unipc_2

INSTALLATION

If you are using a venv, you will need to first run from within your ComfyUI folder (that contains your "venv" folder):

Linux:

source venv/bin/activate

Windows:

venv\Scripts\activate

Then, "cd" into your "custom_nodes" folder and run the following commands:

git clone https://github.com/ClownsharkBatwing/RES4LYF/

cd RES4LYF

If you are using a venv, run these commands:

pip install -r requirements.txt

Alternatively, if you are using the portable version of ComfyUI you will need to replace "pip" with the path to your embedded pip executable. For example, on Windows:

X:\path\to\your\comfy_portable_folder\python_embedded\Scripts\pip.exe install -r requirements.txt

IMPORTANT UPDATE INFO

The previous versions will remain available but with "Legacy" prepended to their names.

If you wish to use the sampler menu shown below, you will need to install https://github.com/rgthree/rgthree-comfy (which I highly recommend you have regardless).

image

If these menus do not show up after restarting ComfyUI and refreshing the page (hit F5, not just "r") verify that these menus are enabled in the rgthree settings (click the gear in the bottom left of ComfyUI, select rgthree, and ensure "Auto Nest Subdirectories" is checked):

image

NEW VERSION DOCUMENTATION

I have prepared a detailed explanation of many of the concepts of sampling with exmaples in this workflow. There's also many tips, explanations of parameters, and all of the most important nodes are laid out for you to see. Some new workflow-enhancing tricks like "chainsamplers" are demonstrated, and regional AND temporal prompting are explained (supporting Flux, HiDream, SD3.5, AuraFlow, and WAN - you can even change the conditioning on a frame-by-frame basis!).

[example_workflows/intro to clownsampling.json ](https://github.com/ClownsharkBatwing/RES4LYF/blob/main/example_workflows/intro%20to%20clownsampling.json)

intro to clownsampling

STYLE TRANSFER

Supported models: HiDream, Flux, Chroma, AuraFlow, SD1.5, SDXL, SD3.5, Stable Cascade, LTXV, and WAN. Also supported: Stable Cascade (and UltraPixel) which has an excellent understanding of style (https://github.com/ClownsharkBatwing/UltraCascade).

Currently, best results are with HiDream or Chroma, or Flux with a style lora (Flux Dev is very lacking with style knowledge). Include some mention of the style you wish to use in the prompt. (Try with the guide off to confirm the prompt is not doing the heavy lifting!)

image

For example, the prompt for the below was simply "a gritty illustration of a japanese woman with traditional hair in traditional clothes". Mostly you just need to make clear whether it's supposed to be a photo or an illustration, etc. so that the conditioning isn't fighting the style guide (every model has its inherent biases).

image

COMPOSITION GUIDE; OUTPUT; STYLE GUIDE

style example

![image](https://github.com/use

Nodes in this pack

100 nodes