smZNodes

A selection of custom nodes for ComfyUI.

1. CLIP Text Encode++
2. Settings

CLIP Text Encode++

CLIP Text Encode++ can generate identical embeddings from stable-diffusion-webui for ComfyUI.

This means you can reproduce the same images generated from stable-diffusion-webui on ComfyUI.

Simple prompts generate identical images. More complex prompts with complex attention/emphasis/weighting may generate images with slight differences due to how ComfyUI denoises images. In that case, you can enable the option to use another denoiser with the Settings node.

Features

• Prompt editing
  • Alternating words
• Weight normalization
• Usage of BREAK and AND keywords
• Optional embedding: identifier

Installation

Three methods are available for installation:

1. Load via ComfyUI Manager
2. Clone the repository directly into the extensions directory.
3. Download the project manually.

Load via ComfyUI Manager

Install via ComfyUI Manager

Clone Repository

cd path/to/your/ComfyUI/custom_nodes
git clone https://github.com/shiimizu/ComfyUI_smZNodes.git

Download Manually

1. Download the project archive from here.
2. Extract the downloaded zip file.
3. Move the extracted files to path/to/your/ComfyUI/custom_nodes.
4. Restart ComfyUI

The folder structure should resemble: path/to/your/ComfyUI/custom_nodes/ComfyUI_smZNodes.

Update

To update the extension, update via ComfyUI Manager or pull the latest changes from the repository:

cd path/to/your/ComfyUI/custom_nodes/ComfyUI_smZNodes
git pull

Comparisons

These images can be dragged into ComfyUI to load their workflows. Each image is done using the Silicon29 (in SD v1.5) checkpoint with 18 steps using the Heun sampler.

stable-diffusion-webui	A1111 parser	Comfy parser

Image slider links:

• https://imgsli.com/MTkxMjE0
• https://imgsli.com/MTkxMjEy

Options

Name	Description
parser	The parser selected to parse prompts into tokens and then transformed (encoded) into embeddings. Taken from automatic.
mean_normalization	Whether to take the mean of your prompt weights. It's true by default on stable-diffusion-webui. This is implemented according to stable-diffusion-webui. (They say that it's probably not the correct way to take the mean.)
multi_conditioning	This is usually set to true for your positive prompt and false for your negative prompt. For each prompt, the list is obtained by splitting the prompt using the AND separator. See Compositional Visual Generation with Composable Diffusion Models
use_old_emphasis_implementation	Use old emphasis implementation. Can be useful to reproduce old seeds.

Important

You can right click the node to show/hide some of the widgets. E.g. the with_SDXL option.

Parser	Description
comfy	The default way ComfyUI handles everything
comfy++	Uses ComfyUI's parser but encodes tokens the way stable-diffusion-webui does, allowing to take the mean as they do.
A1111	The default parser used in stable-diffusion-webui
full	Same as A1111 but whitespaces and newlines are stripped
compel	Uses compel
fixed attention	Prompt is untampered with

Note
Every parser except comfy uses stable-diffusion-webui's encoding pipeline.

Warning
LoRA syntax (<lora:name:1.0>) is not suppprted.

Settings

Settings node workflow

The Settings node can be used to finetune results from CLIP Text Encode++. Some settings apply globally, or just during tokenization, or just for CFGDenoiser. The RNG setting applies globally.

This node can change whenever it is updated, so you may have to recreate the node to prevent issues. Hook it up before CLIP Text Encode++ nodes to apply any changes. Settings can be overridden by using another Settings node somewhere past a previous one. Right click the node for the Hide/show all descriptions menu option.

Tips to get reproducible results on both UIs

• Use the same seed, sampler settings, RNG (CPU or GPU), clip skip (CLIP Set Last Layer), etc.
• Ancestral samplers may not be deterministic.
• If you're using DDIM as your sampler, use the ddim_uniform scheduler.
• There are different unipc configurations. Adjust accordingly on both UIs.

FAQs

• How does this differ from ComfyUI_ADV_CLIP_emb?
  • In regards to stable-diffusion-webui:
    • Mine parses prompts using their parser.
    • Mine takes the mean exactly as they do. ComfyUI_ADV_CLIP_emb probably takes the correct mean but hey, this is for the purpose of reproducible images.
• Where can I learn more about how ComfyUI interprets weights?
  • https://comfyanonymous.github.io/ComfyUI_examples/faq/
  • https://blenderneko.github.io/ComfyUI-docs/Interface/Textprompts/