Extensions/Louis Use
ComfyUI Extension

Louis Use

Practical ComfyUI utility nodes for folder I/O, image tools, color matching, seamless textures, prompt conversion, performance tracking, and optional QwenVL screening.

By LouisLU1997·Created 4 months ago·Updated 2 months ago· 0
LouisLU1997/ComfyUI-louis-use
Nodes20
On cloudLocal install
CategoryLouis_use, Louis_use/ideogram4
Stars0
Updated2 months ago

Nodes (20)

🗃️ Batch Image Saver

The batch saver that keeps your filenames — and your workflow, and your sidecars

Louis_use
🎨 Color Match

Fix the per-tile color drift that tiled upscaling always leaves behind

Louis_use
🎨 Color Palette Extractor

Get the five colors that actually make up an image, as an image you can use

Louis_use
📐 Divisible Crop

Crop to a VAE-friendly size before you sample

Louis_use
📂 Folder Image Loader

Point it at a folder of images and let ComfyUI process them one at a time

Louis_use
📄 Folder Text Loader

Load every prompt in a folder as a list — the lazy way to batch-iterate prompts

Louis_use
📝 Ideogram4 Text Encode

Feed Ideogram 4 its structured JSON prompt — with or without a local LLM doing the thinking

Louis_use/ideogram4
🖼️ Image Composite

Blend, dodge, and burn your way out of the 'paste on top' trap

Louis_use
↔️ Image Flip Horizontal

The mirror node — boring, but batch-safe and dependency-free

Louis_use
🔄 Image Invert

Black to white, white to black — a color invert that doesn't touch your alpha

Louis_use
🖼️ Image Pad Color

Add a solid-color frame without breaking your aspect ratio

Louis_use
🛡️ 版权筛选器

An automated 'don't ship the Mickey image' gate that re-rolls itself

Louis_use
✨ Reflection Extractor

Pull the specular highlights out of a render — the Screen-mode compositing secret

Louis_use
📐 Resolution Selector

Stop typing 1024×1024 — pick an aspect ratio and a megapixel budget instead

Louis_use
🧩 Seamless Tile Fixer

Make any image tileable in one pass — the 50% offset trick explained

Louis_use
📝 Show Text

See the text a node generated — and copy it without digging through the console

Louis_use
✂️ Smart Align & Crop

Normalize a folder of images to one resolution — reference image or 1080p cap

Louis_use
🚗 汽车深度图

A depth map of just the car — Depth Anything, plus a mask, in one pass

Louis_use
✂️ Text Segmenter

Type 'person. cat. red car.' — get a clean cutout of each one, automatically

Louis_use
📊 Performance Tracker

Generation time and peak VRAM, stamped into your PNG — with zero graph changes

Louis_use
Readme

Louis Use - ComfyUI Custom Nodes

A practical ComfyUI custom node pack for batch folder I/O, image utilities, color tools, seamless texture workflows, prompt conversion, performance tracking, and optional QwenVL-based image/IP screening.

Package folder name for manual installs: Louis_use

Features

  • Folder batch loading and saving with preview UI
  • Resolution selector with Chinese aspect-ratio labels
  • Smart crop, divisible crop, padding, compositing, flip and invert tools
  • Color palette extraction and color matching
  • Seamless tile fixer with 1x1, 1x2, 2x1, 2x2, 3x3 and 4x4 preview modes
  • Performance Tracker for generation time and VRAM metadata
  • Ideogram 4 structured prompt encoder with optional QwenVL semantic conversion
  • Optional text-driven segmentation/depth helpers
  • Optional QwenVL image copyright/IP guard (image -> image)

Installation

ComfyUI Manager

Search for Louis Use in ComfyUI Manager and install it.

Manual install

cd ComfyUI/custom_nodes
git clone https://github.com/LouisLU1997/ComfyUI-louis-use Louis_use

Restart ComfyUI after installation.

Dependencies

Most core nodes only need packages that ship with ComfyUI, such as torch, Pillow, numpy, and aiohttp.

For all optional nodes, install:

pip install -r requirements.txt

Optional heavy features may download additional models on first use:

  • Text Segmenter / Text Segmented Depth: Grounding DINO, BiRefNet, Depth-Anything related models
  • Ideogram4 Qwen mode / Copyright Guard: requires the separate ComfyUI-QwenVL custom node and a QwenVL checkpoint

Nodes

Folder I/O

| Node | Description | | --- | --- | | Folder Text Loader | Load .txt files from a folder as strings. | | Folder Image Loader | Load image batches from a folder with frontend thumbnail preview. | | Batch Image Saver | Save image batches, optional .txt/.json sidecars, overwrite toggle, metadata support. |

Resolution and Image Utilities

| Node | Description | | --- | --- | | Resolution Selector | Convert aspect-ratio labels such as 16:9 横屏 to width/height. | | Smart Align & Crop | Align and crop to a target resolution. | | Divisible Crop | Crop to dimensions divisible by 8/16/32 or custom divisor. | | Image Pad Color | Pad image edges with a chosen color. | | Image Composite | Composite foreground over background with common blend modes. | | Image Flip Horizontal | Batch-safe horizontal image flip. | | Image Invert | RGB color inversion. | | Reflection Extractor | Extract bright reflection/highlight regions. |

Color Tools

| Node | Description | | --- | --- | | Color Palette Extractor | Extract dominant colors and output a palette visualization. | | Color Match | Match image color statistics to a reference image. |

Seamless Texture

| Node | Description | | --- | --- | | Seamless Tile Fixer | Offset-blend image into a tileable texture and output a tiled preview. |

Preview modes include 1x1, 1x2, 2x1, 2x2, 3x3, and 4x4.

Text / AI Helpers

| Node | Description | | --- | --- | | Ideogram4 Text Encode | Convert descriptions into Ideogram 4 style structured JSON and CLIP conditioning. Optional QwenVL mode can analyze semantics and reference-image colors. | | Show Text | Display and persist string output directly on the node. | | Text Segmenter | Text-driven object segmentation using detection + matting models. | | Text Segmented Depth | Text-driven target depth-map extraction. | | Copyright Guard | QwenVL-based image -> image screening node. Blocks suspicious images and can cooperate with the frontend auto-requeue helper. |

Performance

| Node | Description | | --- | --- | | Performance Tracker | Image passthrough node that records generation time and peak VRAM, and injects metadata for save nodes. |

QwenVL Optional Features

Ideogram4 Text Encode Qwen mode and Copyright Guard require ComfyUI-QwenVL.

Install ComfyUI-QwenVL separately, then use a VRAM-friendly quantization option such as 4-bit (VRAM-friendly) if needed. These nodes degrade gracefully when QwenVL is not installed, but the Qwen-powered function itself will not run.

Publishing / Comfy Registry

This repository includes:

  • pyproject.toml with [tool.comfy] metadata
  • .github/workflows/publish.yml using Comfy-Org/publish-node-action

To publish to the Comfy Registry, add a GitHub secret named REGISTRY_ACCESS_TOKEN, then push to main/master or run the workflow manually.

File Structure

Louis_use/
├── __init__.py
├── nodes.py
├── ideogram4_text_encode.py
├── ip_copyright_guard.py
├── api.py
├── js/
│   ├── folder_io.js
│   ├── resolution_selector.js
│   ├── show_text.js
│   └── ip_copyright_guard.js
├── requirements.txt
└── pyproject.toml

License

MIT