ComfyUI-louis-use

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