ComfyUI FRED Nodes v2
FRED's enhanced custom nodes for ComfyUI
Nodes (22)
Crop anything to a native SDXL ratio without cutting your subject off
Split an image into tiles from a mask or BBOX, overlap included
A LoRA loader with per-slot on/off/auto switches — the 'auto' is the point
Ten LoRAs, one loader, and it can switch them on from your prompt text
The checkpoint loader that remembers its own name
Crop a face out of an image, with margins you control
A stopwatch for your workflow, hung off the last node
Read a KSampler's settings without touching it — then feed them forward
Pull the applied LoRA names and weights out of any LoRA stacker
An image loader with folder browsing, AVIF support, and metadata passthrough
BRISQUE, blur, SNR, compression
A save node that names files from your actual model, seed, and LoRAs
A real parameters input and unlimited LoRA info
Auto-sharpen until the image stops being blurry, no manual guessing
Paste a processed crop back into its original image, seams hidden
Simulate JPEG artifacts on your images, useful for training and testing
One master slider that drives five parameters at once
Write text to a file from your workflow — append, overwrite, or never clobber
One panel that hands a KSampler everything it needs — noise, sigmas, sampler object included
Turn a tag list into XMP metadata you can embed for Lightroom and Bridge
The FRED UNET loader that hands you the filename
FRED's dynamic wildcard concat, block by block
👑 ComfyUI-FRED-Nodes v2
Custom ComfyUI nodes by Fred — all-in-one utilities and quality-of-life tools for image workflows.
This v2 repo is a clean rewrite of ComfyUI-FRED-Nodes, with improved nodes, bug fixes, and simplified structure.
✨ Features
- Enhanced image saving with metadata + grid support
- Smart image cropping/tiling with ratio/mask awareness
- Integrated quality metrics (BRISQUE, blur, SNR, compression)
- Simplified parameters panel for samplers/schedulers
- Utility nodes for text metadata (XMP) and wildcard concatenation
- Face detection and cropping
- Robust image loading with caching, indexing, and previews
🧩 Nodes Overview
📏 FRED_AutoCropImage_Native_Ratio
Automatically crop and resize images to match Stable Diffusion aspect ratios, or custom sizes. Supports mask preservation and overlay preview.
Key features:
Auto-find closest SDXL ratio or use custom width/height
Mask-preserve mode ensures subject is not cut off
Cropping from center or offset (X/Y %)
Optional resize with upscale/downscale modes
Multiple-of rounding (2, 4, 8, …)
Preview with mask overlay
🪟 FRED_AutoImageTile_from_Mask
Split an image into tiles based on a mask or bounding box. Useful for processing specific regions with controlled overlap.
Key features:
Auto grid from mask/BBOX with confidence adjustment
Manual rows/cols if desired
Overlap in X/Y
Grid preview overlay (bbox + lines)
📊 FRED_ImageQualityInspector
Evaluate image quality using multiple metrics, optionally within a mask.
Metrics:
BRISQUE (no-reference perceptual quality)
Blur (Laplacian variance)
SNR (signal-to-noise ratio, dB)
Compression proxy (raw vs JPEG@95 size)
Modes:
Raw metrics (real units, e.g. dB, Lap variance)
Normalized 0–100 scores (100 = best)
💾 FRED_Image_Saver
Advanced image saver with dynamic naming tokens and metadata embedding.
Key features:
Save single images, grid images, or both
Filename/path tokens: %date, %time, %model_name, %seed, %lora_name_X, …
Metadata embedding:
Automatic1111 “parameters” text
Prompts, workflow JSON, ComfyUI version
Optional JSON sidecar files
Grid saving:
Auto rows/cols up to max
Optional row/column labels
Metadata includes grid info
Outputs last saved image and grid paths.
🖼️ FRED_LoadImage_V8
Robust image loader with indexing, preview, and error logging.
Key features:
Modes: no_folder (direct image) or image_from_folder
Index control: fixed / increment / decrement / randomize
Skip invalid files, with persistent error log (~/.fred_logs)
Cache of folder index for performance
Optional preview images written to temp folder
Outputs include folder path, filename text, counts, skipped report
🙂 FRED_CropFace
Detect and crop faces using RetinaFace (facexlib).
Key features:
Face detection (multi-face; choose by ID)
Expand bounding box with margins
Preview overlay with detected faces
Outputs cropped face, preview, bbox, pixel ratios
⚙️ FRED_Simplified_Parameters_Panel
Expose a compact UI panel for common sampling parameters.
Outputs:
scale, denoise, guidance, steps
proper NOISE object
sampler/scheduler objects + names
sigmas tensor
Great for quick prototyping or simplified workflows.
📝 FRED_Text_to_XMP
Generate and attach XMP metadata to images (or sidecar .xmp).
Key features:
Embed prompts, negatives, parameters into standard XMP fields
Keywords → dc:subject (Lightroom/Bridge searchable)
Ratings, title, description, creator fields
Modes: embed into PNG/JPEG/TIFF or sidecar only
Token support: %date, %time, %model_name, %seed, …
🎲 FRED_WildcardConcat_Dynamic
Dynamic wildcard concatenation for prompt building.
Key features:
Add/remove blocks dynamically
Each block: toggle, wildcard file, line mode (random/fixed), weight
Prefix and suffix support
Custom string delimiter between blocks
Output = concatenated string with tokens expanded
📖 Usage Tips
Each node outputs a help string for quick reference inside ComfyUI.
For tutorials, check the Wiki (todo: add link).
Add screenshots of each node UI for clarity.
🛠️ License
MIT License — free to use and modify. If you use these nodes in your workflows, a ⭐ on this repo is always appreciated.
📦 Installation
Clone or download this repo into your ComfyUI custom_nodes/ folder:
cd ComfyUI/custom_nodes
git clone https://github.com/Poukpalaova/ComfyUI-FRED-Nodes_v2.git