Extensions/ComfyUI-Mutantwork
ComfyUI Extension

ComfyUI-Mutantwork

Prompt optimization, local AI forensics, and image provenance for ComfyUI

By brerereton-beep·Created 3 months ago·Updated 2 months ago· 1
brerereton-beep/ComfyUI-Mutantwork
Nodes4
On cloudLocal install
CategoryMutantwork
Stars1
Updated2 months ago
Readme

🧬 ComfyUI-Mutantwork — The Mutant Power Pack

A professional ComfyUI custom node suite for prompt optimization, local AI forensics, image provenance, and C2PA content credentials.

Built by Mutantwork — AI tools built different.


Nodes

🧬 Mutant Prompt Optimizer

Implements the Word Weight methodology to combat Semantic Saturation.

Most "quality" tokens (masterpiece, 4k, highly detailed) have been seen so often in training data that the model partially ignores them. They consume token budget without activating specific latent directions.

What it does:

  • De-noises your prompt by stripping saturated tokens that dilute latent space
  • Injects weighted replacements — maps weak generic phrases to high-density, precisely targeted equivalents
  • Example: good lighting(cinematic rim lighting:1.3)

Controls: | Input | Description | |-------|-------------| | prompt | Your raw prompt | | enable_denoise | Toggle stripping of inflated tokens | | enable_weighting | Toggle word weight injection | | weight_scale | Global multiplier on all injected weights (0.5–1.5) |

Outputs: optimized_prompt (ready for CLIP Text Encode) + optimization_report


🔬 Mutant Forensic Lab

Local AI image detection. No API keys. No cloud. No cost.

Runs three forensic pillars locally:

| Pillar | Method | What It Detects | |--------|--------|-----------------| | A — Metadata | PNG/EXIF text chunk extraction | ComfyUI workflow JSON, A1111 parameters, Midjourney tags, camera EXIF | | B — FFT Analysis | 2D Fast Fourier Transform | Latent grid artifacts at 8/16/32px intervals from VAE decoder upsampling | | C — ELA | Error Level Analysis | Inconsistent compression cycles indicating composited or edited images |

Outputs: image (pass-through) + forensic_report (scored text report with verdict)


⚗️ Mutant Signature

Injects an invisible provenance key into your PNG before saving.

Bakes a PNG tEXt chunk into every output:

Key:   Mutantwork
Value: Optimized by Prompt Power | mutantwork.com

Images signed by this node can be instantly verified at mutantwork.com/verify.


🔏 Mutant Content Credentials

Embeds a cryptographically signed C2PA manifest into your image.

C2PA (Coalition for Content Provenance and Authenticity) is the open standard backed by Adobe, Microsoft, Google, the BBC, and major media organisations for verifiable media provenance.

What it does:

  • Signs your image with a cryptographic manifest — tamper-evident and verifiable
  • Embeds a "Do Not Train" assertion (AI training mining: notAllowed)
  • Records creator name, AI model, and timestamp
  • Signed images are verified at mutantwork.com/verify — drop in your image and the C2PA badge appears instantly

Ships with bundled test certs — zero setup required. Optionally accepts your own PEM certificate for production signing.

Controls: | Input | Description | |-------|-------------| | creator_name | Your name or brand (default: Mutantwork) | | ai_model | The model used to generate the image | | do_not_train | Embeds C2PA "notAllowed" training assertion when enabled | | filename_prefix | Output filename prefix | | custom_cert_pem | Optional: path to your own PEM certificate | | custom_key_pem | Optional: path to your own PEM private key |

Requires: pip install c2pa-python cryptography


Installation

Option 1 — Manual

  1. Clone into your ComfyUI/custom_nodes/ folder:
cd ComfyUI/custom_nodes
git clone https://github.com/brerereton-beep/ComfyUI-Mutantwork.git
  1. Install dependencies:
pip install -r ComfyUI-Mutantwork/requirements.txt
  1. Restart ComfyUI. Nodes appear under the "Mutantwork" category.

Option 2 — ComfyUI Manager

Search for ComfyUI-Mutantwork in ComfyUI Manager → Install.


Example Workflow

[Load Image] ──→ [Mutant Forensic Lab] ──→ [Mutant Signature] ──→ [Mutant Content Credentials] ──→ [Save Image]
                         ↓                          ↓                           ↓
                  [Display Any]              [Display Any]              [Display Any]
               (forensic_report)           (signature_log)           (credentials_log)

[Mutant Prompt Optimizer] ──→ [CLIP Text Encode] ──→ [KSampler...]

Requirements

numpy>=1.24.0
Pillow>=10.0.0
scipy>=1.11.0
opencv-python>=4.8.0
c2pa-python>=0.32.0

Part of the Mutantwork Suite

| Tool | Platform | Link | |------|----------|------| | Prompt Power Pro | Chrome Extension | Gumroad | | Mutant Verify | Web | mutantwork.com/verify | | NI Trade Guard Pro | Web | ni-trade-guard.streamlit.app | | ComfyUI-Mutantwork | ComfyUI | This repo |


License

MIT — free to use, modify, and distribute.


Built by @Rickbot — mutantwork.com