Native signal processing for ComfyUI

Break the signal.
Keep the frame.

Ten focused nodes carry image tensors through a native Rust NTSC pipeline. Build convincing tape wear, CRT softness, chroma drift, scanlines, and signal noise without leaving your ComfyUI graph.

Install from source View code

10: nodes
14: presets
Rust: native core

Matte television, blue sphere, and red paper ribbon before processing

The same still life after the VHS Classic preset

Real native output

One frame. Three damaged signals.

Every result below came from the compiled extension and the presets in this repository.

VHS Classic output with tape edge distortion and chroma bleed — **VHS Classic**Tape edge shift · color bleed

CRT TV output with softened horizontal detail and scanline behavior — **CRT TV**Soft luma · field texture

Security Camera output with heavy signal noise and reduced clarity — **Security Cam**Noise floor · surveillance wear

Choose the right amount of control

Start simple. Split the signal when needed.

Simple

Pick a preset, set intensity and saturation, and keep moving.

Intermediate

Add scaling, field mode, and deterministic seed controls.

Advanced

Expose the full signal-processing surface in one node.

04—10

Chainable families

Signal · Scanlines · Artifacts · Noise · VHS · Global · Scale

The working path

Tensor in. Analog damage. Tensor out.

01ComfyUIBHWC float tensor
02PyO3validated NumPy boundary
03ntsc-rsnative YIQ signal work
04OutputRGB tensor, same graph

Python 3.12+ · Rust 1.89+

Build for the Python that runs ComfyUI.

The installer targets an explicit interpreter and never copies into a machine-specific folder unless you ask it to install.

git clone https://github.com/gvastethecreator/comfyui-ntsc-rs.git
cd comfyui-ntsc-rs
python -m pip install -r requirements.txt
python build_node.py --install

Break the signal.Keep the frame.