Hermitnoise

A XOR Persistence of Vision Video Effect

A Python tool that applies the mesmerizing XOR persistence of vision effect to videos. Shapes are only visible while they're moving. Pause the video and they disappear into noise!

How It Works

This effect exploits how your brain perceives motion:

1. Edge Detection: Each frame is processed with Canny edge detection
2. XOR Drawing: Edges are XOR'd onto a persistent canvas
3. Pixel Toggle: When a pixel is drawn where one exists, it toggles off
4. Noise Accumulation: Static areas become random noise patterns
5. Motion Visibility: Moving edges remain visible due to persistence of vision

The result: you can clearly see shapes moving, but pause the video and there's no visible shape, you'll just see random dots!

Installation

pip install -r requirements.txt

Usage

Basic usage:

python xor_vision.py input_video.mp4 output.mp4

Remove audio:

python xor_vision.py input_video.mp4 output.mp4 --no-audio

Customize the pixel effect:

python xor_vision.py input_video.mp4 output.mp4 --resolution 80x60 --line-width 2

White edges on black background:

python xor_vision.py input_video.mp4 output.mp4 --invert

Options

Option	Default	Description
--keep-audio	✓	Keep original audio track
--no-audio		Remove audio from output
--resolution WxH	160x120	Processing resolution (lower = bigger pixels)
--line-width 1-5	1	Edge line thickness
--max-duration N	30	Max video length in seconds
--fps N	30	Output frame rate
--invert		White edges on black background

Tips

• Best results: Videos with clear movement and distinct edges
• Lower resolution: Creates bigger, more retro pixels
• Higher line-width: Makes edges more visible but adds more noise
• Invert mode: More dramatic effect, closer to the original demo