This tool takes a folder of sequential PNG frames (e.g. extracted from a video), detects per-pixel movement between consecutive frames, and renders a “motion ghosting” visualization as output frames you can stitch back into a video.

• Movement pixels reset their “age” to 0.
• Non-moving pixels accumulate age each frame.
• Output color is based on age (a heatmap-like HSV mapping), producing a ghosting trail effect over time.

• Python 3.9+ recommended
• Dependencies:
  • numpy
  • Pillow (PIL)

Optional (recommended for video workflows):

• ffmpeg available on your PATH

Install Python deps:

python -m pip install numpy pillow

Create a folder

mkdir frames

Split your input video into frames

ffmpeg -i input.mp4 -vsync 0 frames/frame_%06d.png

Process the frames

python motion_ghosting.py --input frames --output output --fps 60 --fade-seconds 1 --pad-digits 6

Merge processed frames back into a video

ffmpeg -framerate 60 -i output/output_%06d.png -c:v libx264 -pix_fmt yuv420p output.mp4