juliovr / videomation_encoder

Videomation encoder

The video-information encoder converts any file to images (using every 3-bytes chunk as a pixel), and combines them into a video.

ffmpeg commands

Generate video from images

ffmpeg -y -framerate 30 -i image_0000.bmp -c:v libx264rgb -vf "fps=30" -crf 0 -preset veryslow -pix_fmt bgr24 output.mp4

-c:v libx264rgb: Uses the RGB color space for lossless encoding.
-crf 0: Ensures lossless quality.
-preset veryslow: Use the highest quality settings.
-pix_fmt bgr24: Matches the BMP format's pixel format.

Generate images from video

ffmpeg -i output.mp4 -vf "fps=30" -vsync 0 -pix_fmt bgr24 extracted_%04d.bmp

-vsync 0: Extract frames without duplicate or drop any frame.
-pix_fmt bgr24: Matches the BMP format's pixel format.

Compare images

Using compare from imagemagick: compare -metric RMSE image_0000.bmp extracted_0001.bmp difference.png

References:

• https://shotstack.io/learn/use-ffmpeg-to-convert-images-to-video/
• https://www.bannerbear.com/blog/how-to-extract-images-from-a-video-using-ffmpeg/
• ChatGPT (for the extra parameters to ensure the extracted frames matches the original images to construct the video).