I am developing a Java / Processing library to add glitches to images.

This works by first compressing the image, using the jpeg algorithm, then flipping some bits and bytes around, and finally read it again. Because of the singularity of the jpeg compression, the file is still readable, but the image looks destroyed.

Currently I try to speed up the whole process to do live-image glitching (e.g. as Syphon Pipe). The current version on github is not the fastest, but it works if you downscale the image first.

Github: https://github.com/cansik/processing-imageglitch