Meet jit.repos Repos stands for reposition, an efficient way to mangle images. We can use it to turn pictures inside out, or build fun house mirrors, strange lenses, and kaleidoscopes. The principle behind jit.repos is simple. It holds a control matrix with 2 planes and the dimensions of the image to process. The values in the control matrix are the source for each pixel in the final output. For example, if the control matrix contains [0 0] at position [160 120], pixel 160 120 in the output will contain the colors found at [0 0] of the input. This can produce an endless variety of strange images. The secret to using jit.repos is in making control matrices. This is not usually a real time process, so it is best to build them ahead of time and save them to disk. Once they are made up, it is easy to use jit.xfade and jit.matrixset to move them around and animate the process. The first control matrix to make does nothing at all. Each cell contains its own cell position, so jit.repos will just pass the image through. (I’ll call it the pass.jxf matrix) The method used to derive it is unnecessarily complex, but we only have to do it once, and this patch will become our control matrix factory.