Deblur (The Tool Jam 2)

The inner loop of the process essentially:

1. Guesses an un-blurred image (in the first iteration this guess can be anything, e.g. noise or the blurred image itself).
2. Blurs that guess using the specified blurring function.
3. Subtracts the blurred guess from the actual blurred image, to produce an 'error image' (containing both positive and negative values)
4. Blur the error image, and add it to the guess from step 1 (with a randomized & decreasing scale factor) to produce the next guess.
5. And continues iterating from there.

Anti-blur is used during step 2, when the guess is blurred. An anti-blur value of 100% means to blur it at full strength (i.e. full blur radius) and lower means to blur it less. So it affects the way that error propagates back to the guesses (in probably a pretty minor way).

Thanks!