The problem with the new conversion code has been fixed and the output is looking very nice. If I could have found it before spending several days porting the original code it would have been even better.
Finished the code to apply 3×3 convolution matrices to the output image, and replaced the old sharpening code with a matrix. Eventually the user will be able to enter a custom matrix to be used (i.e. for blurring/alternate sharpening/edge detection/mean removal etc). I’ve also got an unsharpening mask working, but it’s a little disappointing (2xoriginal – blurred, right?).
Aside from that, there’s lots of little things done that are hardly worth mentioning (drag drop image onto the form/executable, resizeable form so you can go full screen or use it in 800×600 mode, validation…)