Since I would be using said cell-shader on entire images or maybe rectangular selections only, this method could help me out, right? Other than the fact that the dll will run more than once, slowing the entire process down for no gain at all except for removing the occasional mistake, would there be many drawbacks, for my purposes, to this method? I do understand that, since PDN seems to be breaking a blank canvas with a line drawn down it into about 10 separate ROI's, the runtime should be around 10 times longer...
Or maybe I can trash my crappy edge-detection and go for a more accurate built-in edge-detection (is it emboss, or is it actually called edge detection... eh, cant remember) and pull out the differences between that and the original image...
Dont have PDN with me at the moment, but maybe some of what I have in my head...
I did see the tutorial page on these, but just sketching the concept. Please mind the obviously fake code written in 2 minutes :
dst = someFunctionThatWillEdgeDetect(src);
that built-in for loop on the pixels{
if(CurrentPixel != dst[x,y]){ makePixelBlack }
}
Just showing the concept there.