[image processing] - A New Kind of Science: The NKS Forum

A New Kind of Science: The NKS Forum

Pages:1


image processing

(Click here to view the original thread with full colors/images)


Posted by: Philip Ronald Dutton

I suspect that most of the basic filters in the field of image processing are "based upon simple 'rules.' " Maybe not quite literally as in the NKS sense but still very basic processes such as what is required to produce an inverted (negative) image. I suggest image processing experiments in which image processing filters are applied under certain rules to a basic image. It would be interesting to see the evolution of such graphics. Perhaps I should post an example. I will try to do so as soon as possible.

(hint: layers and layer transfer modes may come in handy if photoshop is your artistic weapon of choice)


Posted by: dan wills

you're absolutely right.. there's some wicked effects that can happen when repeating simple image filters, I've found lots of interesting behaviours using sharpen, blur, edge-detection, median and dilate/erode filtering.

I find that sharpen and blur are particularly simple and cool when repeated in a sequence, depending on the relative strengths of each filter it's sometimes neccesary to balance them by doing mor blurs than sharpens or vice versa. I like to think of repeating sharpen and blur as a kind of rough analogue to reaction and diffusion, although I have never seen the spiral forms, I think mainly because the graphics apps I've used have the same size kernel in both their sharpen and blur filters.

The patterns you see are often quite biological - very Turing-esque with smooth blobs membranes and dots and such. Too much sharpen and you end up at a pixel-wide labyrinth, too little and it all fuzzes out. But all rather fascinating :D

The app I used mostly for this was the image viewer Irfanview. This was because the filters can all be done repeptetively with the keyboard. I have also written some macros for Shake that build long sequences of filter nodes, which give a similar effect.

I've also started experimenting with some of these ideas in a Java Applet.. which is better because often with image programs you either quickly lose all color, or get separated evolution in R G and B, which isn't that interesting.. in my applet I'm using an array of color as part of the processing to try to guide the colors into more interesting basins... sometimes works ;)

Good point man! :D NKS is so often right there underneath the tools ppl use every day! ;).




Forum Sponsored by Wolfram Research

© 2004-2008 Wolfram Research, Inc. | Powered by vBulletin 2.3.0 © 2000-2002 Jelsoft Enterprises, Ltd. | Disclaimer
vB Easy Archive Final - Created by Xenon and modified/released by SkuZZy from the Job Openings