Below are the 2 most recent journal entries recorded in Mark Pazolli's LiveJournal:

Sunday, March 5th, 2006
6:30 pm	Of strawberries and other sweet things There are some interesting projects in my life right now and they're all sort of coming together (I think). Often I have some good ideas or someone passes along an interesting tidbit but no-one else knows about them. I keep meaning to create a personal website but never do, so I figure a personal blog is the next best thing. This way people can stay in-tune with what I'm thinking and doing and ideally offer me some fresh perspectives on my thoughts or just flat out correct them when they're wrong. I was sent something really interesting today. It was a list of GIMP plug-ins that can make textures out of existing images. I was so delighted with these that I considered implementing one of the plug-ins then and there. Only to realize I'd need to extend Seashore's plug-in framework to do so (the plug-in I wanted to implement created a new image something not currently supported by Seashore plug-ins). Nevertheless, if the samples are anything to go by, what they do is really impressive. Take a look at this sample from the Texturize webpage: That may not seem impressive until you realize the borders of the original image do not simply reappear in the final product. By that I mean if the image was tiled it would look like this: Seashore, by the way, is an image editor I made loosely based upon the GIMP. This is really impressive and special thanks to "zipppy" who sent me the article. :-) There are other things in my engineering life to talk about - some related to Seashore, some not - so stay tuned amigos. (Comment on this)
Wednesday, February 15th, 2006
10:27 pm	On trees and such I have been scouring the 'net recently for insights into final year engineering because my feeling has been "I can't believe so many people make it through something this long and difficult." I had a minor breakthrough today. I have this binary decision tree and the path through the decision tree is based on the result of a lengthy calculation at each node. The logical approach seemed to be to do the calculation then move to the next node do the calculation there and so on. This created a long stream of calculations that would significantly delay the final output. My breakthrough was to realize that I could do all the calculations in parallel in a small subtree of the tree and, based on the result, move to the next subtree and do the same there. There would be lots of redundant calculations but since I'm targeting silicon not software this is okay. The diagram illustrates. I have more interesting things to say but this is what dominated my day today - so this is what I am writing about. (Comment on this)