Comments on: Roll Over, Delaunay: Voronoi Library Goes Open-Source http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/ Thu, 05 Jan 2012 17:15:00 +0000 hourly 1 http://wordpress.org/?v=3.0 By: Anonymous http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-183892 Anonymous Tue, 15 Nov 2011 04:41:00 +0000 http://nodename.com/blog/?p=377#comment-183892 Thanks for the link! Great to see such creative things being built with Voronoi! Thanks for the link! Great to see such creative things being built with Voronoi!

]]> By: Marco Quaggiotto http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-183891 Marco Quaggiotto Mon, 14 Nov 2011 18:52:00 +0000 http://nodename.com/blog/?p=377#comment-183891 It's been a while since you released this, but I wanted to let you know that it's still appreciated and used.. Here's your third (official) user: http://www.datainterfaces.org/2011/11/twitter-orographies/ Thanks! It’s been a while since you released this, but I wanted to let you know that it’s still appreciated and used..
Here’s your third (official) user: http://www.datainterfaces.org/2011/11/twitter-orographies/
Thanks!

]]> By: Quick Sketch – Spanning Tree - Ross Kettle's Blog http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-183879 Quick Sketch – Spanning Tree - Ross Kettle's Blog Sat, 16 Jul 2011 03:20:53 +0000 http://nodename.com/blog/?p=377#comment-183879 [...] old sketch from a while back. This made use of nodename’s excellent AS3 Delaunay library. It is the spanning tree from a grid of points that was generated as a regular grid with noise added [...] [...] old sketch from a while back. This made use of nodename’s excellent AS3 Delaunay library. It is the spanning tree from a grid of points that was generated as a regular grid with noise added [...]

]]> By: alan http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-181666 alan Wed, 09 Mar 2011 07:45:38 +0000 http://nodename.com/blog/?p=377#comment-181666 Hi Amit, I would certainly say the incremental algorithm is simpler. Also I think it's more productive in terms of where it can lead you (into topology-based improvements that can handle many many points, for example, leaving numerical stability problems behind, or so I hear) than the rather confusing sweepline - beachfront - parabolas Fortune algorithm, which for me was a bit of a dead end. Fortune claimed that generalizing his algorithm to the weighted version was straightforward but I haven't grasped exactly how! Check out this <a href="http://www.anderswallin.net/2011/02/vd-algorithm-animation-zoomed-out/" rel="nofollow">animation</a> of the incremental algorithm. Hi Amit,
I would certainly say the incremental algorithm is simpler. Also I think it’s more productive in terms of where it can lead you (into topology-based improvements that can handle many many points, for example, leaving numerical stability problems behind, or so I hear) than the rather confusing sweepline – beachfront – parabolas Fortune algorithm, which for me was a bit of a dead end. Fortune claimed that generalizing his algorithm to the weighted version was straightforward but I haven’t grasped exactly how! Check out this animation of the incremental algorithm.

]]> By: Amit Patel http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-181409 Amit Patel Sun, 06 Mar 2011 04:52:15 +0000 http://nodename.com/blog/?p=377#comment-181409 Alan, is the incremental algorithm simpler than Fortune's algorithm? I've been considering porting my map generation project to C++ so that it can run on a server, and haven't decided how I want to do this (or even whether I should, given all the other projects I want to work on). Alan, is the incremental algorithm simpler than Fortune’s algorithm? I’ve been considering porting my map generation project to C++ so that it can run on a server, and haven’t decided how I want to do this (or even whether I should, given all the other projects I want to work on).

]]> By: alan http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-180827 alan Sun, 27 Feb 2011 06:36:19 +0000 http://nodename.com/blog/?p=377#comment-180827 Hi Darrell, Yes, the clipping to the bounding rect was a pain, and I'm not that thrilled with my code there. I had started to rework it using more general polygon clipping as you mention, but then I got into implementing the incremental Voronoi algorithm, which handles changes to points in a localized way & therefore avoids recalculating the entire diagram, and also starts with three seed points outside the clipping bounds & therefore makes the clipping simpler -- much like what you describe! But that was several distractions (and IDE evaluations) ago, and it's not moving forward at the moment... Hi Darrell,
Yes, the clipping to the bounding rect was a pain, and I’m not that thrilled with my code there. I had started to rework it using more general polygon clipping as you mention, but then I got into implementing the incremental Voronoi algorithm, which handles changes to points in a localized way & therefore avoids recalculating the entire diagram, and also starts with three seed points outside the clipping bounds & therefore makes the clipping simpler — much like what you describe! But that was several distractions (and IDE evaluations) ago, and it’s not moving forward at the moment…

]]> By: Darrell Plank http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-180793 Darrell Plank Sat, 26 Feb 2011 17:20:05 +0000 http://nodename.com/blog/?p=377#comment-180793 Okay, I did actually take time to look through your code and it looks like you had to stand on your head a bit to take care of this also. I did happen to think of a cheap workaround kind of. I've already got a routine which clips convex polys to convex polys so the only cells I have any problem with are the infinite ones and they only occur only on points of the convex hull. If I put four points at the corners or diagonally out some distance from the corners, they would be the only ones on the convex hull so none of the sites within my rectangle would be infinite. That would solve my problem, but I think it's worth pursuing the "correct" solution here - esp if I put this in public domain which I plan on doing. BTW, now that I've looked at the code, nice job! It's been a long time since I've programmed in ActionScript - this brought back memories. Okay, I did actually take time to look through your code and it looks like you had to stand on your head a bit to take care of this also. I did happen to think of a cheap workaround kind of. I’ve already got a routine which clips convex polys to convex polys so the only cells I have any problem with are the infinite ones and they only occur only on points of the convex hull. If I put four points at the corners or diagonally out some distance from the corners, they would be the only ones on the convex hull so none of the sites within my rectangle would be infinite. That would solve my problem, but I think it’s worth pursuing the “correct” solution here – esp if I put this in public domain which I plan on doing.

BTW, now that I’ve looked at the code, nice job! It’s been a long time since I’ve programmed in ActionScript – this brought back memories.

]]> By: Darrell Plank http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-180760 Darrell Plank Sat, 26 Feb 2011 11:32:48 +0000 http://nodename.com/blog/?p=377#comment-180760 Hi- I've done a Fortune algorithm in C# and it's working fine, but I'm messing around a bit with trying to properly clip the infinite cells to a box. This is mainly so I can do Lloyd relaxation. I think you return your cells clipped so you must have done this already and I'm wondering if you ran into the same problems I'm running into. Without going into a lot of detail, after dealing with the potential of 1 or possibly 2 doubly infinite lines surrounding your cell due to collinear input points, rays that may originate outside the box and intersect 0, 1 (at one of the box corners) or 2 times or may originate inside the box in which case life becomes a bit easier - anyway, lots of little nitpicky cases. I'm wondering if you found some elegant way to take care of all these cases, or whether you ignored several of them, or whether you handled all those little nitpicky cases like I find myself doing. I can handle them, but it feels like I'm spending an incredibly amount of time and effort on some very corner cases. I suppose I should take a look at your code, but I've got my head buried in my own right now. It seems very quick from the map demo I looked at. Nice job! Hi-
I’ve done a Fortune algorithm in C# and it’s working fine, but I’m messing around a bit with trying to properly clip the infinite cells to a box. This is mainly so I can do Lloyd relaxation. I think you return your cells clipped so you must have done this already and I’m wondering if you ran into the same problems I’m running into. Without going into a lot of detail, after dealing with the potential of 1 or possibly 2 doubly infinite lines surrounding your cell due to collinear input points, rays that may originate outside the box and intersect 0, 1 (at one of the box corners) or 2 times or may originate inside the box in which case life becomes a bit easier – anyway, lots of little nitpicky cases. I’m wondering if you found some elegant way to take care of all these cases, or whether you ignored several of them, or whether you handled all those little nitpicky cases like I find myself doing. I can handle them, but it feels like I’m spending an incredibly amount of time and effort on some very corner cases.

I suppose I should take a look at your code, but I’ve got my head buried in my own right now. It seems very quick from the map demo I looked at. Nice job!

]]> By: basics in generative art 6 : TRIANGLE « HIDIHO! http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-175630 basics in generative art 6 : TRIANGLE « HIDIHO! Wed, 29 Dec 2010 15:47:31 +0000 http://nodename.com/blog/?p=377#comment-175630 [...] words about Alan shaw: “he [...] [...] words about Alan shaw: “he [...]

]]> By: jordan http://nodename.com/blog/2009/12/03/roll-over-delaunay-voronoi-library-goes-open-source/comment-page-1/#comment-170026 jordan Sun, 22 Aug 2010 03:17:40 +0000 http://nodename.com/blog/?p=377#comment-170026 thanks for sharing this library ... was having fun playing with the toy so i'm looking forward to using this for some VJ tricks! thanks for sharing this library … was having fun playing with the toy so i’m looking forward to using this for some VJ tricks!

]]>