breadbox Posted June 5, 2011 Share Posted June 5, 2011 hello, I have been influenced by a technique called reaction diffusion. I made 2 examples of reaction diffusion with photoshop. this one with an added spiral - looks like a finger print! i know little of houdini Image OP's but it looks like they have a blur and a sharpen. and it can work to create images that are similar, however i don't see a "for loop" in the IMG like they have in the SOPs so i have to copy and paste manually each blur and sharpen. you need around 50 or so iterations to get some results like this. is there a way to loop through images. my next step will be to try to create geometry that's built off this method but i don't really know how I'm going to approach that yet. could just be interesting texture/ displacement maps. of course the effect looks really cool when the source image would evolve a bit those wormy lines will crawl. 2 Quote Link to comment Share on other sites More sharing options...
Macha Posted June 6, 2011 Share Posted June 6, 2011 I don't know if this helps but it will blur and sharpen volumes repeatedly. blursharpen.hip Quote Link to comment Share on other sites More sharing options...
Andz Posted June 6, 2011 Share Posted June 6, 2011 (edited) my next step will be to try to create geometry that's built off this method but i don't really know how I'm going to approach that yet. You can try the trace sop and/or an L-system. *edit ... or even a iso surface I guess. Edited June 6, 2011 by Andz Quote Link to comment Share on other sites More sharing options...
nlapinski Posted November 28, 2013 Share Posted November 28, 2013 Using a smooth sop will let you blur color values, you can then use a attribute composite to blend it with the original. Drop it in a solver sop and it will run every frame. Also attribute transfer can be used to blur when both inputs are attached to the same node. The color values can then be used for displacement or clipping of some geometry. I have attached a hip, its not really gray scott reaction diffusion I dont think, but it gives an aproximate look. Also using very large values in a vertex color pattern and running it through a smooth sop (smoothing color) will give the patterns simmilar to the ones posted. rd_test.hipnc 1 Quote Link to comment Share on other sites More sharing options...
nlapinski Posted October 10, 2015 Share Posted October 10, 2015 actual reaction diffusion (attached hip) - quite slow made it awhile ago never got around to optimizing it gsrd.hipnc 1 Quote Link to comment Share on other sites More sharing options...
Jason Posted October 13, 2015 Share Posted October 13, 2015 Nice! There is also this out there : https://www.orbolt.com/asset/animatrix::reactionDiffusion::1.00 Quote Link to comment Share on other sites More sharing options...
dan wills Posted December 6, 2015 Share Posted December 6, 2015 Hiyas, Reaction Diffusion is a super interesting technique, I've done a few things with it, here's some clips: I've attached my Gray-Scott Vop OTLs.. they were used in the first two clips above, and can be connected in a myriad of ways. No doubt they could also be adapted to different formulae and heaps of speedups without too much trouble Apologies if they're not as well structured or documented as they could be. These went public some time ago on the 'reaction-diffusion' google group (gollygang/ready). Which has some instructions if you're wondering how to use them: https://groups.google.com/forum/#!msg/reaction-diffusion/6nSGH3CPRZM/plXsCITLpvMJ If you are interested in reaction-diffusion, then I highly recommend checking out Ready!! : (pc/mac/linux/src) https://github.com/GollyGang/ready It's an awesome and fun playground for RD systems. I've made a (very basic so far) houdini plugin to capture Ready's simulation data into houdini as voxels. Source is already out there but hard to build so I've been working out build details to try to make it as simple as toggling a boolean cmake flag which should hopefully make it a lot easier for people to play with. It'll probably be announced in the release notes if/when it's accepted into the repo. Happy diffusing, in reaction, Dan reactionDiffusion.otl reactionDiffusionInit.otl 7 Quote Link to comment Share on other sites More sharing options...
mestela Posted December 6, 2015 Share Posted December 6, 2015 Awesome stuff Dan, thanks for sharing! Quote Link to comment Share on other sites More sharing options...
msyugo Posted December 7, 2015 Share Posted December 7, 2015 Amazing! Thanks for sharing!! Quote Link to comment Share on other sites More sharing options...
animatrix Posted December 7, 2015 Share Posted December 7, 2015 I also made a VEX implementation of this before. The speed is consistent across different contexts, i.e. COPs vs a slice of a volume with the same resolution. Some of the VOPs shown in the video: http://www.orbolt.com/asset/animatrix::reactionDiffusion::1.00http://www.orbolt.com/asset/animatrix::pointCloudLaplacian::1.00 There is also a higher level Reaction Diffusion Solver that uses all of these internally with the ability to override certain parameters using attributes or volumes with the same name (fr -> Feed Rate, kr -> Kill Rate, etc). It will be on Orbolt soon hopefully 2 Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.