kensonuken Posted January 28, 2009 Share Posted January 28, 2009 I would to create dense particle effects but they are hard to render .... like Im expecting high density particle count with motion blur.... how is it possible? Quote Link to comment Share on other sites More sharing options...
old school Posted January 28, 2009 Share Posted January 28, 2009 Not much info to start with. I assume your problem is with rendering in Mantra exclusively and not generating the ifd file. - Make sure you are rendering the object as points - If your particles have opacity, try the mantra opacity threshold but be careful. Too aggresive leads to popping. - I am sure you aren't raytracing but be sure you aren't. - don't use ambient occlusion. Use a deep shadow map and a couple well placed lights to get depth. - instance smaller point clumps to each particle with random scale and rotate to amplify the number of particles at render time. Quote Link to comment Share on other sites More sharing options...
Jason Posted January 29, 2009 Share Posted January 29, 2009 Additionally, you might be interested in this plugin called clusterThis by Mork Starry http://www.digitalcinemaarts.com/dev/clusterThis/index.html It can replicate clouds of particles at render time if you need to bolster the numbers of points you're rendering. Quote Link to comment Share on other sites More sharing options...
GallenWolf Posted January 30, 2009 Share Posted January 30, 2009 I've also been giving this some thought, it may be possible to run the same particle simulation say, 10-20 times, with each sim running a different seed. And then use the Mantra: Delayed Load procedural to load 'em. Not sure if you can get motion blur this way though... GW Quote Link to comment Share on other sites More sharing options...
kensonuken Posted January 30, 2009 Author Share Posted January 30, 2009 I read a paper on surf up where they have used some DSO to calculate dense points generated at rendertime... But my point is how do we get the direction of each particle to be random according to the parent particle... ( i mean the default way would be very uniform looking than in different directions ... what could be the solution for this... how about the Day after Tomorrow effects.. how they might have handled tons of these things... Quote Link to comment Share on other sites More sharing options...
Marc Posted January 30, 2009 Share Posted January 30, 2009 The Surf's Up dso just (just.. hah!) instanced new points around the base point based on some noise value. I'm pretty sure the id of the particle was used as the seed for the noise so that it didn't change on a frame by frame basis. If those points inherit the base points velocity then you have a fancy cluster. As for the Day After Tomorrow, they used Storm!!!1! to generate the points/volumes. It's basically a voxel based renderer that fills the volume with points instead of with clouds. They're both very similar ideas but approached from different directions, especially when you throw in the frame dicing that Surfs Up used. I've also been giving this some thought, it may be possible to run the same particle simulation say, 10-20 times, with each sim running a different seed. And then use the Mantra: Delayed Load procedural to load 'em. Not sure if you can get motion blur this way though... This works great and mantra can handle a good few million particles at rendertime without completely falling over, which is nice. Any more than that and you'll need to start dicing up the frame to get it to render. I've successfully used the wedge ROP to generate a bunch of sims. Cheers M oh P.S. I had some motion blur issues with clusterThis, which I couldn't quite track down. I was trying to get a file together for Mark but sadly was derailed before I could... just a heads up if you give it a try. Quote Link to comment Share on other sites More sharing options...
kensonuken Posted January 31, 2009 Author Share Posted January 31, 2009 (edited) http://weblogs.variety.com/photos/uncatego...xplodingcar.jpg is this quality of fireball possible with H? I want to do something like this.. Im stuck Edited January 31, 2009 by kensonuken Quote Link to comment Share on other sites More sharing options...
Marc Posted January 31, 2009 Share Posted January 31, 2009 um... yes? It's kind of hard to help you move forward when we don't know where you are. How far have you gone, what have you tried? Where exactly are you stuck? Unless you're hoping for a finished hip file, in which case I'm fairly certain you'll be disappointed . M Quote Link to comment Share on other sites More sharing options...
odforceuser42 Posted February 1, 2009 Share Posted February 1, 2009 here are a few links that i came across not too long ago that i believe are actually related to the thread! From Hosuk Chang's website: Data Expansion 1 - "Simply, what I want to do is that I want to make a particle simulation with small amount of particles, and in render time, they can be expanded to larger number of particles." http://www.hosok2.com/project/dataExpansio...xpansion_1.html Data Expansion 2 - "This page shows distribution method for second particles from the seed particle and implementation using RiFilter and Procedural Primitive DSO in Renderman." http://www.hosok2.com/project/dataExpansio...xpansion_2.html Data Expansion 3 - skipping to the end... "4. Conclusion - I can say using this solution I can render almost unlimited number of particles. It is independent from the amount of memory. If I want to render more particles, I only need more time" http://www.hosok2.com/project/dataExpansio...xpansion_3.html Quote Link to comment Share on other sites More sharing options...
GallenWolf Posted February 2, 2009 Share Posted February 2, 2009 ken: I think it is possible. My "research" shows that breaking down the FX Tools: Heavy Smoke OTL may be the key ;-) Marc: Is that all the Surf's Up dso does? Instance a heavy cluster of points to a particle? I think i remember reading the waves were rendered in layers right? So do you guys combine multiple deep shadow maps into one or something like that? (Drat I need to find that pdf again...) odforceuser42: Hosok's stuff is amazing. I've been following his site for a bit but unfortunately alot of the prman rib related stuff does not seem to translate very well to the ifd stream. Any thoughts? I've managed to get the Mantra Program Procedural to "work" with python, but I'm not getting basic stuff like motion blur or even attributes to be passed in. GW Quote Link to comment Share on other sites More sharing options...
Marc Posted February 2, 2009 Share Posted February 2, 2009 Well the waves and the spray are two totally separate issues really. But the spray dso at it's most basic level does what Hosuk's does. Cheers Marc Quote Link to comment Share on other sites More sharing options...
GallenWolf Posted February 2, 2009 Share Posted February 2, 2009 Thanks Marc! Alvin Quote Link to comment Share on other sites More sharing options...
Guest xionmark Posted February 2, 2009 Share Posted February 2, 2009 Additionally, you might be interested in this plugin called clusterThis by Mork Starry Nanoo Nanoo Quote Link to comment Share on other sites More sharing options...
Guest xionmark Posted February 2, 2009 Share Posted February 2, 2009 I'm pretty sure the id of the particle was used as the seed for the noise so that it didn't change on a frame by frame basis. <SNIP>oh P.S. I had some motion blur issues with clusterThis, which I couldn't quite track down. I was trying to get a file together for Mark but sadly was derailed before I could... just a heads up if you give it a try. Gotta a new release just about to go out. The motion blur is a bugger, Mark E. clarified a few things for me, for now we're limited to deformation blur. So I've got a few options on this. The new version has CVEX processing for the instanced geo, which gives the user the capacity to throw VEX code into the mix. Presently the motion blur is derived (hap hazardly) from the velocity of the birth particle. Lots of work to do there! So I'm partly wondering if I should use the CVEX hook for motion blur processing (move P to where you want), or keep it simple ... I'll write up more details when I release this next version. Basically the new features are: * CVEX processing of instanced geo (limited to moving P at the moment) * Optimization for deep shadow passes. * hmmmm .... I thought there were more, still waking up ... too much code last night! :-) Should be later today. Mark Quote Link to comment Share on other sites More sharing options...
Guest xionmark Posted February 2, 2009 Share Posted February 2, 2009 Marc: Is that all the Surf's Up dso does? Instance a heavy cluster of points to a particle? I think i remember reading the waves were rendered in layers right? So do you guys combine multiple deep shadow maps into one or something like that? (Drat I need to find that pdf again...) Yes, it did more than just instance points. It could also "slice" the scene in Z and generate multiple layers of instanced geo which then would have it's own deep shad map. I think this was talked about in the paper that Matt Hausman explained the pipeline. The waves were "hand sculpted" and then various layers of deformation/texture mapping & displacement added (much of it based on Jerry Tessendorf's work, IWave), add foam balls, lip spray, streaks, crests, etc. ... it was a fun show to be on to be sure! Debbie Carlson did much of the heavy lifting on the wave R&D, Steve Blakey really got to know (and love!) the cluster DSO! Neato. Mark Quote Link to comment Share on other sites More sharing options...
GallenWolf Posted February 3, 2009 Share Posted February 3, 2009 Hello Mark! Thanks for releasing clusterThis! Been testing it out recently at home. It's fun! Curiously, what kind of particle limits have you been able to push out? I'm topping out a 16gb machine with about 20million points. (10 particles, 2 recursion, 1 million base particles) Quote Link to comment Share on other sites More sharing options...
Guest xionmark Posted February 5, 2009 Share Posted February 5, 2009 Hello Mark!Thanks for releasing clusterThis! Been testing it out recently at home. It's fun! Curiously, what kind of particle limits have you been able to push out? I'm topping out a 16gb machine with about 20million points. (10 particles, 2 recursion, 1 million base particles) Hi, I've been able to get around 20million+ on a 4 core/5G machine. Need to do much more testing, there's a lot of circumstances that can cause some real problems. A new update is available (v1.4.4): http://digitalcinemaarts.com/dev/clusterThis/index.html I will update the posting ( http://forums.odforce.net/index.php?showto...24&start=24 ) with info about changes, etc. Mark 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.