Guest tar Posted May 9, 2014 Share Posted May 9, 2014 Just curves. Been building tools for fur and hair lately, creating modifiers for guides generated in xgen, zbrush and Houdini. Maybe we should move this to another thread. Cool - thanks! A new thread will work well but probably not needed yet. - I'm confident that it can be broken Quote Link to comment Share on other sites More sharing options...
magneto Posted May 10, 2014 Share Posted May 10, 2014 Do you guys also get an issue where Houdini switches to handle tool automatically? I always use the view tool unless I want to use the handles, then I press Enter, and then after I am done, I press Esc. But I know some actions do this automatically which I don't want. I submitted to support but they couldn't reproduce it I think this is sort of thing that happens if you use Houdini for a while, i.e. 10 mins. Quote Link to comment Share on other sites More sharing options...
Guest tar Posted May 10, 2014 Share Posted May 10, 2014 Do you guys also get an issue where Houdini switches to handle tool automatically? I always use the view tool unless I want to use the handles, then I press Enter, and then after I am done, I press Esc. But I know some actions do this automatically which I don't want. I submitted to support but they couldn't reproduce it I think this is sort of thing that happens if you use Houdini for a while, i.e. 10 mins. Good news is I did just see this whilst testing on Linux. Will have to find reproducible steps. There is a topic now for the last few H13 bugs. Please post any H13 bugs there. Thanks! The Bug is Dead, Long Live the Bug (H13.0.X) http://forums.odforce.net/topic/20105-the-bug-is-dead-long-live-the-bug-h130x/ Quote Link to comment Share on other sites More sharing options...
Guest mantragora Posted May 16, 2014 Share Posted May 16, 2014 Add possibility to create node presets in initialization script with python, so we can add a little AI to them. Quote Link to comment Share on other sites More sharing options...
LaidlawFX Posted May 29, 2014 Author Share Posted May 29, 2014 Viewport support for Edge Display. You can display points and prim numbers which makes logical sense, where as edges are a derivative of the two, so it's not as straight forward. But it be cool if there were a way to show p8-9 at the middle of an edge(the two associated points with that edge) so you could selective edit and modify it. Quote Link to comment Share on other sites More sharing options...
PatW Posted June 1, 2014 Share Posted June 1, 2014 Improve auto layout network. I never like to hit 'L' key in network view unless I'm sure it won't mess up the network setup. If there is branching or complex wiring, it's highly that the command will not sort nodes as I want. Quote Link to comment Share on other sites More sharing options...
paxsonsa Posted June 3, 2014 Share Posted June 3, 2014 I tried search to see if anybody mentioned it and i couldn't find anyone who did. - I would love to see a the stereo rig re-done with options for a toe-in, parallel, or Off Axis setup. I would also like to be able to view stereo plates in the viewer (i.e. If I am working on a native/converted stereo shot I want to be able to view the image plane with the stereo plates) - Lastly I would love to be able to source dispairty and render disparity maps in screen space. Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 3, 2014 Share Posted June 3, 2014 Some interesting stuff could be - find a bi-directionnal path tracing solution to get physicall caustics without relying on photon-mapping - add the possibility to setup light temperature in kelvin for the lights - possibility to choose the rendertime like in RIS and Maxwell and go back to the calcul to reduce noise without recomputing all that was already done if needed. 1 Quote Link to comment Share on other sites More sharing options...
LaidlawFX Posted June 3, 2014 Author Share Posted June 3, 2014 Some interesting stuff could be - find a bi-directionnal path tracing solution to get physicall caustics without relying on photon-mapping - add the possibility to setup light temperature in kelvin for the lights - possibility to choose the rendertime like in RIS and Maxwell and go back to the calcul to reduce noise without recomputing all that was already done if needed. Have you been reading the new Renderman stuff??? http://www.fxguide.com/featured/pixars-renderman-turns-25/ 1 Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 3, 2014 Share Posted June 3, 2014 (edited) Yes i have And i am starting to be quite envious about this new RIS stuff ... Having something similar in Mantra would be cool ! not necessarly something like the one Pixar looks to use http://cgg.mff.cuni.cz/~jaroslav/papers/2012-vcm/ But something that would bring some Physicall caustics at a lower price than Maxwell in H ! Edited June 3, 2014 by sebkaine Quote Link to comment Share on other sites More sharing options...
Guest tar Posted June 3, 2014 Share Posted June 3, 2014 - find a bi-directionnal path tracing solution to get physicall caustics without relying on photon-mapping There's an option on the PBR node, just not turned on by default and looks a bit noisy Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 3, 2014 Share Posted June 3, 2014 Thanks for the hint Marty , i have seen this one in the PBR doc , but i have doubt on the bi-directional nature of this feature ? It looks that you keep firing from the camera and not from the light source ? I could be completely wrong as the doc is quite vague about this. Quote Link to comment Share on other sites More sharing options...
Guest tar Posted June 3, 2014 Share Posted June 3, 2014 good question, what's a proper test for bi? Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 4, 2014 Share Posted June 4, 2014 (edited) Well i have no idea ... A first clue is that it is definitly slower ! What make me think it is still uni-directional path tracing, is that like in Arnold if you want noise free caustics you must have a huge number of samples. https://support.solidangle.com/display/AFMUG/Caustics "The caustics switches in the Standard shader mean that you can tell diffuse GI rays to 'see' the mirror reflection, glossy reflection and refraction from the shader By default only direct and indirect diffuse rays are seen by GI rays" the Mantra option looks quite similar so i would tend to believe that Mantra is also uni-directional path tracer One possible way would be to check how Mantra handle interior scene but it's not bullet proof either ... We need a true nerds to check this out ! Cheers E Edited June 4, 2014 by sebkaine Quote Link to comment Share on other sites More sharing options...
LaidlawFX Posted June 4, 2014 Author Share Posted June 4, 2014 You might have to ask sidefx, for a true nerd at least a mantra nerd of that level. I imagine they may tell you it's bi-directional straight up ... but I think it is bi-directional as what is mentioned above, plus also I believe that was one of the steps in how they solved shadow mask in PBR with h13... There code follows very loosely http://www.pbrt.org/ with as you can imagine a whole bunch of optimized algorithms plus Houdini history wrapped it up to be different... 1 Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 4, 2014 Share Posted June 4, 2014 Thanks for the input Ben , you're right a little mail to Janet could be a good idea ! Quote Link to comment Share on other sites More sharing options...
Guest tar Posted June 4, 2014 Share Posted June 4, 2014 (edited) A little testing and it would seem to be like the Arnold implementations. Adding a Caustic Light with photon map is the only way to create nice caustics. Please prove me wrong That's the big deal with R19 and it's VCM http://iliyan.com/publications/VertexMerging/comparison/ Edit: attached a sample of non-photon mapping cautics on sphere Caustics_Non_PM.hip.zip Edited June 4, 2014 by tar Quote Link to comment Share on other sites More sharing options...
LaidlawFX Posted June 4, 2014 Author Share Posted June 4, 2014 A comparison of our new progressive vertex connection and merging (VCM) algorithm against bidirectional path tracing (BPT) and stochastic progressive photon mapping (PPM) after 30 minutes of rendering They missed the boat on the real time faster is better rendering concept.... Not longer and better. It's def a step forward, but not a dramatic step. The default for hard surface will still be using vray, people will not switch back to renderman for better caustic rendering ability especially for hard surface rendering. Willing to bet as long as Renderman doesn't have a patent on certain parts of this method already, SideFX can take the paper if it was a school based research project, and probably make a better cut down version of it to implement. I think it would be just modifications on their current system. My biggest problem with looking at these research project is they compare the scientific method of doing it alongside each other, as oppose to current real world methods/cheats. Easy to get a doctorate if you got the time going down these roots. Quote Link to comment Share on other sites More sharing options...
kev2 Posted June 5, 2014 Share Posted June 5, 2014 SideFX adopts Swift, deprecates VEX, cVex, Python, and hScript in one fell swoop. Also adds language operator which quickly becomes known as Loopy, sole purpose is to allow older users to comment on the "good old days..." 1 Quote Link to comment Share on other sites More sharing options...
sebkaine Posted June 5, 2014 Share Posted June 5, 2014 (edited) This is where is disagree with you Ben. In fact for many artist, rendering must focus on speed and thus optimisation with various techniques to get nearly real-time rendering. And thus progress in this area is mesure on how fast you get your pictures. Engine like Redshift / Clarisse are really the next gen engine serving this purpose with clever biased techniques mixed together to get the fastest experience in a very clean and optimized environment ! But to me (and it's quite subjective ) , you are still thinking like a 3D guy , and you have the same way to approach rendering than in the 1990ies or 2000ies when we have to think on how to cheats reality. Now that computing power is really there , i would be more in the camp of those who said. A render engine has one goal mimic reality to simulate light. So the absolute goal would be to have a light simulator that has this nearly perfect replication of what would happen in the real world. For this purpose you build a physicall shading / lighting system were there is no tricks anymore. You want 5K tungsten spot you got it , you want orange juice shader you got it and you don't have to touch anything. So now you focus your mind only on making pictures you are now a DOP more than a 3D guy. Maxwell and Arnold are the best representative of this. The bad is that they are slow but you can shade a perfectly photoreal car in 1 hour if you know what you are doing , and you know that there will be no glitches , you will get your images without bug cause they are predictable. Maxwell is by far for me the best engine, but has serious lack in flexibility in a vfx workflow This is were Arnold really shines ! But i'm quite sure SESI has build a LEGO BOX with many element that would allow us to build a kickass workflow that would be very close from Maxwell and far more flexible than Arnold. Physicall Caustics is something that is really missing IMO in H. And photon mapping should be replace, cause i agree with arnold guys it often sucks in production. But even if this new implementation by SESI would multiply render time by 2, to me it would be a progress cause we are closer from reality. People also said well caustics are useless, we can fake them , but they are wrong ... It's like saying DVD is enough compare to blu-ray. But in shots with glass or refractive materials the beauty you get in Maxwell with physicall caustics is stellar. It's like the difference beetween a lemon and an L lens. Sure it's expensive but it's more real and the pictures are better ... So please SESI a new PBR caustics system would be great (even if it's slow ) Cheers E PS : @Marty Thanks for your test ! Edited June 5, 2014 by sebkaine 1 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.