robert.magee Posted July 27, 2005 Share Posted July 27, 2005 Side Effects Software, is pleased to announce the availability of Houdini 8 in public beta. Houdini 8 delivers many significant enhancements to its expanding CG artist community with particular emphasis on dynamics, lighting and character animation. CG artists from around the world can download the Houdini 8 public beta and run it for free as part of the Houdini Apprentice Program and learn about all the new features. Visit us at www.sidefx.com for more details and to download the Houdini 8 Public beta. Robert Magee Senior Product Specialist Side Effects Software Quote Link to comment Share on other sites More sharing options...
altbighead Posted July 27, 2005 Share Posted July 27, 2005 :notworthy: :nuke: :notworthy: Quote Link to comment Share on other sites More sharing options...
Szymon Posted July 28, 2005 Share Posted July 28, 2005 Am I wrong or DOP's are the slowest thing around? Look at attached scene. Simple Cloth, 20 seconds a frame, that must be a joke guys! It does not compare to Syflex in any way. It's even slower than 3dsmax7. test3.rar Quote Link to comment Share on other sites More sharing options...
TheUsualAlex Posted July 28, 2005 Share Posted July 28, 2005 Hi, Can you please repost your attachment in .zip format...? I am on Linux box... No Rar here... Quote Link to comment Share on other sites More sharing options...
edward Posted July 28, 2005 Share Posted July 28, 2005 http://www.sidefx.com/index.php?option=com...iewtopic&t=4175 I don't think Houdini's cloth is up to Syflex's standards quite yet. They've been at this game much longer than we have. We're still doing as much as we can to improve the solver prior to the release of Houdini 8.0. The weak points we're focusing on are cloth/volume collisions and cloth/cloth collisions. Cloth/volume collisions in particular have seen a fair bit of work over the past few weeks, and should hopefully be getting more usable.We are working on clothing animation, and we have had some successful simulations there, but I don't want to make any promises about how far we'll get on that front. Unfortunately I can't tell you that we've got any rabbits up our sleeves. You see it as we write it. But we have lots of ideas on how to improve things, and some time left to implement them. Hopefully it will be enough for the Cloth Solver to be useful, even if it's not yet a match for Syflex. Quote Link to comment Share on other sites More sharing options...
Guest Guest Posted July 28, 2005 Share Posted July 28, 2005 lets not forget that we are talking about beta version Quote Link to comment Share on other sites More sharing options...
peship Posted July 28, 2005 Share Posted July 28, 2005 lets not forget that we are talking about beta version 19964[/snapback] sorry for the anonymous post Quote Link to comment Share on other sites More sharing options...
stevenong Posted July 28, 2005 Share Posted July 28, 2005 Hi there, Am I wrong or DOP's are the slowest thing around?Look at attached scene. Simple Cloth, 20 seconds a frame, that must be a joke guys! It does not compare to Syflex in any way. It's even slower than 3dsmax7. 19959[/snapback] If you lower the number of Maximum Substeps in the Cloth Solver DOP down to 16 or 32, it will run faster but it's less accurate. Cheers! steven Quote Link to comment Share on other sites More sharing options...
Szymon Posted July 28, 2005 Share Posted July 28, 2005 Here is same scene in zip file as requested. Also look at test2. That's fast? Ouch. I know it's all beta but I expected something more revolutionary. test3.zip test2.zip Quote Link to comment Share on other sites More sharing options...
sibarrick Posted July 28, 2005 Share Posted July 28, 2005 I think the revolution lies in the fact that any sim can interact with any other sim, and you don't get that anywhere else AFAIK. I sure they will work on performance its a know issue. Quote Link to comment Share on other sites More sharing options...
Guest Guest Posted July 28, 2005 Share Posted July 28, 2005 Here is same scene in zip file as requested.Also look at test2. That's fast? Ouch. I know it's all beta but I expected something more revolutionary. 19976[/snapback] test2 has "Use Volume Based Collision Detection" turned off. So it is basically using a cloth-like collision detection algorithm (surface collisions) rather than a Volume Based approach, which has faster performance in contact situations provided that both objects have volume . Thanks for the feedback, Jack Quote Link to comment Share on other sites More sharing options...
bugzpodder Posted July 28, 2005 Share Posted July 28, 2005 oops... sorry about the Guest post above. Thanks, Jack Quote Link to comment Share on other sites More sharing options...
bugzpodder Posted July 28, 2005 Share Posted July 28, 2005 and for those of you who waited patiently for the balls to collide together, you'll notice that they actually inter-penetrate. This is because the sphere's type is "primitive". Change it to polygon should avoid this from happening. RBD Objects likes polygonal stuff much better (with points and edges, something primitives are missing). oh and, I dont believe you need to have "use animated geometry" on since AFAIK it is for non-static SOP Geo. Since we are on topic about that demo, one more thing, you might want to turn Min and Max substeps on the RBD Solver to at least 10. This is necessary for resolving tangles in contact cases when Use Volume Collision Detection is off. Thanks, Jack Quote Link to comment Share on other sites More sharing options...
peship Posted July 28, 2005 Share Posted July 28, 2005 Heya bugzpodder, maybe you can bring some light here ... I noticed some problems with RBD's collision detection, which causes penetrating and popping objects. I got enough experience with DOPs testing the alpha version of H8 at studio few monts ago, why that yesterday i directly started doing some more complicated stuff. Then i noticed some issues with the simulation quality. Step by step i simplified my tests and at last i stuck with the most simple scenario - bouncing ball over a plane. I created a meshSphere 20x20 and meshGrid 10x10, turned them to rigidBodies and played simulation. After the sphere stops bouncing, it starts shifting and sliding over the plane ... And here is the question - do you know how to prevent this behaviour ? Any input will be very welcome. BTW - great stuff, nice conception and design. Quote Link to comment Share on other sites More sharing options...
bugzpodder Posted July 28, 2005 Share Posted July 28, 2005 Heya bugzpodder, maybe you can bring some light here ...I noticed some problems with RBD's collision detection, which causes penetrating and popping objects. I got enough experience with DOPs testing the alpha version of H8 at studio few monts ago, why that yesterday i directly started doing some more complicated stuff. Then i noticed some issues with the simulation quality. Step by step i simplified my tests and at last i stuck with the most simple scenario - bouncing ball over a plane. I created a meshSphere 20x20 and meshGrid 10x10, turned them to rigidBodies and played simulation. After the sphere stops bouncing, it starts shifting and sliding over the plane ... And here is the question - do you know how to prevent this behaviour ? Any input will be very welcome. BTW - great stuff, nice conception and design. 19986[/snapback] Thats a pretty big sphere! Anyways I am not exactly sure of what you are seeing. It would help if you send me your example file or post it in the new DOPs forum and i'll take a look at it. Thanks, Jack Quote Link to comment Share on other sites More sharing options...
michael Posted July 28, 2005 Share Posted July 28, 2005 friction baby...friction Quote Link to comment Share on other sites More sharing options...
peship Posted July 28, 2005 Share Posted July 28, 2005 heh, when i said sphere 20x20 i ment resolution, not scale. i really dont think it's needed to post the scene file. just spend 10 seconds - create a geometryOP, inside it create a meshSphere and grid. go to DOPs turn them to rigidBodies, apply gravity, merge, RBD solver, applyData, play the animation - wait a second until the sphere stop bouncing ... that's not a friction related issue. Quote Link to comment Share on other sites More sharing options...
bugzpodder Posted July 28, 2005 Share Posted July 28, 2005 heh, when i said sphere 20x20 i ment resolution, not scale.i really dont think it's needed to post the scene file. just spend 10 seconds - create a geometryOP, inside it create a meshSphere and grid. go to DOPs turn them to rigidBodies, apply gravity, merge, RBD solver, applyData, play the animation - wait a second until the sphere stop bouncing ... that's not a friction related issue. 19989[/snapback] I assume its a 10x10 msh grid? And sphere is at some height (maybe 5?) and that merge is applied in the very end? (like Sphere -> gravity -> solver -> merge, grid -> merge) ? You were talking about "penetration" and "popping"?. Then you mentioned "shifting" and "sliding". I definitely dont see penetration (which was what I am confused about). I assume you meant that the sphere was shifting slightly from the resting position after it stops bouncing? Thats probably due to the sphere's imperfect geometry. You can disable it by turning its rotational stiffness off. Thanks, Jack Quote Link to comment Share on other sites More sharing options...
peship Posted July 28, 2005 Share Posted July 28, 2005 yes, the setup is like what you said. I talked about "penetration" and "popping" because i noticed these issues. I decided to track them down to see what is going on and all that finished up with the bouncing ball thing. I had not chance to make proper looking simulation, why that decided to ask here. My problem is not how to fix the bouncing ball somehow, i just used this simple setup, trying to understand how to improve the dynamic simulation quality. I believe that if i can manage to get the bouncing ball work right, i can avoid the "penetration" and "popping" issues in similar manner. Sorry for the confusion, i hope it makes sense now. Quote Link to comment Share on other sites More sharing options...
bugzpodder Posted July 28, 2005 Share Posted July 28, 2005 the setup is like what you said.I talked about "penetration" and "popping" because i noticed these issues. I decided to track them down to see what is going on and all that finished up with the bouncing ball thing. I had not chance to make proper looking simulation why that decided to ask here. Sorry for the confusion. 19992[/snapback] If you still see those issues, feel free to post it here and I'll look at them for you. Thanks, Jack 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.