Pixar on OpenSubDiv: and Sub Division Surfaces in General
wancow
Posts: 2,708
http://www.youtube.com/watch?v=Y-3L9BOTEtw
This is a great video which explains a whole lot about Sub Division Surfaces.
http://graphics.pixar.com/opensubdiv/index.html
What they show in that video is flippping insane! It'll make you drool!
Post edited by wancow on
Daz 3D is part of 
Connect
DAZ Productions, Inc.
7533 S Center View Ct #4664
West Jordan, UT 84084
Licensing Agreement | Terms of Service | Privacy Policy | EULA
© 2024 Daz Productions Inc. All Rights Reserved.
Comments
Wow, that really is impressive.
This would be awesome in Daz Studio. The current SubD module is very laggy, not to mention limited to only two steps.
Since each level of SubD quadruples the meshes polygon count you really don't want to go beyond two levels.
jestmart, did you even look at the links? It's totally new technology using among other things, adaptive subdivision, i.e. does not quadruple the mesh polygon count.
The examples shown only used uniform subdivision, it was still the beta software being used. The adaptive subdivision is planned for Version 1.0.
The big things for me were the move to GPU processing (which produced the amazing performance gain), making it open source and the really big thing... PTEX!!!!
Open source standards take a long time to filter through.
In Cinema 4D the equivalent to the "semi-sharp edges" tag is the "subdiv weighting" tag. There's currently no standard format to get that weighting information into studio or any other software. I'm restricted to basic, subdiv modelling techniques and putting in extra edge loops and polys.
So all that has to happen is that everybody converts to OpenSubDiv and OpenSubDiv gets supported by FBX... eventually it will happen but I'm not holding my breath.
I wouldn't be surprised though, if all these things get into DS before it they get into Cinema 4D! :coolsmile:
I think this presentation demonstrates and explains adaptive subivision more clearly. You can also see a technician model a telephone hand set from scratch using the technology. Furthermore, it doesn't have all those annoying Powerpoint slides which dominate the original video!
This is incorrect. It has parameter limits that limit it to 0-2 levels to protect users that don't know what they're doing from killing their machine setting it too high. But if you turn off the parameter limits for the subdivision level on an object (click the little gear icon next to the name of the parameter), Studio will happily let you specify a higher subdivision level.
Yes that too cwichura, forgot about that part ;)
Thanks wancow and hiro for the links, and everyone for the input. I love looking at and discussing exiting new technology like this. It just creates a bit of 'I want it now' sensation at times ;p
I noticed in the presentation re: Hiro's link that the tessellation has high ngon & tri count. I don't remember how high the extraordinary vertices count was if any. The point being, the resulting mesh seems to break many of the rules to good topology as we now work. I'm wondering what the implications and impact this will be, and if the general rules will still hold true when developing the original mesh before subdiv.
Looks like this was the laptop they where using in the presentation.
You ain't kidding about the "want it now" sensation! I saw that and I just about spit out my coffee when I saw the bit about adaptive subdiv and edge creasing... the whole thing with displacement in display render is...
DAHROOOL... (have I slobbered on you yet?)
that's a great question. I'd pay attention to what the guys on CGTalk say about modelling using it once this makes its way into general use. I'm suspecting the excitement level over the technology is enough that it'll happen pretty fast.
That was totally cool and yes, want now. Genesis would benefit from this greatly!
Actually, I picked this link up in a thread on the need for good topology, and the commenter suggested that traditional concepts of good topology (which I always try to follow when I'm modelling), or at least the idea of maintaining good topology as the model becomes more complex, may become redundant with this sort of technology.
I would have thought that it would still be desirable to begin with good topology at the base level. I don't know though, perhaps it will be like sculpting often is: start with a sphere or a cube and let the software worry about the rest. As one of the reasons for having good topology is to enable the model to be rigged and animated well, and this adaptive subd is said to animate well (which it will have to do, if Pixar are behind it) then perhaps such concerns will become a thing of the past.
You explained well what I was thinking but didn't put as well :)
It just occurred to me that the release of OpenSubDiv is actually a way for Pixar to force the 3D industry to bend its will.
With the release of OpenSubDiv, what they we are very likely to see is that modellers, the software, will have to implement it to remain viable as it's very likely to become an industry standard very, very quickly.
Since PRman already has a leg up on OpenSubDiv, it will, if I'm right, remain the 800lbs Gorilla in the industry, thereby securing its place for years to come.
I don't think I'd say "bend to it's will" since they made it open source!