What do you think of this?
Build once, run anywhere 3D content?
Slashdot link to the link
Remember, Web3D X3D (before that VRML)
I actually worked on a huge DOD project called SEDRIS 10 years ago that was trying to do this as well. Most massive bloat format EVER.
I just saw this on Slashdot. Sounds interesting.
That’ll have about as much success as a universal image format. Just me give verts, texture cooridinate and triangle indices, thats all I need… for now at least :-/
I agree this is a monumental task as well, there is one flaw in the idea as I can see it.
Those other format success stories (even say .doc and .ppt) are de facto standards, not standards by design. It’s true a group established them, but there “universal-ness” happen by use not because they said so.
I imagine a 3D standard will happen the same way. Is Sony on that team? How about Doug Twilleager from the original Java3D guys 
This is just Shockwave3D, and it’s going to be very far from universal write once run anywhere.
We had the displeasure of working with Intel on this when they approached the Web3D consortium about this just over a year ago. At the time they focussed it on the CAD market and brought a heap of people to “work” on it. My business partner, Alan and I, spent a lot of time in face to face meetings with the Intel people over this stuff trying to get it adapted to CAD usage (simple summary, in it’s current form it would fail miserably). Intel were eventually forced to leave the consortium due to some very underhanded stuff they were doing (rigging elections and all sorts of really nasty stuff). The funny thing is that if you dig through Intel’s press release archive, you’ll see that they announce exactly the same thing about every 2 years. It’s still the same technology which started with Intel/Adobe Shockwave3D development.
The ideas and philosophy behind the format are quite interesting from an academic standpoint. The concept of non-transformation based scene graph and separate runtime structures for relating behavioural aspects are quite a nifty idea. However these strengths are also it’s greatest weakness too. Because of this architecture, it is close to impossible to use any hardware accelaration. Everything must be done at the CPU level. Taking this one step further, it would be impossible to add programmable shaders into it. The core problem is the separate runtime structure (known as Modifier Chains) which is a huge message passing structure between various stages. This whole setup assumes callbacks between the stages and a fixed memory layout. For example, you set up a modifier chain for animating a human. The chain starts with a part assembly module (collects all the pieces of geometry together from the structural relation graph), then puts it through a skinning module, then some various other modules, finally endig up with the IK module to do the animation before pushing it through to the rendering rasteriser. This completely eliminates any ability to put in shaders or other hardware assisted structure at the runtime level. Other problems include things like only having fixed culling options - the content must define what culling implementation to use from a fixed list that can’t change or be extended, which means content is fixed by the author’s knowledge, not by what is best on the platform the content is run on.
Almost a complete waste of time. But more than that I cannot say.
Cas 
[quote]I agree this is a monumental task as well, there is one flaw in the idea as I can see it.
Those other format success stories (even say .doc and .ppt) are de facto standards, not standards by design. It’s true a group established them, but there “universal-ness” happen by use not because they said so.
I imagine a 3D standard will happen the same way. Is Sony on that team? How about Doug Twilleager from the original Java3D guys
[/quote]
Agreed, in principle. But history suggests the easiest way to juggernaut something like this through to success is simply to make it the only format used by a new tool - and make that tool dominant.
People don’t actually use formats, they use tools. Once they pick a tool, they’ll use whatever that tool prefers. Given that modern 3D modelling tools are still, on the whole, poor (especially in terms of their arcane GUI’s), someone who brings out a kick-ass tool is likely to have their pick of whatever format they wish to turn into the standard.
(NB this is of course how we ended up with word docs as a “standard” file format despite their lack of basic features like explicit format-versioning - MS leveraged “word” dominance into “.doc” dominance)
I’d have to say that Maya is a kick-ass 3D tool.
However, the reason the Maya format isn’t popular is because it is technical an Maya Embedded Langauge script (MEL) that was the process (and history) of making the 3D scene you have. Just trying to read the format is INCERDIBLE difficult, you have to basically build a MEL VM like Maya.
You want 3D models? Exporting to something simplier of your choice is the typical path.