[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Public WebGL] A Declarative node set for WebGL?

Interstingly, the response to every one of your points below is the same. Why discuss this here? Why this scene graph rather than any other? How can you use this to produce other scene rendering techniques? I'm discussing a declarative form of WebGL. It has all the benefits and characteristics of WebGL with the additional benefits of native traversal, DOM and CSS integration and events.


On Feb 24, 2010, at 6:33 PM, Gregg Tavares <gman@google.com> wrote:

Personally I don't think this discussion even belongs in the WebGL mailing list. Has there ever been a discussion of adding a scene graph to OpenGL?  I hope not.

But, to weigh in on it anyway I think this a HORRIBLE idea.

The problem is there are infinite ways of representing 3d data. No matter which one you pick you'll only be covering some small set of use cases. Typically the least interesting use cases.  A scene node representation will pretty much only get you some spinning models at best. (I'm sure the X3D guys will disagree).

Some examples: BSP, Portals, Heightmaps, PVSs, LODs, Dynamic LOD generation, Skinning, BlendTargets, Subdivision Surfaces, Nurbs, Metaballs, Shrubs, Billboards, Particles, Voxels,

Then people will start asking for collisions, picking, physics, animation...

Every one of those items is extremely application specific. Hard coding one solution into stone in the browser is just not the right way to go IMO.

I think the GLGE demo

Which shows loading collada files shows that it's possible to write a scenegraph system that fits your particular apps needs directly in _javascript_.

Also CopperLicht

Sure there are going to be some places that are slower than a built in scenegraph would but many of those can be solved by creative optimization of _javascript_ and further performance enhancements of the various browsers' _javascript_ engines.