Posted by campbell
on July 28, 2003 at 9:46 PM PDT
A continuation of Chris Adamson's ponderings on the state of JOGL and Java 3D; how JOGL affects Java 2D; how it's way past my dinner time and how I'll pass out if I don't eat in the next five minutes.
[I was going to reply to Chris's excellent weblog in the talkback section, but I started rambling and it touched on some other thoughts I've had, so I decided to ramble here instead... Keep in mind that I'm only half-wearing my Sun cap right now (kind of like one of those green and brown, half A's, half Giants caps that were popular in the '89 Series), so I'm not speaking entirely on behalf of the company.]
In response to Chris Adamson's recent blog entry, The End (of Java3D) and the Beginning (of JOGL) :
I'd just like to point out that JOGL is not an all-out replacement for Java3D. The two can co-exist, and one could potentially rewrite the platform-specific layer of J3D to sit atop JOGL. Java3D does indeed act as an "isolation layer" for the underlying platform when a developer uses its "immediate mode" APIs, but more importantly Java3D offers a high-level scene graph API. Many educational and corporate institutions have chosen Java3D because of its scene graph offerings, in addition to the appeal of its cross-platform nature.
On the other side of the coin you have the traditional game shops, who want to get as close to the graphics platforms/hardware as possible. Many of these folks are finding JOGL a better fit because it's a lower-level API, and they can make use of their existing OpenGL knowledge/code base. So I think it depends on the type of application you're developing which API best suits your needs. The gaming community has been clamoring for official Java bindings for OpenGL for quite some time, so that's where Sun's efforts seem to be heading, but don't count Java3D out for good; it still serves its purpose quite well as a higher level 3D graphics library.
Related to this discussion, we're also starting to see some folks on the javagaming.org forums asking whether JOGL would be a better fit than Java 2D for their apps/games. Again, JOGL is not the end-all and be-all Java graphics library. Many people don't realize that OpenGL is actually an expressive 2D library, despite its tight association with the 3D world. However, there's so much more to 2D graphics than rendering lines and sprites really fast (think medical imaging, complete support for any image format or color/sample model, printing, text rendering, stable offscreen rendering, etc). This is where Java 2D really blows the proverbial socks off all the other 2D libraries out there.
My answer to those folks on javagaming.org is the same as my J3D response: Java 2D is a higher-level, easier-to-use, more robust, more full-featured 2D rendering API than JOGL. Like J3D, the two technologies can play well together (if we do our job correctly, there should be no reason why the two API's couldn't be used in the same application). Also like J3D, we use hardware-accelerated graphics libraries (such as Direct3D and OpenGL) under the hood, so for many applications, performance should be virtually the same whether you use Java 2D or JOGL. As I mentioned earlier for J3D, we could also port our OpenGL-based Java 2D pipeline to sit atop JOGL (in fact, we're exploring this idea for a future release, which should further decrease our dependence on native C code). If you want to access the very latest in hardware technology, such as programmable shaders, or if you have large data sets of vertices and you're comfortable with the increased complexity of OpenGL, then JOGL would certainly be a better fit.
So the choice is yours... Each API has its benefits; it's up to you to evaluate which one is best-suited for your next project!