Scientific Visualization? GLCanvas vs. GLJPanel

I work in the scientific visualization field; specifically molecular modeling tools. These programs typically are multi-document apps having several windows open showing 3D graphics rendering of molecule(s), plots and spreadsheet views of data, buttons, etc.

Our existing package is a Win32 app using OpenGL but we’re now working on a new Swing-based program. Our initial experiments rendering molecular structures on a GLCanvas were encouraging (until we discovered the Z-order problem with mixing heavyweight and lightweight components).

It looks like our only option is to use the GLJPanel. However, experiments with GLJPanel in our app were not encouraging due to performance and stability problems. (I can reproducibly blue-screen my XP laptop by not turning off doubleBuffering in the GLCapabilities object!!)

Finally to my question: We must have a Swing app with an MDI look and feel. Is there anyway to do our 3D renderings in this app using JOGL that is hardware accelerated (or at least “fast”) ? So far it looks like the answer is no, but I’ve only been researching JOGL for a few weeks.

It appears that enhancing GLJPanel to include hardware acceleration is not a high-priority item for the JOGL dev team. Prior threads in this forum talk about getting the pbuffer rendering to work with GLJPanels, but still with performance problems. (btw its gotta work on OSX, Windows, and Linux :slight_smile: )

Thanks in advance for any help.
Mark

GLJPanel is never going to be fast, though with PCIExpress video cards it may get very useable. What kind of frame rates were you looking for? What size gljpanel are you looking to use? Once I get a little extra time I am going to try one last nvidia specific trick to get my implementation of GLJPanel to run a little faster but right now it is just slow. The doublebuffering bug is fixed in the cvs but I don’t think it is in the precompiled binaries yet.