Posted by ThomasR on Feb 12, 2016; 8:03pm
URL: https://forum.jogamp.org/OpenGL-OS-X-tp4036147p4036210.html

Hi Julien,

Thank you for this excellent background. Sorry I meant to say JogAmp Ardor3D Continuation. I'm a scientist and software developer here at the University of Wisconsin-Madison working on VisAD: https://en.wikipedia.org/wiki/VisAD. VisAD is a Java api primarily used in interactive scientific data fusion and visualization on workstation and laptop computers. It's the basis of many systems including Unidata's Integrated Data Viewer, Julien Chastang has posted here, VisBio (Curtis Rueden, now maintaining Java3d on MavenCentral), HYDRA (HYperspectral viewer for the Development of Research Applications), McIDAS-V and several others. Its dependent on Java3D so we have concerns going forward that it may become less stable or stop working all-together. To be fair, Java3D has proven extremely reliable to this point, as many of these system are deployed around the world on Windows, Linux and OS X. Mac is an extremely important platform as they've become the defacto standard for scientists and engineers at least in our domain, hence my questions on OpenGL and JOGL on OS X in the future.

Stability is an extremely important characteristic, for example HYDRA is used as a professional development tool for remote sensing scientists in seminars sponsored by NASA and NOAA, but it would be nice to begin to leverage new capabilities provided by the gpu via shaders. Currently VisAD renders via the fixed pipeline generating its own textures and geometry from science data which are passed to Java3D. We have our own marching cube algorithm to generate isosurfaces from 3D scalar fields, but Java3D doesn't handle depth sorting of semi-transparent triangles very well. We have texture-based volume rendering (first demonstrated by Bill Hibbard here in Vis5D) but it's slow and requires alot of memory - maybe a ray-casting approach is better. Would Ardor3D Continuation provide better capabilities here? But we also have needs for headless and off-screen rendering. If we can't stay with Java3D, I would prefer not to have to write our own scene api based on JOGL, but to start with something professional and contribute back to the community. My hope is that Java3D can be our base for the next 2 or 3 years while we possibly make a  transition.

Tom