GVDB Voxels Release 1.0 goes live!

We are pleased to announce the first release 1.0 of NVIDIA® GVDB Voxels went live at the 2017 GPU Technology Conference.

The official website is: http://developer.nvidia.com/gvdb
Here you will find the download, github link, samples page and programming guide.

GVDB Voxels is now available open source for Windows and Linux.

This forum is for discussion on all topics related to GVDB.

Regards,

Rama Hoetzlein
NVIDIA

Hi Rama,

GVDB looks great. We will definitely be evaluating for our voxel modeling applications.

Do you have a GVDB roadmap? Would you consider publishing one?

One quick note on the installer. The installer uses “GVDB 1.0 SDK” in the install path. Then when you use CMake and build in VS 2015, you will get a link error about this path. Fix is to take the spaces out of the path. For example, change “GVDB 1.0 SDK” to “GVDB”.

Tommy

Hi Tommy,

Nice to hear from you.
We are looking to get a release 2.0 by the end of 2017 with a focus on simulation on gvdb, as this is a common request. Beyond that I can’t give anything more specific about roadmap.

Part of that future depends on the feedback we get from developers like you about desirable features.

We tested on install paths with spaces, so I’m surprised. There are still a few kinks as we get started since this is a 1.0 release. Making build easy is a high priority, so this feedback is useful.

Can confirm the CMake process WILL NOT SUCCEED on Windows with spaces in the path - suggest you update the Readme accordingly!!
Very excited to see GVDB in the wild after the long wait!
Mike Pelton

Mike,
Can you tell me which sample you are trying?
And which source point, the Win installer or Github?

From the Windows installer, rebuilding GVDB works fine for me with the cmake base path:
“C:\ProgramData\NVIDIA Corporation\GVDB SDK 1.0\source\gvdb_library”

Thanks,
Rama

Hi Rama - I definitely installed from the Win installer, and the failures come when there are spaces in paths in the command line parameters to the CUDA compiler as it processes .cu files during a Visual Studio build of the CMake-generated VS Solution on Windows 10. The spaces issue has been with us a long time - I wonder over time how many people (like me) have cloned their CUDA tree out of “NVidia Corporation” and into something with no spaces in the path! When I get a moment I’ll get you an exact repro - clearly there’s something different in our two environments. (There’s a separate issue too about why in stark contrast to everyone else NVidia puts everything into ProgramData, but you and I aren’t going to fix that!!)

Hi Rama, Thanks for this excellent work on GVDB. We’ve been playing with it a bit and it has great potential. As Tommy mentioned earlier, I’m also interested in any roadmap, timeframe, and/or list of planned features you could provide. For us, we’re curious about expanded OpenVD support beyond version 3.0, and the ability for multiple GVDB objects (i.e. not just one singleton instance).

Thanks again and keep-up the good work. Andrew