Closed rfeinman closed 4 months ago
Hi, Thanks for your issue! We just upload nksr-cuda and you could try it again. And for the nanoVDB, we are working on fixing this.
Thank you! I ran into a few more issues during install that I thought I'd point out.
openvdb/fvdb/fvdb/include
. However I find that this include directory is not automatically generated by the fvdb install. I had to copy it over manually to the expected location.We update openvdb/fvdb, could you try to git pull and install again? Thanks!
Yes, the nanovdb issue is solved now, thanks! The other 2 fvdb issues still persist (c++20 and missing include dirs) but I was able to figure out work-arounds and I'm sure others will too. Thanks for your quick feedback!
@rfeinman Could you clarify a bit more on which files you copied into folder openvdb/fvdb/fvdb/include? I'm finally able to compile the fvdb, and able to import fvdb in python. But I am not able to compile nksr-cuda, with the following error (looking like it's related to include directory you mentioned):
XXX/XCube/ext/nksr-cuda/csrc/meshing/functions.h:3:10: fatal error: fvdb/GridBatch.h: No such file or directory 3 | #include <fvdb/GridBatch.h> | ^~~~~~~~~~~~~~~~~~ compilation terminated.
Hi @yshen47, I will also have a look to fix it.
Hi @yshen47
I actually just wrote an explanation in issue #2 here: https://github.com/nv-tlabs/XCube/issues/2#issuecomment-2189035531
Does this solve it for you?
yea, i just figured out myself too, thanks so much!
Thanks for the awesome paper and code!
Your README conda environment setup references a directory
ext/nksr-cuda
which does not exist. Where can I find the NKSR package? The pip packages from the NKSR github are built for CUDA 11 and are incompatible with this environment.Also, just FYI: the fvdb setup.py script references a github repository (NVIDIA-Omniverse/NanoVDB) that does not exist, or that at least is not public.
Just a few small fixes and the code should be runnable. Thanks!