pKrime / brignet

Automatic rigging using neural network from RigNet
GNU General Public License v3.0
407 stars 39 forks source link

Some modules not found, please check bRigNet preferences #7

Closed mitchellgordon95 closed 2 years ago

mitchellgordon95 commented 3 years ago

I am not sure what this step means: "Enable bRigNet in the blender addons, the preferences will show up. Set the Modules path properties to the RigNet environment from the previous step."

I have tried setting this to:

C:\Program Files\Blender Foundation\Blender 2.91\2.91\python C:\Program Files\Blender Foundation\Blender 2.91\2.91\python\bin C:\Program Files\Blender Foundation\Blender 2.91\2.91\python\lib

C:\Users\mitch\AppData\Roaming\Blender Foundation\Blender\2.91\scripts\addons\brignet-main C:\Users\mitch\AppData\Roaming\Blender Foundation\Blender\2.91\scripts\addons\brignet-main\RigNet

But none of them seem to resolve the error.

xiaochuan-li commented 3 years ago

same problem, have you resolved it ?

pKrime commented 3 years ago

Hi, I have not fixed those issues yet: the silver bullet would be rewriting RigNet and converting the model so that no additional dependencies are required, but I had no time during the past months.

the user Eritojai, has shared how he made it work on this blenderartist thread

Before Blender / setting env thing :

  • Install Python 3.7.7 (and dont forget using Blender 2.90 later, because i use 2.83 previously)
  • Check cuda version in powershell with command nvcc --version
  • Then proceed to install like Rignet Setup installion, but i change it to pytorch 1.7.0 instead and other torch libraries adjusted accordingly Edit : i’m using open3d==0.11.2

conda install pytorch torchvision torchaudio cudatoolkit=10.2 -c pytorch pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.7.0+cu102.html pip install torch-sparse -f https://pytorch-geometric.com/whl/torch-1.7.0+cu102.html pip install torch-spline-conv -f https://pytorch-geometric.com/whl/torch-1.7.0+cu102.html pip install torch-geometric

Going to Blender : After installing the addon, and i already make sure the Rignet work fine after running the Quick_start.py like stated in Rignet Github tuts , And Following all the steps in bRigNet github tutorial then i install the pytorch 1.7.0 too , to blender with

.\bin\python -m pip install torch===1.7.0 torchvision===0.8.1 torchaudio===0.7.0 -f https://download.pytorch.org/whl/torch_stable.html

and dont forgot all the libraries too

.\bin\python -m pip install torch-scatter -f https://pytorch-geometric.com/whl/torch-1.7.0+cu102.html

and other 4 (sparse, spline-conv, Cluster, Geometric )

did any of that help?

fire commented 3 years ago

Is there some way you can package a self contained blender archive? It's very difficult to get setup.

pKrime commented 3 years ago

Hi,

I tried, but I ended up with about 2GB in dependencies, and it's not even guaranteed to work due to differences in system environment: if virtualenv+pip don't work, there are reasons to think that my packaging wouldn't either.

Hang on, I'm planning to be back on this tool as soon as possible

fire commented 3 years ago

2GB uncompressed isn't that large in the large scheme of things. Text and binary code is compressible. It can be handled by Google Drive and Github releases.

pKrime commented 2 years ago

Hi,

the main issue with a prepackaged bundle is that you might need different builds of pytorch, depending on your system anc cuda version.

I have added an autoinstall button in the latest release, I'm afraid it's the best I can do for now.

I will look for more efficient solutions in the future

Cheers, Paolo