Skip to content

Just some suggestions about setting up the environment #9

@lihaolin88

Description

@lihaolin88

Just a kind reminder for anyone experiencing similar difficulties setting up the environment:

  1. I highly recommend installing Python version 3.8 or higher on a Linux machine.
  2. Many problems can arise with PyTorch3D, especially when installing a new version of PyTorch. The original command "conda install pytorch3d -c pytorch3d" only works for "torch < 1.6." For newer versions of PyTorch, I recommend installing PyTorch3D manually from the GitHub repository.
  3. When installing Kaolin, it is crucial to keep your CUDA toolkit consistent with the Kaolin version. I recommend using the command "pip install kaolin==0.13.0 -f https://nvidia-kaolin.s3.us-east-2.amazonaws.com/torch-{your torch version, for example: 1.10.0}_cu{your cuda toolkit version, for example: 111}.html". Using the wrong version can cause a lot of problems.
  4. To avoid potential conflicts, I highly recommend creating a virtual environment with Conda for this project.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions