Rendering the result in Videos using Pytorch3d

Question

from lib.colab_util import generate_video_from_obj, set_renderer, video

renderer = set_renderer() generate_video_from_obj(obj_path, out_img_path, video_path, renderer)

we cannot play a mp4 video generated by cv2

!ffmpeg -i $video_path -vcodec libx264 $video_display_path -y -loglevel quiet video(video_display_path)

ImportError Traceback (most recent call last) in () ----> 1 from lib.colab_util import generate_video_from_obj, set_renderer, video 2 3 renderer = set_renderer() 4 generate_video_from_obj(obj_path, out_img_path, video_path, renderer) 5

4 frames /usr/local/lib/python3.7/dist-packages/pytorch3d/renderer/blending.py in () 7 8 # pyre-fixme[21]: Could not find name _C in pytorch3d. ----> 9 from pytorch3d import _C 10 11

ImportError: /usr/local/lib/python3.7/dist-packages/pytorch3d/_C.cpython-37m-x86_64-linux-gnu.so: undefined symbol: _ZNK2at6Tensor7is_cudaEv

---

Answer 1

I found the same problem for Google Colab. Then I solved it from this documentation https://github.com/facebookresearch/pytorch3d/blob/main/INSTALL.md

In general, from inside IPython, or in Google Colab or a Jupyter notebook, you can install with

import sys
import torch
pyt_version_str=torch.__version__.split("+")[0].replace(".", "")
version_str="".join([
    f"py3{sys.version_info.minor}_cu",
    torch.version.cuda.replace(".",""),
    f"_pyt{pyt_version_str}"
])
!pip install --no-index --no-cache-dir pytorch3d -f https://dl.fbaipublicfiles.com/pytorch3d/packaging/wheels/{version_str}/download.html

You have to restart your runtime environment after installing this. In my case, it worked perfectly in Google Colab with GPU instance.

Answer 2

The users is running, the PIFuHD collab.

Try, the workaround provided by https://colab.research.google.com/drive/11z58bl3meSzo6kFqkahMa35G5jmh2Wgt?usp=sharing

# If you get an error in the next cell, you can instead try the following command (don't forget to comment out the one above!).
# Note that this error is caused by inconsistent cuda version for the pytorch3d package and pytorch in Colab environment.
# Thus, this issue may persist unless pytorch3d in the pip package is updated with the cuda version consistent with pytorch in Colab.
# Also please be aware that the following command is much slower as it builds pytorch3d from scratch.

# !pip install 'git+https://github.com/facebookresearch/pytorch3d.git@stable'

# You can try another solution below as well. This is also slow and requires you to restart the runtime.

# !pip install 'torch==1.6.0+cu101' -f https://download.pytorch.org/whl/torch_stable.html
# !pip install 'torchvision==0.7.0+cu101' -f https://download.pytorch.org/whl/torch_stable.html
# !pip install 'pytorch3d==0.2.5'