search

yihua7 / SC-GS

[CVPR 2024] Code for SC-GS: Sparse-Controlled Gaussian Splatting for Editable Dynamic Scenes
https://yihua7.github.io/SC-GS-web/
MIT License
427 stars 21 forks source link

Occur an error when install requirements.txt #10

Closed irisfreesiri closed 3 months ago

irisfreesiri commented 4 months ago

error: command 'D:\Program Files\CUDA\v11.8\bin\nvcc.exe' failed with exit code 1 ERROR: Failed building wheel for pytorch3d

I've installed CUDA 11.8. When I run the 3D Gaussian Splatting code (https://github.com/jonstephens85/gaussian-splatting-Windows), everything runs well. However, when I installed requirements.txt, I encountered an error. I've tried on my own laptop and the lab's desktop, but both failed.

What else do I need to provide?

yihua7 commented 4 months ago

Hi,

  1. The problem is caused by the failure installation or the incompatibility of pytorch3d.
  2. You may try to build and install pytorch3d from source following the instruction in here by running: 
    # For Ubuntu
git clone https://github.com/facebookresearch/pytorch3d.git
cd pytorch3d && pip install -e .
yihua7 commented 4 months ago

Hi Does the above help? Any further problems?

irisfreesiri commented 4 months ago

Hi, Thank you for asking. It doesn't work. Which cuda version should I install?

[cid:77c440ea-80a7-4bc9-b437-48a7308df48e]

From: Yihua @.> Sent: Thursday, March 7, 2024 3:19 AM To: yihua7/SC-GS @.> Cc: Iris @.>; Author @.> Subject: Re: [yihua7/SC-GS] Occur an error when install requirements.txt (Issue #10)

Hi Does the above help? Any further problems?

— Reply to this email directly, view it on GitHubhttps://github.com/yihua7/SC-GS/issues/10#issuecomment-1980975176, or unsubscribehttps://github.com/notifications/unsubscribe-auth/A5SK6SI6BKIXSMK34AVG2T3YW4QV3AVCNFSM6AAAAABEGB7NP6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBQHE3TKMJXGY. You are receiving this because you authored the thread.Message ID: @.***>

yihua7 commented 4 months ago

I am sorry to hear that. Is your CUDA installation compatible with the required version of PyTorch? It is important to ensure that CUDA and pyTorch are compatible with each other to successfully build pyTorch3D.

irisfreesiri commented 4 months ago

So which version of CUDA and PyTorch you recommend?

On Thu, Mar 7, 2024 at 3:54 PM Yihua @.***> wrote:

I am sorry to hear that. Is your CUDA installation compatible with the required version of PyTorch? It is important to ensure that CUDA and pyTorch are compatible with each other to successfully build pyTorch3D.

— Reply to this email directly, view it on GitHub https://github.com/yihua7/SC-GS/issues/10#issuecomment-1982244017, or unsubscribe https://github.com/notifications/unsubscribe-auth/A5SK6SNRD6X4MU4PY3JWMDDYW7JGRAVCNFSM6AAAAABEGB7NP6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBSGI2DIMBRG4 . You are receiving this because you authored the thread.Message ID: @.***>

yihua7 commented 4 months ago

Since you already installed CUDA 11.8, I recommend the following version of pytorch to install:

# CUDA 11.8
pip install torch==2.2.0 torchvision==0.17.0 torchaudio==2.2.0 --index-url https://download.pytorch.org/whl/cu118

Then you can move on to build and install pytorch3d locally. Hope this helps. : )

irisfreesiri commented 3 months ago

Hi, thank you for your help. It's already been addressed. I followed these steps that you sent me the link: [image: image.png]

  1. conda create -n pytorch3d python=3.8

  2. conda activate pytorch3d

  3. conda install pytorch=1.13.0 torchvision pytorch-cuda=11.6 -c pytorch -c nvidia

  4. conda install -c fvcore -c iopath -c conda-forge fvcore iopath

  5. pip install torch==1.12.1+cu113 torchvision==0.13.1+cu113 --extra-index-url https://download.pytorch.org/whl/cu113

  6. pip install "git+https://github.com/facebookresearch/pytorch3d.git"

  7. pip install -r requirements.txt [image: image.png]

It's finally successfully installed.

Thank you for your time and assistance. I really appreciate it!

On Thu, Mar 7, 2024 at 5:22 PM Yihua @.***> wrote:

Since you already installed CUDA 11.8, I recommend the following version of pytorch to install:

CUDA 11.8

pip install torch==2.2.0 torchvision==0.17.0 torchaudio==2.2.0 --index-url https://download.pytorch.org/whl/cu118

Then you can move on to build and install pytorch3d locally. Hope this helps. : )

— Reply to this email directly, view it on GitHub https://github.com/yihua7/SC-GS/issues/10#issuecomment-1982322022, or unsubscribe https://github.com/notifications/unsubscribe-auth/A5SK6SONWZYT3RVZVSP57XTYW7TRLAVCNFSM6AAAAABEGB7NP6VHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBSGMZDEMBSGI . You are receiving this because you authored the thread.Message ID: @.***>