Roop

[Ghofx]

Select Topic Area

Question

Body

Please can anyone help me out in fixing and getting pass this on Roop program?
Please i need help

C:\Users\Cletus\Downloads\faceswap\roop>python run.py
Traceback (most recent call last):
File "run.py", line 3, in
from roop import core
File "C:\Users\Cletus\Downloads\faceswap\roop\roop\core.py", line 20, in
import roop.ui as ui
File "C:\Users\Cletus\Downloads\faceswap\roop\roop\ui.py", line 12, in
from roop.face_analyser import get_one_face
File "C:\Users\Cletus\Downloads\faceswap\roop\roop\face_analyser.py", line 7, in
from roop.typing import Frame, Face
File "C:\Users\Cletus\Downloads\faceswap\roop\roop\typing.py", line 7, in
Frame = numpy.ndarray[Any, Any]
TypeError: Type subscription requires python >= 3.9

C:\Users\Cletus\Downloads\faceswap\roop>

[davevad93]

Hi @Ghofx ,

That error indicates that the numpy.ndarray class used in this program requires at least Python 3.9, which hints that you have an older version installed on your system. Upgrading Python to a newer version should fix it, but keep in mind that it might break some programs you have that depend on older versions of Python, in that case you should install the newer version in a virtual environment. If you are sure that you don't have any application that depends on an older version of Python you can safely upgrade to 3.12 which is the latest version.

[Ghofx]

thank you, i was able to bypass that but saw this next please advice

(base) C:\deep>pip install onnxruntime-gpu==1.15.1
Requirement already satisfied: onnxruntime-gpu==1.15.1 in c:\users\cletus\miniconda3\lib\site-packages (1.15.1)
Requirement already satisfied: coloredlogs in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (15.0.1)
Requirement already satisfied: flatbuffers in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (24.3.25)
Requirement already satisfied: numpy>=1.21.6 in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (1.24.3)
Requirement already satisfied: packaging in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (23.0)
Requirement already satisfied: protobuf in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (4.23.4)
Requirement already satisfied: sympy in c:\users\cletus\miniconda3\lib\site-packages (from onnxruntime-gpu==1.15.1) (1.12)
Requirement already satisfied: humanfriendly>=9.1 in c:\users\cletus\miniconda3\lib\site-packages (from coloredlogs->onnxruntime-gpu==1.15.1) (10.0)
Requirement already satisfied: mpmath>=0.19 in c:\users\cletus\miniconda3\lib\site-packages (from sympy->onnxruntime-gpu==1.15.1) (1.3.0)
Requirement already satisfied: pyreadline3 in c:\users\cletus\miniconda3\lib\site-packages (from humanfriendly>=9.1->coloredlogs->onnxruntime-gpu==1.15.1) (3.4.1)
WARNING: There was an error checking the latest version of pip.

(base) C:\deep>python run.py --execution-provider cuda
Traceback (most recent call last):
File "run.py", line 3, in
from modules import core
File "C:\deep\modules\core.py", line 14, in
import torch
File "C:\Users\Cletus\miniconda3\lib\site-packages\torch_init_.py", line 130, in
raise err
OSError: [WinError 127] The specified procedure could not be found. Error loading "C:\Users\Cletus\miniconda3\lib\site-packages\torch\lib\c10_cuda.dll" or one of its dependencies.

(base) C:\deep>

Im just trying to learn the roop program

[dom557]

The error message indicates that there's an issue with the CUDA setup for PyTorch. This often happens when the CUDA version installed on your system is incompatible with the version of PyTorch you're using. Here’s how you can address this issue:

Steps to Resolve the Issue:

1. Verify CUDA Installation:

   • Ensure that the CUDA toolkit is installed on your system and that the version is compatible with the version of PyTorch you are using. You can download CUDA from the NVIDIA website.

2. Install the Correct Version of PyTorch with CUDA Support:

   • Visit the PyTorch website and use the selector to choose the correct version of PyTorch compatible with your CUDA version. For example:

     conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch

   • Replace cudatoolkit=11.7 with the version that matches your CUDA installation.

3. Check Dependencies:

   • Ensure that all necessary dependencies for CUDA are installed. This can include NVIDIA drivers, cuDNN, etc. You can find the installation instructions for these on the CUDA toolkit installation guide.

4. Verify the DLL Path:

   • Ensure that the c10_cuda.dll file exists in the specified directory (C:\Users\Cletus\miniconda3\lib\site-packages\torch\lib\). If it does not, reinstalling PyTorch with the correct CUDA version should resolve this.

5. Reinstall PyTorch:

   • If the above steps do not resolve the issue, try uninstalling and reinstalling PyTorch:

     pip uninstall torch
     pip cache purge  # Clear pip cache to ensure a clean installation
     conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch

6. Check Environment Variables:

   • Ensure that your system environment variables are set correctly. This includes PATH and CUDA_HOME. You can set these variables in your environment or add them to your .bashrc or .bash_profile.

Example Commands:

# Install CUDA toolkit and PyTorch with matching versions
conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch

# Verify CUDA installation
nvcc --version

Additional Tips:

• Restart Your System: After making changes to the CUDA setup, it’s a good idea to restart your computer to ensure all changes take effect.
• Check GPU Availability: Ensure your GPU is correctly recognized by running:

  python -c "import torch; print(torch.cuda.is_available())"

By following these steps, you should be able to resolve the OSError: [WinError 127] related to c10_cuda.dll and successfully run the roop program with CUDA support. If you continue to face issues, ensure all components (PyTorch, CUDA, cuDNN, and NVIDIA drivers) are properly installed and compatible with each other.

[davevad93]

@Ghofx this could be for several reasons, there seems to be an issue with loading c10_cuda.dll needed for PyTorch. It could be a problem with your Conda environment, your GPU or with PyTorch installation I'm sorry but this is beyond my knowledge, maybe someone else here in the discussions can help you better regarding this.

[Ghofx]

hello, thanks for helping out, i tried everything as stated above and still got this error message please

(base) C:\deep>pip cache purge
Files removed: 306

(base) C:\deep>conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch
Channels:

• pytorch
• defaults
  Platform: win-64
  Collecting package metadata (repodata.json): done
  Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

• cudatoolkit=11.7*

Current channels:

• https://conda.anaconda.org/pytorch
• defaults

To search for alternate channels that may provide the conda package you're
looking for, navigate to

https://anaconda.org

and use the search bar at the top of the page.

(base) C:\deep>conda install pytorch torchvision torchaudio cudatoolkit=11.7 -c pytorch
Channels:

• pytorch
• defaults
  Platform: win-64
  Collecting package metadata (repodata.json): done
  Solving environment: failed

PackagesNotFoundError: The following packages are not available from current channels:

• cudatoolkit=11.7*

Current channels:

• https://conda.anaconda.org/pytorch
• defaults

To search for alternate channels that may provide the conda package you're
looking for, navigate to

https://anaconda.org

and use the search bar at the top of the page.

(base) C:\deep>nvcc --version
nvcc: NVIDIA (R) Cuda compiler driver
Copyright (c) 2005-2022 NVIDIA Corporation
Built on Wed_Sep_21_10:41:10_Pacific_Daylight_Time_2022
Cuda compilation tools, release 11.8, V11.8.89
Build cuda_11.8.r11.8/compiler.31833905_0

(base) C:\deep>

[dom557]

reasons why this error pop out :

1. CUDA Toolkit Version: The error suggests that your system is looking for CUDA toolkit version 11.7, but you have version 11.8 installed. PyTorch might not yet support CUDA 11.8.

2. PyTorch Installation: Since PyTorch doesn't seem to support CUDA 11.8, you'll need to install a different version of PyTorch that's compatible.

Here's what you can do:

• Check PyTorch Compatibility: Visit the PyTorch website to see if there's a version compatible with CUDA 11.8. If there is, follow the installation instructions.

• Alternative Channels: If you can't find a compatible version from the usual sources, try alternative channels or repositories.

• Manual Installation: If needed, you can manually install PyTorch from its source code to ensure compatibility with CUDA 11.8, but this requires more technical knowledge.

After installing a compatible PyTorch version, try running your program again. If you still have issues, feel free to ask for more help!