Re: Microsoft Visual C++ Redist error
Thanks for this, this is useful information. Hopefully others can confirm whether this fixes their issue.
The place to discuss Faceswap and Deepfakes
https://forum.faceswap.dev/
Thanks for this, this is useful information. Hopefully others can confirm whether this fixes their issue.
Has anyone found a solution?
funkytownn wrote: ↑Thu Oct 29, 2020 5:14 pmHad the same DLL-error for a while in all tensorflow-related installs.
It's because of having an older CPU (I have an i7 870) with no AVX2 support.To fix it use : https://github.com/fo40225/tensorflow-windows-wheel
I installed the cuda102cudnn76sse2 wheel with conda and Faceswap is working now!
First of all thank you for the instructions. Does the order of installing things matter. Can i first install faceswap and then cuda102cudnn76sse2? And how do i install it? Do i just need to install this cuda102cudnn76sse2 or all these what are there where this link you shared takes us? Thank you in advance.
Hi guys! Might have the solution. I thought i had the same issues; checked the anaconda prompt log, tensorflow stuff etc. went back to the visual C++ page https://support.microsoft.com/en-us/hel ... -downloads and downloaded " x64: vc_redist.x64.exe "
i realized earlier i had hastily downloaded the 86 one well now i can finally run faceswap!
sorry if this isn't the solution for u. hopefully the next fella doesn't make the same mistake i did
martmah wrote: ↑Fri Nov 06, 2020 8:20 amfunkytownn wrote: ↑Thu Oct 29, 2020 5:14 pmHad the same DLL-error for a while in all tensorflow-related installs.
It's because of having an older CPU (I have an i7 870) with no AVX2 support.To fix it use : https://github.com/fo40225/tensorflow-windows-wheel
I installed the cuda102cudnn76sse2 wheel with conda and Faceswap is working now!First of all thank you for the instructions. Does the order of installing things matter. Can i first install faceswap and then cuda102cudnn76sse2? And how do i install it? Do i just need to install this cuda102cudnn76sse2 or all these what are there where this link you shared takes us? Thank you in advance.
This issue tends to only happen on Windows, as we cannot currently use Conda to install Tensorflow under Windows, as they do not have a compatible release yet.
To install a custom Tensorflow wheel that does not require AVX, do the following:
Check if you CPU supports AVX/AVX2 instruction sets (google it). If it does, then this will not fix your issue and you will be wasting your time.
Download and install Faceswap as usual (if not done already). The below instructions assume that you installed with the default settings. If you changed your environment name at install time, then please update the commands accordingly.
Download this version of Tensorflow (you need all 3 parts downloaded to the same folder):
https://github.com/fo40225/tensorflow-w ... udnn76sse2
Open the file that ends with .001
with 7zip and drag the contents to somewhere easy to access. (NB: Only do this with the file that ends with .001.
7zip will automatically build the file from the other parts).
Start an Anaconda prompt by going Start
> Anaconda Prompt
(from the windows start menu)
Activate your faceswap environment with the following command (once done, the cli should have (faceswap)
displayed at the start:
Code: Select all
conda activate faceswap
The custom version of Tensorflow requires a different version of Cuda/cuDNN, so we need to remove our installed version, and install the correct version. We can do this with one command:
Code: Select all
conda install cudatoolkit=10.2
Remove the installed version of Tensorflow:
Code: Select all
pip uninstall tensorflow-gpu -y
Go to the location that you extracted your custom Tensorflow Wheel to in Windows Explorer
. Hold down shift
, right click the file and select Copy as path
Go back to the Anaconda Prompt
window and enter pip install
and paste the contents of the clipboard to the end of this line. (It should read something like pip install C:\<tensorflow_path\tensorflow-2.2.0-cp38-none-win_amd64.whl
). Hit Enter to run the command and install.
This should have replaced the version of Tensorflow you are using, and you should be good to go.