search

deepfakes / faceswap

Deepfakes Software For All
https://www.faceswap.dev
GNU General Public License v3.0
52.55k stars 13.24k forks source link

An update for current python versions [Linux] #1395

Closed XDM-Inc closed 4 months ago

XDM-Inc commented 4 months ago

because of the updating of python the linux version no longer works because it is asking for the older version of python.

torzdf commented 4 months ago

Don't use system Python. Use the Linux installer and/or Conda if you want to install manually and be able to run different versions of Python

XDM-Inc commented 4 months ago

Don't use system Python. Use the Linux installer and/or Conda if you want to install manually and be able to run different versions of Python

i tried followed the instructions from the sh file. you cant get away from the python requirement. i also cant make it select different versions of python that i have installed as it only wants to use the system one. and i cant downgrade the python on my system as its a system app that other system apps depend on.

torzdf commented 4 months ago

you cant get away from the python requirement.

This is just plain incorrect. You can and we do. We use Conda specifically because it allows you to use different versions of Python, and manage different local Cuda installs.

If you want to run a newer version of Python, then you are going to need to take it up with Tensorflow, as the last version that supports Windows, natively, is 2.10. The highest version of Python that this supports is 3.10.

Ultimately, I am a Linux user myself. My system Python is version 3.12. I also run (and develop) faceswap. This runs, in Conda, on version 3.10.

We will only provide support for people who follow the standard install procedure and manage their faceswap install through Conda.

That is not to say that you can't run it with system Python, just that we will not support it.

torzdf commented 4 months ago

If you want to discuss this further, you are welcome to engage with me on Discord, where I replied to your message. Ultimately this issue is closed because it is not a bug.