How to upscale textures using the ESRGAN neural network?

Question

Textures

Good afternoon!

There is such a thing on github - https://github.com/Calinou/quake2-neural-upscale/
Textures for Quake 2, improved with the help of a neural network.

"Quake 2 textures upscaled using ESRGAN (old-arch tag) and the Manga109Attempt model."

I'm looking for a way to increase the textures of two official mods for Quake 2 - Ground Zero and The Reckoning in the same way.

Can anyone explain how to set everything up yourself?
How to use ESRGAN and Manga model to get texture B from texture A?

I am a web programmer and linuxoid (debian), so I think I will master this task, but I don’t know how to approach it yet. In terms of basic concepts - what is a Manga model, how does it connect to a neural network, does this require a modern quantum computer, etc.

Answer 1

Wrong question category, okay.
You need to install libraries and python 3.7
and also don’t forget about c++ because without a library in c++ it won’t work
https://dtf.ru/science/37841-uluchshaem-tekstury-s...
But I recommend the old one model because there are already scientific models of behavior.
If you still cannot compile, I can throw off the compiled version, but the ana is modified and with an interface.
also this miracle eats a lot of RAM. and specifically on 100x100 about 2 GB.
Therefore, it is better to go directly to Google servers