zigimg VS ImageSharp

zigimg continues to get improvements, with support for JPEG, PNG, QOI, TGA and more image formats.

You can use std.mem.eql. Dunno what you are trying to do with an PNG image but maybe my library can be useful? https://github.com/zigimg/zigimg

See https://github.com/SixLabors/ImageSharp/discussions/2151 for previous discussions on the subject.

Looks like everyone is using bloated Skia library. Why not use ImageSharp?

I use this library for image handling, it's pretty good https://github.com/SixLabors/ImageSharp

I'm using ImageSharp for loading the texture, but I don't imagine that's where the problem arises. It's been working fine for the previous chapters, loading different, albeit smaller, textures.

From the release notes here: https://github.com/SixLabors/ImageSharp/releases/tag/v2.0.0

This probably won't fit your needs, but ImageSharp is very interesting imo. You can learn a lot of low level and optimization stuff there (SIMD, pointers/memory hackery, etc.).

You could use something like ImageSharp to load the image, scan over it's pixels, create the new image and save it.

While there are scenarios where using a native library certainly makes sense, that doesn't come for free. Using native code adds overhead (in many cases, having a full implementation in C# can be faster than invoking a C++ assembly), and it kills your portability, and makes maintenance harder. There are libraries out there, such as ImageSharp that are great examples of how you can have a fully managed implementation of a native image processing library that has speed that are on the same level (or actually, faster in some cases) than native libraries, with also the advantage of being able to run everywhere.