I have been playing around with tools that could enlarge/upscale images with better quality that the simple bicubic method. I don't need it often, but every once in a while a good upscale can save the day.
There are quite some tools out there that use neural networks to guesstimate new pixels, and most of the time the result is quite good; a lot better than I expected anyway.
The first hits in Google were for commercial solutions like https://letsenhance.io/
These work fine, but they turned out to be based on open source projects like https://github.com/nagadomi/waifu2x
- using those makes more sense. I've played around with a Windows port of waifu2x from https://github.com/DeadSix27/waifu2x-converter-cpp
and it worked like a charm.
However, it turned out that Adobe Photoshop CC 2018 has an experimental feature called Preserve Details 2.0, which does a very similar job. And it is built into Photoshop, for which I happen to have a license, so I'll be using that.
More info can be found at https://www.photoshopessentials.com/basics/upscale-images-photoshop-cc-2018/