tl;dr: I have no idea how to properly process those images. If you want to see the result, convert to bitmap (to kill the alpha channel) and floodfill the black with white.

Yes and I've used the .atlas files to separate them. However, I know nothing about graphics processing (heck, unstitching the images took me two days of googling for ancient code) so it's very possible I damaged something that might have helped. Even if I did, I wouldn't know what to do with that information.

Using Irfanview, I can apparently see the alpha channel. However, I know nothing about how to use it.

The best my seriously-out-of-date self can do is convert to bitmap, use a 26-year-old programming language to separate them and use Image Magick to floodfill the background to transparent and convert back to png. This works for some (if the image uses no black) but not many.

I guess I need someone that actually knows what they're doing.

Amusingly enough, I could use Irfanview to make a mask from the original and then I could use code to merge mask and image for display but not to save it. I could relatively easily create any combination of the images I wanted, including automating the generation of the thumbnails that I painstakingly cropped from screenshots, I just wouldn't be able to save them to anything but .bmp.