This kind of thing was very common in MSDOS back when I was learning to code. There were a set of basic hardware screen modes that could be accessed through simple assembly instructions.
For those supporting ASCII character sets, it was possible to modify the actual glyph images for the characters in memory. Each available character was a simple bitmap, which could be modified in memory to look however you want.
For example, in colour screen modes (e.g. CGA, EGA), the text could be rendered via a specific memory area where one byte specified the character and the other byte specified a colour (foreground and background as 4-bits each for EGA). You just write the values in there and they show on screen. If you had modified your character table, then the modified characters would show.
I can't believe I remembered this just now, but the memory address A000
is apparently burned into my brain. You can read more at https://en.wikipedia.org/wiki/Enhanced_Graphics_Adapter
Of course, there are higher screen modes that support writing actual pixels, and that's how we used to write games. Famously the 320x200 VGA mode was super easy and fast to get "realistic" and smooth animated graphics in mind-blowing 256 colours without worrying about bit planes or anything.
And this is essentially what's happening at the low level. BIOS support for these forgotten graphics modes has existed for a very long time, but modern hardware has abstracted us away from such modes and we now tend to communicate directly with graphics cards via a driver or OS abstractions.