Base64 is NOT a modern 64 bit technique, it’s a 6 bit (64 value) compatibility translation for 1960’s and ‘70’s computing!
This shouldn’t be something anyone has to come out and say, but repeatedly I come across people diving for base64 and claiming it’s some fancy modern “64 bit” way of embedding data… Utter nonsense when it was basically created for 8 bit data to be communicated across DarpaNet decades ago in a format that was safe from corruption on mainframes that only had 6 bit character spaces. It’s called “64” because it is built on a 6 bit / 64 possible value table of characters.
Where this delusion that it’s something “recent” comes from I have no clue, but I keep hearing it parroted over and over. I suspect this is like the misunderstanding some people have over the equally incorrect notion that the scam artists at W3Schools have anything to do with the W3C… they don’t. People see a couple letters or a number, and fill in the blanks themselves.
That it seems to have become the darling of people for whom “separation of concerns” means dick-all, throwing proper caching in the trash and slopping presentational images into their markup where they have zero blasted business being in the first place, is no real shock. Those who cannot grasp the most basic principles of HTML, CSS or caching models believing bald-faced lies like “putting above the fold style and static images in the markup speeds up pages” are most always first in line to fail to understand much of anything else.
Realistically you should avoid it if at all possible, leaving it for e-mail attachments where it’s basically required due to mail protocols having their craniums wedged up 1984’s rectum. It inherently increases file sizes 25%, is painfully slow to decode due to the need for the “shift/and” with bitwise tracking, and on the whole is a relic of the past best left there.