Mojibake: noun, borrowed from Japanese 文字化け (mojibake): Corrupt characters or letters, especially from display or transfer through an inappropriate character set or encoding.
With the tools on this page you can make mojibake of your own! Example strings you could try: “τɘχт ɡöες ɦєɾε”, «tête-à-tête», 「亂碼字符」
Interprets the input as UTF-8 characters, with the astral plane characters (code point ≥ U+10000) being four bytes, not a pair of three-byte surrogates. Interprets the bytes as if they were characters from the Windows-1252 code page, for many years the default code page and still considered the default by web browsers if no other encoding is specified.
Simplified details: UTF-16 is a format for storing text where each character is stored as two bytes, AA-BB, for example "a" is stored as 00-61 and "☂" as 26-02. This flips these bytes around to BB-AA, so that 00-61 "a" becomes character 61-00 "愀", and 26-02 "☂" becomes 02-26 "Ȧ". Most basic letters become random Chinese characters, most of them rarely-used; don't use this to "translate" your name to Chinese and get a tattoo! Rather interestingly, a space (00-20) just becomes a wider space (20-00, "en quad").
In most cases, you can copy the output of this function and feed it back in as input, and you'll get the original text back.
To simulate transport of UTF-8 text thru an email system that can only handle 7-bit ASCII text.
Like in URLs, non-ASCII characters and many ASCII characters too get percent encoded.
Stuff outside basic ASCII, and some basic ASCII characters with significance in HTML, get encoded with &-entities.