CESU-8 is non-standard UNICODE character encoding format close to UTF-8 with the exception that the characters points above U+FFFF are represented with UNICODE surrogate pairs encoded as 16 bit characters. Officially UTF-8 should contain those characters directly without using surrogate pairs as intermediate encoding.
CESU-8 defined by http://fileformats.archiveteam.org/:
CESU-8 is an inefficient Unicode character encoding related to UTF-8. It is not an accepted standard, but has been documented in the interest of practicality. It's what you get if you take UTF-16 data, reinterpret it as UCS-2, then convert it to UTF-8 (while ignoring any rules forbidding the use of code points in the range U+D800 to U+DFFF). A code point thus uses 1, 2, 3, or 6 bytes. It is sometimes used by accident, but may be used deliberately to accommodate systems that don't support 4-byte UTF-8 sequences, or when a close correspondence between UTF-16 and a UTF-8-like encoding is deemed necessary.
