Text to hex, two digits to a byte

Four settings differ between tools and this page hands you all four. Capitals or lowercase for the digits a to f, a 0x or backslash-x in front of each byte or nothing at all, the character sitting between bytes, and above all the encoding — UTF-8 and UTF-16 produce genuinely different bytes for the same sentence, not just a different layout. Hello is 48 65 6c 6c 6f in UTF-8 and 00 48 00 65 00 6c 00 6c 00 6f in UTF-16 big-endian. Match a tool you have to copy by setting the encoding first and the cosmetics second.

Two per byte, and the number of bytes depends on the character and the encoding. In UTF-8 every character of unaccented English is one byte and therefore two hex digits, an accented letter such as é is two bytes and four digits, a Chinese, Japanese or Korean character is normally three bytes and six digits, and every emoji is four bytes and eight digits. UTF-16 works the other way around: two bytes and four digits for almost everything, four for emoji. That is the whole reason a 10-character sentence can produce 40 hex digits without anything being wrong.

It is a label saying the digits that follow are base 16, and it is not part of the value — 0x48 and 48 are the same byte. The notation came from the C language in the early 1970s and spread from there into most programming languages, which is why hex copied out of source code so often arrives wearing it. Other traditions mark the same thing differently: assemblers write 48h, HTML colours use a leading #, and URLs use a per-byte %. The decoder here strips 0x, backslash-x, U+ and # before it looks at anything, so pasted code needs no cleaning up first.

A digit went missing, because a byte is always exactly two hex digits and an odd count cannot be split. Two causes account for nearly all of it: a leading zero dropped by hand or by a spreadsheet, so 0f was written as f, or a character lost when the string was copied across a line break. The converter names the total it counted so you can work out where the gap is. If the digit count is even and it still fails, the problem is a stray character rather than a missing one, and that is reported by position.

No — this page converts characters, and hex to decimal is arithmetic on a plain number. The distinction matters because one string answers both questions differently: the byte 41 spells the letter A when read against a character table, and counts as 65 when added up as base 16. If what you are holding is a quantity rather than a message — a colour value, a memory address, a byte count — the number panel on the binary translator does that conversion at any length. Everything on this page treats the digits as text.

Only the order of the two bytes inside each character, but get it wrong and the text comes back as unreadable Chinese characters rather than as a warning. Big-endian writes the high byte first, so the letter A is 00 41; little-endian writes the low byte first, giving 41 00. Neither is more correct — it is a hardware convention, and Intel and AMD processors are little-endian while network protocols traditionally are not. A file often announces its choice with two bytes at the very start, FE FF or FF FE, called a byte order mark, and this page tells you when it spots one instead of quietly decoding it as a character.

You can, but the result is meaningless, and it is worth knowing why. A colour code is three numbers side by side — ff is 255 units of red, 66 is 102 of green, 00 is none of blue — and they were never characters to begin with. Decoding them as UTF-8 produces a broken-byte error, because ff cannot start a valid UTF-8 character. The bytes only mean something as text when they were written as text in the first place.