Emoji to Unicode, one code point at a time

Unicode gives each base emoji one code point. Everything beyond the base, a skin tone, a gender, a second person, a flag, gets built by stacking extra code points behind the base. Your phone or browser reads the stack and draws one picture. If the font has no picture for the stack, you see the parts in a row instead.

The breakdown cards in the converter color each piece by its job, using the key above the cards. Load sample fills the box with a short line holding seven different cases. The 👩🏽‍💻 card lists four code points: U+1F469 woman, U+1F3FD medium skin tone, U+200D joiner, U+1F4BB laptop. Delete the joiner and most systems draw two separate emoji, a woman and a laptop.

The England flag on the same line is the heaviest case: seven code points, 28 bytes, for one small flag.

The stats strip under the converter shows all four counts for whatever you paste. Three places where the gap bites:

Form limits

Browsers measure maxlength in UTF-16 units. A field set to maxlength="10" accepts only five 😀, and one family emoji uses eight of the ten slots.

Database columns

MySQL sizes VARCHAR(n) in code points. A column of VARCHAR(20) holds twenty plain smileys but only four family emoji. The older utf8 charset (an alias for utf8mb3) stores at most 3 bytes per character, so every emoji above U+FFFF fails with Incorrect string value: '\xF0\x9F\x98\x80'. Use utf8mb4 for the column, the table and the connection.

Truncation

Cutting a string with substring(0, n) in JavaScript or Java splits surrogate pairs and leaves a broken half. Cut on visible characters with Intl.Segmenter instead.

The highlighted JSON row trips people up most. JSON has no \u{1F600} form, and strict parsers throw an error. Write the surrogate pair, or skip escaping and save the raw emoji, since JSON files are UTF-8 by default. PHP's json_encode writes the surrogate pair unless you pass JSON_UNESCAPED_UNICODE.

The CSS escape is padded to six digits on purpose. A shorter \1F600 followed by a letter from a to f would swallow the letter as part of the number.

With Leave letters and digits readable ticked, a line like Hi 😀 becomes Hi \u{1F600}, ready to drop into a string literal. Quotes, backslashes and angle brackets get escaped either way, so the output never closes your string or tag early.

😀Garbled Latin letters

Mojibake. The four UTF-8 bytes of 😀 (F0 9F 98 80) got read as Windows-1252, one letter per byte. The data is fine. The reader picked the wrong encoding. Send Content-Type: text/html; charset=utf-8, or reopen the file as UTF-8.

????Question marks after saving

The database column or connection uses a 3-byte charset. In MySQL, convert the table to utf8mb4 and run SET NAMES utf8mb4 on the connection. Text already saved as question marks stays lost.

PKTwo letters instead of a flag

Windows ships no country flag pictures in its emoji font, so regional indicator pairs show as letters. The code points are correct. Open the same text on a phone and the flag appears.

👨 👩 👧A family split into people

Either the font lacks the combined picture, or something stripped the U+200D joiners. Paste the text in and check the card. No joiner in the chain points to a sanitizer or an old storage layer.

❤︎A black outline heart

The sequence is missing U+FE0F, so some apps draw the text form. Add U+FE0F after U+2764 to request the red emoji.

�A diamond with a question mark

A string got cut between the two halves of a surrogate pair, then encoded. Code to emoji mode flags a lone half such as \uD83D in red with a short reason.

1. No emoji names. Cards show the Unicode block and the job of each piece, not names such as "grinning face". Use the Unicode Character Finder to look a name up.
2. No check against the recommended list. Any chain of code points decodes, including combinations no vendor draws as one picture. Your device decides what shows up.
3. Display follows your fonts. Unicode adds new emoji each September and phone makers ship them months later. A brand new emoji shows as a box on an older device even when every code point is right.
4. Bare hex means code points. 2764 FE0F decodes to a heart. A run of two-digit values such as F0 9F 98 80 decodes as UTF-8 bytes only when the run forms valid UTF-8. Prefix bytes with \x or % to remove doubt. Spaces between U+ codes or bare hex count as separators, so write U+0020 for a real space. Spaces between escapes such as \u{1F600} stay in the result.
5. Cards stop at 120 characters. The output box and the counters still cover the full text.
6. Text stays in your browser. Conversion runs in JavaScript on this page. Copy link puts your text in the URL, so skip the link for anything private.