Text to Hex Converter

Hex is the notation every debugger, packet capture and memory viewer speaks, and each one punctuates its bytes differently. The Text to Hex Converter writes UTF-8 bytes as uppercase two-digit pairs and lets you choose the shape, so the output pastes straight into whichever tool you are feeding without a round of manual editing first.

Decoding is deliberately looser than encoding. Before anything is parsed the input drops every 0x prefix and every comma and whitespace character, which means a C array initialiser, a Wireshark column and a bare run of digits all decode as they are. What survives that cleanup has to be an even number of hex digits, and you get told when it is not.

Both directions travel through UTF-8. A Cyrillic or CJK character encodes as the 2 or 3 bytes it genuinely occupies rather than as one oversized value, and the strict decoder refuses to invent replacement glyphs for sequences that are not real text. No input is uploaded.

The Four Separator Options

• Space (FF AA) is the default, and the layout nearly every hex dump and protocol document uses.
• Comma (FF,AA) drops into a spreadsheet column or a comma-separated literal with no further editing.
• 0x (0xFF 0xAA) prefixes each byte, which is the form C, Rust and most assemblers expect inside an array.
• None (FFAA) gives one unbroken string, the shape used by checksums, colour values and opaque identifiers.
• The separator row is visible in encode mode only. Decoding accepts all four at once, so you never have to declare which one you pasted.

What Happens When the Hex Is Malformed

• An odd digit count returns Hex string length must be even. Losing one character while selecting a dump is the usual cause.
• Anything outside 0-9 and A-F after cleanup returns Input contains non-hex characters. A mistyped g, or a | column divider copied from a hex editor, both land there.
• Valid hex that is not valid text returns These bytes aren't valid UTF-8 text., so FF FE is refused instead of being shown as two question marks.
• Case is ignored coming in and normalised to uppercase going out, so 48 65 6c and 48 65 6C both decode to Hel.
• The 0x strip is global, which means a literal 0x intended as data vanishes with the prefixes. That is the price of pasting real source code unedited.