Convert Data URI to UTF-8

The generic decoder honours whatever charset the URI declares (UTF-8, Windows-1252, ISO-8859-1, etc.) – it converts the bytes using that charset. This tool always decodes as UTF-8 and additionally exposes a per-codepoint UTF-8 byte breakdown. Pick this one when you know the payload is UTF-8 and want to see the byte-level structure; pick the generic one when the source charset is legacy or unknown.

Because UTF-8 packs larger Unicode code points into more bytes. U+1F600 (😀) sits in the supplementary planes above U+FFFF, which requires UTF-8’s 4-byte form (F0 9F 98 80). The decoded string has 1 code point, but any string-length measurement that counts code units (JavaScript’s .length) will say 2, and the UTF-8 byte count is 4. The breakdown panel exposes all three numbers.

UTF-8 uses variable-width encoding: U+0000-U+007F is 1 byte (plain ASCII), U+0080-U+07FF is 2 bytes (Latin extended, Greek, Cyrillic, Arabic), U+0800-U+FFFF is 3 bytes (CJK, most BMP characters), U+10000-U+10FFFF is 4 bytes (emoji, supplementary scripts). The stats panel breaks down your payload by these classes.

The decoder hit bytes that aren’t valid UTF-8. Most common cause: the source was actually some other charset (e.g., Windows-1252) but this tool always assumes UTF-8. If you see lots of U+FFFD, try the generic Data URI to ASCII decoder with the original charset, or check whether the URI was truncated mid-sequence.

After decoding, check if the output starts with U+FEFF (a zero-width no-break space that serves as a byte-order mark). Tick the “Strip UTF-8 BOM” box to remove it automatically. In Base64, the BOM appears as the prefix 77u/ (which decodes to EF BB BF).

Yes. RFC 2397 technically specifies the standard Base64 alphabet (+ and /), but URL-safe Base64 (- and _) leaks into data URIs that were built for web contexts. This tool normalises both alphabets and auto-pads missing = characters.

Because UTF-8 bytes, UTF-16 code units (what JavaScript’s .length counts), and Unicode code points (Array.from(s).length) are three different things. For ASCII-only text they match. For anything else they diverge: “é” is 2 UTF-8 bytes but 1 code point and 1 code unit; “😀” is 4 UTF-8 bytes, 1 code point, but 2 UTF-16 code units. The breakdown panel shows all the math.

Percent-encoded multi-byte UTF-8 is handled correctly. data:,%E6%97%A5%E6%9C%AC (6 percent-escapes, 18 chars of payload) decodes to 日本 (2 code points, 6 UTF-8 bytes). decodeURIComponent is UTF-8-aware by default, so this just works.

You can, but you’ll get U+FFFD replacement characters wherever the bytes don’t form valid UTF-8. The tool flags that in stats. For lossless legacy-charset decoding, use the generic Data URI to ASCII decoder which honours the declared charset. This tool is UTF-8-only by design so you can see the byte structure clearly.

Yes. Decoding uses the browser’s native atob, decodeURIComponent, and TextDecoder. Nothing is uploaded, nothing is logged, no account is needed. Load the page once and the tool works offline indefinitely.