B64Base64 Encode/Decode

How it works

Base64 groups the input into 3-byte (24-bit) chunks and slices each chunk into four 6-bit pieces. Six bits hold a value from 0 to 63, which maps neatly onto a 64-character alphabet (A–Z, a–z, 0–9, +, /). Three bytes become four characters, so the output is 4/3 the size of the input — about 33% larger.

When the input length isn't a multiple of three, the final group is short and the missing positions are filled with '=' characters. That's why Base64 strings end with zero, one, or two equals signs: one leftover byte produces '==', two leftover bytes produce '='.

'H' = 0x48 = 01001000
'i' = 0x69 = 01101001

Concatenate bits:  01001000 01101001
Slice into 6 bits: 010010 000110 1001(+00 pad)
Decimal:           18     6      36
Alphabet:          S      G      k
Pad the short group with '=' → SGk=

How to use it

1. Paste the text you want to encode, or the Base64 string you want to decode.
2. Choose the Encode or Decode direction.
3. Copy the result. When decoding, invalid Base64 input produces an error message.
4. If the output goes into a URL or filename, read the URL-safe section below first.

Standard Base64 versus URL-safe Base64

The + and / characters used by standard Base64 both mean something special in URLs: + can be read as a space in a query string, and / is a path separator. RFC 4648 therefore defines a 'base64url' variant that swaps those two characters for - and _. JWTs use this variant, and they also drop the = padding.

Why non-ASCII text breaks

Base64 is a specification about bytes. Turning characters into bytes is the job of the layer beneath it — character encoding. To Base64-encode 'héllo', you first encode it as UTF-8 to get the actual bytes, then Base64 those bytes.

The browser's legacy btoa() skips that step and treats each character as a single-byte code point, so anything above U+00FF — CJK text, emoji — throws an InvalidCharacterError. This tool runs input through TextEncoder to get UTF-8 bytes first, so Korean, Japanese, and emoji all work.

If a decode produces mojibake instead, the original bytes were probably encoded with something other than UTF-8, such as EUC-KR, CP949, or Latin-1.

// encode
const bytes = new TextEncoder().encode("héllo");
const b64 = btoa(String.fromCharCode(...bytes));

// decode
const raw = Uint8Array.from(atob(b64), c => c.charCodeAt(0));
const text = new TextDecoder().decode(raw);

Where you'll encounter it

• Data URIs: embedding a small icon directly in CSS or HTML to save a request
• HTTP Basic auth: the 'Basic dXNlcjpwYXNz' in an Authorization header is Base64 of user:pass
• JWTs: header and payload are base64url-encoded and joined with dots
• Email attachments: MIME assumes a 7-bit channel, so binary attachments are Base64'd
• Putting an entire certificate or key file on one line in a config file or env var
• Comparing HMAC results as text when verifying webhook signatures

A performance caveat

Embedding images as data URIs saves an HTTP request but often costs more than it saves: the payload grows 33%, the browser can no longer cache the asset separately, and the containing CSS or HTML file gets bigger, delaying first render. As a rule of thumb, icons under 1–2KB are reasonable to inline; anything larger belongs in its own file with cache headers.

Frequently Asked Questions

Is Base64 encryption?

No. There is no key and anyone can reverse it instantly, so it provides no security whatsoever. A Base64-wrapped password should be treated as plaintext — this is exactly why HTTP Basic auth is considered unsafe without TLS. Use real encryption like AES to protect a value, or a hash if it should not be reversible.

What do the trailing = characters mean?

They're padding, added when the input byte count isn't a multiple of three so the final four-character block is complete. One leftover byte yields '==', two leftover bytes yield '='. Some implementations decode fine without padding, but strict parsers require it, so don't strip it arbitrarily.

My decoded output is gibberish. Why?

Check three things. First, the original may be binary — an image or archive — rather than text. Second, it may use a character encoding other than UTF-8, such as EUC-KR or Latin-1. Third, the string may be the base64url variant decoded as standard Base64, or vice versa; if you see - or _ characters, it's base64url.

Can a long Base64 string contain line breaks?

The MIME spec (RFC 2045) actually mandates a line break every 76 characters, and most decoders ignore whitespace and newlines. In strict contexts like JWTs, however, an inserted newline invalidates the token.

Does Base64 compress data?

The opposite — it turns 3 bytes into 4 characters, growing the payload by about 33%. If you need compression, gzip first and then Base64 the compressed bytes. Doing it the other way round performs poorly, because Base64 output has high entropy and compresses badly.

Can I encode a file?

This tool takes text input. For turning an image into a data URI, the Image to Base64 tool is more convenient. On the command line, use base64 -i file.png on macOS or base64 -w 0 file.png on Linux.