Base64 Encoder & Decoder for Text, Images, Data URI and JWT Payloads
Encode UTF-8 text and images to Base64 in your browser, and decode Base64 back to text or preview it as an image. Handles Chinese, emoji and multi-line input, but this is standard Base64 only: URL-safe variants and bare JWT segments need fixing first, and binary data decodes as garbled UTF-8.
What this Base64 tool handles — and what it does not
UTF-8 text in both directions
Typing text encodes it immediately: TextEncoder converts it to UTF-8 bytes before Base64, so Chinese, Japanese and emoji come out correct instead of throwing the way raw btoa does on non-Latin-1 characters.
Images become Data URIs
Switch to Image, pick a file, and it is read as data:image/...;base64,... . Paste that Data URI back into the Base64 box to preview it inline; an invalid source shows a fallback message instead of a broken image.
Standard Base64 only, no URL-safe mode
The - and _ characters are rejected, so JWT payloads and URL-safe tokens must be rewritten to + and / before decoding. Missing = padding is accepted, and spaces, tabs and line breaks are ignored.
Decoding assumes UTF-8
Base64 of binary data is decoded as UTF-8 text, producing replacement characters instead of an error. Only UTF-8 text round-trips cleanly, so images and other bytes cannot be recovered this way.
Runs locally, but large input can fail
Nothing is uploaded; text is converted on every keystroke. There is no size guard, and very large text can hit a call-stack error because all bytes are spread into btoa at once.
Base64 encoding and decoding FAQ
Does it support URL-safe Base64 (- and _)?
No. Only the standard + / alphabet works; - or _ throws InvalidCharacterError. Replace - with +, _ with /, then pad the string to a multiple of 4 before pasting it here.
Can I decode a JWT payload here?
Only after conversion. A JWT payload segment is URL-safe Base64 without padding, and this page neither splits a token nor adds missing characters. Paste a rewritten segment; the result is plain text, not parsed JSON.
Can I convert files to Base64 and back?
Only images, and only one direction for files: upload an image to get a Data URI. There is no text-file picker, and decoding never produces a downloadable file — the result is shown as text.
Why does Chinese or emoji work when btoa usually fails?
Because the code converts text to UTF-8 bytes with TextEncoder before calling btoa, then reverses it with TextDecoder. Calling btoa directly on 中文 throws InvalidCharacterError; this page does not.
What happens with binary (non-UTF-8) Base64?
It is decoded as UTF-8, so arbitrary bytes become replacement characters; nothing fails and no error is shown. Use the image preview or a dedicated binary decoder when the bytes must be preserved.
Is there a size limit, and does it upload anything?
Nothing is uploaded and there is no explicit limit; conversion runs synchronously on every keystroke. Large inputs can throw 'Maximum call stack size exceeded', and multi-megabyte text may freeze the tab.
How are padding, whitespace and invalid input treated?
Missing = padding is accepted, and atob ignores spaces, tabs and line breaks. A length of % 4 == 1, or any character outside the alphabet, throws InvalidCharacterError and the UI shows no error message.

