About this converter
Pikkola, tool 29. Written 18 November 2026
A converter is judged by what it says when the input is wrong.
Why another one
Base64 arrives in the middle of other work. A token in a log line, a payload in a webhook, an icon someone wants inline, a query string that lost its encoding on the way through a chat window. The task in front of you is never "base64"; it is the thing underneath, and the tool should be small enough to open without a decision.
UTF-8 is most of the job
The browser function that writes base64, btoa, takes a string of bytes and refuses anything above 255. Hand it Korean and it throws. That is where a great deal of broken text comes from: a converter either passes the string straight in and fails, or encodes one character per byte in latin1, so 한글 comes back as a row of odd symbols on the other side.
Here the order is fixed. The text becomes UTF-8 bytes first, then those bytes are written in the chosen alphabet. Coming back, the bytes are decoded as UTF-8 with errors set to fail, so a sequence that cannot be text is reported instead of being replaced by question marks and saved over your data.
Five alphabets, one byte string
Base64, URL-safe Base64, percent-encoding, hex and Unicode escapes carry the same bytes in different clothes. Base64 is for JSON fields, mail attachments and tokens. The URL-safe form exists because plus and slash mean something in a URL, and padding equals signs are unwelcome there. Percent-encoding is what the URL standard asks for in a parameter or a path. Hex is what a byte dump, a colour value or a packet log is written in. Escapes are for JavaScript strings, JSON and properties files.
Errors you can act on
A base64 string of the wrong length, a stray character, an odd number of hex digits, a percent sign without two digits behind it, bytes that are not UTF-8: each comes back as a sentence that names the character, in the pane that caused it. Silent repair is worse than a refusal, because the string that reaches the other end is then wrong in a way nobody notices until later.
Files stay in the tab
A dropped file is read with the browser's own FileReader and turned into a Data URI in memory. There is no upload, and the content security policy blocks requests to every other origin, so there is no route out even if a mistake were made. Files above two megabytes get a warning, because a Data URI that large makes the document holding it slow to open and slow to paint.
What it does not do
It does not encrypt. Base64 is not a lock and this page will not pretend otherwise; if the string has to stay secret, encrypt it first. It does not hash either, which is hash.pikkola.com. It keeps no history, no drafts and no account. There is no analytics code on the page. Between visits it keeps two things in local storage: the theme and the language.
For the curious
The conversion functions are attached to the page as window.PikkolaBase64, so you can check any result from the browser console, for example PikkolaBase64.encodeBase64(new TextEncoder().encode("안녕")). The engine has no dependency and no build step.
Say something
A bug report is most useful when it carries the exact string that failed, the format and the direction. Write to hello@pikkola.com.