Paste it as messily as you like
What you actually have in hand usually comes wrapped in something. All of these can go straight in.
background: url("data:image/png;base64,…")— the whole CSS declaration<img src="data:image/png;base64,…">— the whole HTML tagiVBORw0KGgo…— just the Base64, nodata:prefix- A value broken across lines — common once it has been through JSON or a log
- base64url using
-and_, or with the=padding cut off data:image/svg+xml,%3Csvg…— percent-encoded SVG- Raw SVG source
It tells you what the format really is
Sometimes the type declared in the data: prefix and the actual content disagree.
When that happens you are told it claims one thing but contains another. Nearly always the file
was truncated, or the start or end went missing during the copy.
Things that are not pictures — PDFs, fonts — arrive as data URIs too. Those cannot be previewed, but they can be saved.
Why no error about the missing = padding?
It is added for you. Dropping = is normal for JWT segments and the URL-safe
alphabet, so refusing them would make the tool useless.
How is the saved file's extension decided?
From the type in the data: prefix first, and if there is none, by reading the
leading bytes. PNG becomes .png, JPEG becomes .jpg.
Does what I paste go to a server?
No. Decoding and previewing both finish on this page. Even a value with a token or a key mixed into it stays here.