Convert Base64 to Image Online
Paste a base64 string or data URL and decode it back into a viewable, downloadable image. Choose the format for raw base64 input. Decoding runs locally on your device.
Your image stays on your device. This tool runs entirely in your browser. Your file is never uploaded to our server.
How do you decode base64 back into an image?
Decoding base64 into an image here is a three-step loop: paste the string, check the preview, and download the file. The page accepts a full data URL or a raw base64 string and decodes it with the Canvas API inside your browser, so the string and the recovered image never leave your device. A data URL is detected and decoded automatically, while a raw string just needs a format chosen from JPG, PNG, or WebP. The preview updates the moment you paste, so you can confirm the string decodes before you download anything.
- 1Paste a data URL or a raw base64 string into the field.
- 2For a raw string, pick the format: JPG, PNG, or WebP. Data URLs are detected automatically.
- 3Check the preview to confirm the image decoded correctly.
- 4Download the recovered image. The original string is not modified.
Data URL or raw base64: which do you have?
A data URL begins with data:image/jpeg;base64 or the matching png or webp type, and it carries the format inside the prefix, so the page decodes it with no extra input. A raw base64 string has no prefix, so the page needs you to pick JPG, PNG, or WebP before it can render the pixels. If the preview looks wrong, the format guess was the issue, and switching the choice re-renders instantly. You can also just paste first and read the preview; if the colors or shape look off, the format is the first thing to change.
Common use cases for base64 to image
Recovering an image from a JSON API response
APIs often return small images as base64 inside a JSON field, which is awkward to view or save by hand. Pasting the string here turns it back into a real file. A 16 KB base64 thumbnail decodes to a 12 KB PNG that you can preview and download in a few seconds, and it is the fastest route when the raw image was never saved to disk in the first place.
Extracting an image embedded in HTML or email
Inlined images in HTML and some email clients sit inside the markup as data URLs. Copy the data:image string and paste it here to get the picture back. A logo inlined at 6 KB of base64 returns as a 4.5 KB PNG you can reuse elsewhere.
Turning a canvas or clipboard data URL into a file
Screenshot tools and canvas exports often hand you a data URL instead of a file. Pasting it here converts that clipboard string into a downloadable image. A 1080x1080 canvas export around 800 KB of base64 comes back as a roughly 600 KB PNG.
Debugging an encoded asset before you ship it
When a data URL in your code does not render, pasting it here shows whether the string is valid and what the image actually is. A truncated string previews as a broken or half-loaded image, which is a fast way to confirm the payload is the problem, and it confirms the asset is the image you think it is before you deploy it.
Is decoding base64 lossless?
Yes. Decoding base64 is an exact, reversible operation: the bytes that come out are the same bytes that went in. A 133 KB base64 string decodes to a 100 KB binary image with nothing lost, because the decode itself never touches the pixels. The only quality loss that can appear happened earlier, when the original image was saved in a lossy format such as JPG.
Which format should you pick for a raw string?
Raw base64 strings have no header, so the page cannot guess the format for you. The three choices below cover every format the decoder outputs, and the preview updates instantly when you switch.
- PNG: the most common raw base64 format, and the right guess for screenshots, graphics, and anything with transparency.
- JPG: for photographs, where the smaller file and the lossy save are expected.
- WebP: for modern web assets that may carry transparency at a smaller size than PNG.
Why decode base64 in the browser?
The decode runs with the Canvas API inside your browser, so the pasted string and the recovered image never leave your device. That matters when the base64 came from a private email, an internal API response, or a screenshot on your clipboard that you would not send to a third-party service. There is no upload, no account, and no queue, and the preview appears the moment you paste.
Frequently asked questions
How do I convert a base64 string to an image file?
Paste the string into the page, pick the format if it is a raw string, and download the recovered image. A data URL is detected automatically, so you only need to choose a format when there is no prefix.
Do I need to know the format of my base64 string?
Not if it is a data URL, because the MIME type is inside the prefix and detected automatically. For a raw string, pick JPG, PNG, or WebP; if the preview looks wrong, switch the choice and it re-renders instantly.
Will the decoded image be identical to the original?
Yes. Decoding base64 is an exact, lossless operation, so the bytes come back the same as the ones that were encoded. Any difference from the original comes from a lossy format the image was saved in, not from the decode.
What can I do with the recovered image?
Preview it on the page to confirm it is the right one, then download it as a file. It is the fastest way to restore an image from an HTML email, a JSON API response, or a canvas or clipboard data URL.
Why is my decoded file smaller than the base64 string?
Base64 text is about 33% larger than the binary it encodes, so a 13 KB string decodes to a 10 KB image. If the result still seems small, the source was probably a low-resolution or heavily compressed image.
Does decoding base64 upload my data anywhere?
No. Decoding runs entirely in your browser with the Canvas API, so the pasted string and the recovered image never leave your device.
Can I decode very large base64 strings?
Single images decode fine, but very large strings use a lot of memory because the browser builds the full image before previewing it. If a multi-megabyte string feels slow, it is the size of the source image, not the decoder.