Metadata remover

How the metadata remover reads a file

Every image format keeps its metadata in labelled blocks, and all three formats here can be walked without decoding a single pixel. A PNG is a run of chunks, each one carrying its length and a four-letter type. A JPEG is a run of marker segments, each starting FF and carrying its own length. A WebP is a RIFF container with four-character chunk ids. The tool follows those length fields to the end, lists every block it recognises as metadata, then writes a new file out of the blocks that are left. The pixels are copied across untouched, so this is not a re-encode and there is no quality loss.

What it cannot do

• It cannot remove a watermark that lives in the pixels. Google's SynthID and similar schemes encode their signal into the image content itself, where it survives cropping, resizing and re-compression. Metadata removal does nothing to it, and any tool claiming otherwise is worth doubting.
• PNG, JPEG and WebP only. AVIF, HEIC, TIFF and PDF are not handled, and a file it does not recognise is handed back untouched rather than guessed at.
• Colour profiles are kept on purpose. An ICC profile is metadata in the technical sense, but deleting it changes how the image looks, so it stays.

Questions people actually ask

What counts as metadata here?

EXIF, XMP, IPTC and Photoshop resource blocks, free-text comments, modification timestamps, and C2PA Content Credentials. Each one is listed by name with its byte count before anything is deleted, so you can see what the file was actually carrying rather than trusting a summary.

Does removing metadata change the image itself?

No. The pixel data is copied from the original byte for byte, and the tool never decodes or re-encodes it. That is the difference between this and running a photo through a compressor to shed its EXIF: compression strips metadata as a side effect and costs you quality, while this only removes the blocks you were shown.

Why keep the colour profile?

Because deleting it visibly changes the image. An ICC profile tells the display how to interpret the colour values, and a photo that loses it can shift noticeably, especially one shot in a wide gamut. It carries nothing about you, your camera or your location, so there is no privacy reason to remove it.

Is the file uploaded?

No. The file is read into memory in your tab and the new one is built there. There is no upload endpoint on this site to receive it, which is the point for a tool whose whole job is handling files you would rather not hand to a stranger.

How do I check it worked?

Drop the cleaned file back in. The tool will report that it found nothing, which is the same check it ran the first time rather than a different, friendlier one. If you want a second opinion, any EXIF viewer will do.

Why does the file barely get smaller?

Because metadata is small compared with the picture. A typical EXIF block is one to twenty kilobytes against a photograph of several megabytes, so the saving rounds to nothing. The exceptions are a large embedded preview thumbnail, which can run to tens of kilobytes, and a C2PA manifest, which in some pipelines carries its own thumbnail as well. If you came here to make a file smaller rather than quieter, the image compressor is the tool that does that.

Which formats are handled, and what happens to the rest?

PNG, JPEG and WebP. Those three cover the overwhelming majority of images that travel around the internet, and each one is walked using its own container rules rather than a shared guess. Anything else, including AVIF, HEIC, TIFF, GIF and PDF, is reported as an unsupported format and handed back exactly as it arrived. Refusing to touch a file the tool cannot read is deliberate: a byte-level edit made on a wrong assumption produces a corrupt file that still looks fine in a listing.