01 / IMAGE
Image Metadata Cleaner
Remove embedded metadata (EXIF, GPS, ICC, C2PA Content Credentials, XMP, thumbnails) from JPEG, PNG and WebP images entirely in your browser. Fast mode keeps pixels intact; Deep mode also re-encodes pixels to defeat perceptual hash matching.
JPG, PNG, WebP
Drop files here or click to upload
Supported formats: JPG, PNG, WebP
How to use
- 1Drop one or more JPG, PNG, or WebP files into the upload area. The tool accepts files up to 100 MB.
- 2Review the metadata preview shown on each file card — it lists C2PA Content Credentials, EXIF (including GPS, camera model, capture time) and XMP fields detected in the original image.
- 3Choose a mode: Fast clean only removes metadata containers (keeps original pixels and file format); Deep clean also re-encodes pixels to defeat perceptual hash matching.
- 4Open Advanced options if you need to retain the ICC color profile or EXIF Orientation tag.
- 5Click Strip all metadata. Each file is processed locally; nothing is uploaded.
- 6Download the cleaned file or use Download all when every file is done. Cleaned files are named <original>-clean.<ext>.
Frequently asked questions
Does this tool upload my images?
No. All processing — parsing, metadata stripping, optional canvas redraw — happens in your browser. Your image bytes never leave the device.
What is C2PA and why might I want to remove it?
C2PA (Content Credentials) is a signed metadata container that records where an image came from and which tools edited it. AI tools like Photoshop's Generative Fill and Adobe Firefly embed it by default, and platforms such as LinkedIn surface it as an 'AI Info' label. Stripping the C2PA chunk removes that signed provenance trail.
What's the difference between Fast and Deep clean?
Fast clean removes EXIF/XMP/ICC/C2PA/IPTC containers from the file binary while leaving the pixel data byte-for-byte identical. Deep clean does the same and then re-encodes the pixels through an HTML canvas, which slightly perturbs the pixel hash — useful against platforms that match images by perceptual hash rather than by metadata.
Will Deep clean degrade my image?
Deep clean re-encodes the image. PNG output stays lossless; JPEG and WebP are re-encoded at quality 0.92, which is visually indistinguishable on most photographs but is technically a re-compression.
Which formats are supported?
JPEG, PNG, and WebP. AVIF, HEIC, TIFF and GIF are not supported in this version.
Why does my JPEG file size barely change after Fast clean?
Most JPEG metadata is small (a few KB). The big size savings come from removing embedded thumbnails or large ICC profiles — both are stripped by default. If a file is mostly pixel data, the size difference will be modest.
Does this work on images from Chinese AI platforms (Yuanbao, Wenxin, Doubao, etc.)?
Yes. Chinese AI platforms are required by GB45438 / TC260 regulations to embed an AIGC identifier (a JSON blob with platform, content ID, and SM2 signature) inside the EXIF UserComment field. The metadata preview surfaces this as 'AIGC', and Fast clean removes the entire EXIF segment — so the AIGC label disappears along with the rest of EXIF. Note that this does not prevent visual AI detectors on social platforms; those use image features, not metadata.