What an image metadata viewer reads
A digital image carries two separate records. The pixels decide how it looks. The metadata describes everything around them: the device that captured the file, the settings used, the time and place of the shot, the software that edited it and, in many AI-generated files, the tool or model that produced it. An image metadata viewer parses that second record and reports what it holds in ordinary language.
The oldest layer is EXIF, the Exchangeable Image File Format that CIPA maintains as standard DC-008 (https://www.cipa.jp/e/std/std-sec.html). Camera and phone makers write EXIF blocks into JPEG, TIFF and HEIF files. The format defines fields for camera make and model, lens, exposure settings, orientation and timestamps. It also defines an optional GPS block that can hold latitude, longitude, altitude, the direction the camera faced and the time of the location fix.
IPTC fields sit beside EXIF in most professional workflows. The IPTC Photo Metadata standard (https://iptc.org/standards/photo-metadata/) defines creator, credit line, copyright notice, captions, keywords and a Digital Source Type field that some tools use to label synthetic media. News desks and stock libraries read these fields for attribution, and that is why they survive many publishing pipelines that drop everything else.
C2PA Content Credentials are the newest layer. C2PA is an open standard from the Coalition for Content Provenance and Authenticity (https://c2pa.org/). A manifest is signed metadata that records how a file was made or edited, including the tool, the model and the sequence of changes. Many AI generators attach one automatically. The viewer reports what the manifest declares and separately notes whether the structure is intact.
How it works
How to read the metadata inside an image
Step 1: Upload the image
Choose a PNG, JPEG, WebP, BMP, TIFF, HEIC, HEIF or AVIF file up to 20 MB. The upload goes from your browser straight to object storage, and that copy is what the reading runs on.
Tip
Read the original file when you can. A screenshot or a re-shared copy usually carries less metadata than the source.
Step 2: Run the listing
The forensic metadata operation parses the file and groups the fields it finds: file basics, camera and lens settings, GPS, timestamps, editing history, IPTC fields, XMP packets and C2PA records. It runs on the CPU tier and turns the result into a report.
Tip
The operation only reads. No pixels are changed and no cleaned file is produced.
Step 3: Read each field
The report explains each value and marks the layers that are missing. A file with no C2PA manifest, for example, says so instead of leaving a blank column. If you sign in, the reading is also saved in your workspace history with its status.
Tip
An empty section is a finding. It tells you what this file does not declare.
What the report covers
Camera and lens fields
Make, model, lens, focal length, aperture, shutter speed, ISO and the exposure program the camera recorded. These fields answer which device took the picture and how it was set.
GPS and location records
Latitude, longitude, altitude, direction and the time of the location fix when the camera wrote them. Many phones record a coordinate by default, and the report lists every GPS field the file contains.
Timestamps and editing history
The capture time, the time of each edit and the software that saved the file. Editors, converters and messaging apps often leave their names and versions in this layer.
C2PA Content Credentials
The signed manifest, the claim generator, the signer and the actions the manifest records. The report shows what the credential declares and whether the structure is intact.
AI generator tags
Generator names and model identifiers written into XMP, IPTC or the C2PA manifest. Some tools also record prompt or seed data, and the report shows those fields when they exist.
IPTC rights and caption fields
Creator, credit line, copyright notice, caption and keywords. These are the fields that news desks and stock libraries read when they check attribution.
File basics and structure
Format, dimensions, color model, bit depth, file size and a content hash. These values help you compare two versions of a file and confirm which one you are holding.
What happens to the file you upload
The reading is a server-side job. Your file uploads from the browser to Alibaba Cloud OSS in the US West region through a presigned URL, and the forensic operation reads that stored copy. The upload is not published, and no public result link exists because a reading produces no output image.
Uploads are kept for up to 30 days and then deleted, following the retention policy in our privacy page. A signed-in job also appears in your workspace history, where its status and report remain for the same period. We do not use uploaded content to train AI models.
What the metadata can and cannot tell you
After the reading: strip the layer or clean the pixels
A metadata reading often ends with a decision. If the fields you want gone are metadata, the Metadata Remover strips EXIF, GPS, editing history and C2PA records in one pass and returns a cleaned file. If the concern is the provenance records an AI tool wrote, the AI Metadata Remover concentrates on C2PA, IPTC Digital Source Type and generator fields. If you only want the signed credential removed, the C2PA Remover handles that layer and explains what a signature does and does not prove.
A metadata strip works on the file structure, so invisible watermarks such as SynthID survive it because the pixels never change. Removing those signals needs a full cleanup that rebuilds the image, and no tool can promise a particular detector result. Read the metadata first, then choose the layer you actually need to change.
FAQ
Frequently asked questions
Common questions about reading the metadata inside an image file.
Explore more tools
Metadata Remover
Strips EXIF, GPS, editing history and C2PA records from photos in one pass, with the pixels left as they were.
AI Metadata Remover
The specialist page for the records AI tools write: C2PA, IPTC Digital Source Type and generator fields.
C2PA Remover
Inspect Content Credentials, then strip the signed provenance record while the image keeps its dimensions.
Remove AI Label
The platform label this metadata can trigger, and the fields Instagram and Facebook read before applying it.
AI Watermark Remover
When the report shows a pixel watermark rather than a metadata field, this image pass handles that layer.
Apple Image Playground Metadata Remover
Clears the Image Playground metadata Apple writes at export. The page also explains the pixel signal Apple documents.
Bing Image Creator Metadata Remover
Clears the metadata Microsoft writes into its image exports, with the covered Bing surfaces listed on the page.
Pinterest AI Label Remover
Strips the metadata Pinterest reads at upload, with the Gen AI label rules explained on the page.
Samsung Galaxy AI Watermark Remover
The Galaxy page covers the localized AI marks and the C2PA record on S25 exports in one run.
All tools
Every remover and detector on one page, grouped by what each one cleans or checks.
Read the metadata before you share the file
One listing shows the camera fields, GPS records, C2PA credentials and AI tags a file carries. If you find something you would rather not share, the metadata pages strip it and return a clean copy.