The short answer
Choose a JPEG, PNG, or WebP file and select Inspect metadata to read supported fields inside the current browser tab. Viewing comes before cleaning: you should know which fields exist, which are removable, and which are retained before creating another copy.
A local viewer is useful when checking a client delivery, downloaded attachment, edited export, portfolio image, or photo prepared for a public post.
Read the report from container to field
Begin with the signature-confirmed format, file size, and pixel dimensions. Those facts help detect selection mistakes: a tiny WebP thumbnail is not the same file as a large JPEG export even if both show the same scene. Then read group headings. Location and serial-number findings carry the strongest warning, while time, device, authorship, description, and free-form text require contextual review.
Each row states whether this tool can remove the item. An ICC profile or standard color information is retained because it supports predictable display rather than ordinary personal identification. Do not interpret “retained” as “ignored.” It means the cleaner intentionally preserves that item and the report gives you enough context to decide whether a more specialized workflow is necessary.
Why local viewing changes the privacy tradeoff
Many metadata websites require an upload, which gives another service the image bytes and any sensitive fields you wanted to inspect. A browser-only viewer can parse the selected ArrayBuffer in memory without a file-processing request. The webpage itself still loads through ordinary hosting and security systems, but the selected photo and extracted values do not need to be included in those page requests.
Local execution is not magic protection against a compromised device, browser extension, or operating system. It is a narrower data-flow choice: the application code does not transmit the selected bytes. You can reinforce that choice by keeping the browser updated, using a trusted device, closing the tab after use, and avoiding unknown extensions when reviewing sensitive material.
Use the browser tool on the final file
Confirm the file name and size in the picker, inspect once, and read the grouped report from top to bottom. Choose the exact JPEG, PNG, or WebP copy you intend to share and select Inspect metadata. Read the grouped results and distinguish removable findings from retained color information. If removal is appropriate, create a cleaned copy. The tool rewrites supported container metadata without canvas recompression, reparses its output, and enables download only after verification reports no supported removable metadata.
Keep the original separately, download the “-clean” copy, and inspect that downloaded file again when the decision is sensitive. This workflow gives evidence about one exact file; it does not alter earlier exports or copies already sent elsewhere. The selected bytes and parsed values stay in the current browser tab rather than being uploaded to an image-processing endpoint.
Interpret the result within its scope
The viewer dispatches from file signatures rather than trusting an extension, then validates the relevant container structure before showing findings. The viewer is intentionally not a universal forensic parser and does not decode every proprietary maker note or arbitrary ancillary payload.
A clean result is deliberately narrow: it means the parser did not find a supported privacy-sensitive or removable item in that file. It is not proof against every proprietary block, concealed payload, visible clue, filename disclosure, remote copy, or steganographic technique. Review pixels, context, audience, and destination separately, and use a purpose-built forensic workflow for high-risk decisions.