Image Metadata Remover

The tool reads metadata with the exif-js library: it loads the file via FileReader, runs EXIF.getAllTags, and sorts every tag into one of five categories that each carry a fixed privacy threat level. Location (GPS) is rated 'critical' and matches tags like GPSLatitude, GPSLongitude, GPSAltitude and GPSTimeStamp; Identity ('high') covers Artist, Creator, Copyright, OwnerName, CameraOwnerName and ByLine; Device ('medium') covers Make, Model, LensModel, SerialNumber, BodySerialNumber and LensSerialNumber; Date/Time ('medium') covers DateTimeOriginal, CreateDate and ModifyDate; Software ('low') covers Software, ProcessingSoftware and CreatorTool; and Technical ('low') covers ExposureTime, FNumber, ISO, FocalLength, Orientation and ColorSpace. The overall threat score shown in the banner is the worst level present (critical if any GPS tag exists, otherwise high, medium, low, or 'safe' when no tags are found), and the file count plus a 'Contains GPS location!' warning is surfaced for any image that carries coordinates.

Removal is done by re-encoding rather than by surgically editing the file. cleanImage draws the decoded image onto an off-screen HTML canvas at its natural pixel dimensions and calls canvas.toBlob in the chosen output format. Because a canvas bitmap holds only pixels, this discards every ancillary segment at once - all EXIF, IPTC and XMP, and also the embedded ICC color profile and any EXIF thumbnail (which can otherwise retain an older, uncropped version of an edited photo). The trade-off is that this is all-or-nothing: although the interface shows the five categories, the cleaning step strips everything regardless of which categories are expanded, so you cannot keep, say, the color profile while dropping only GPS. After cleaning, the tool re-runs the same exif-js extraction on the output and shows a before/after panel with the resulting field count (typically zero) alongside the file-size delta and percent saved.

Output format is selectable between PNG (re-encoded at quality 1.0, labelled 'lossless, always clean'), JPEG, WebP and AVIF, with a quality slider (50-100, default 92) that appears only for JPEG and WebP. A batch tab accepts multiple images via drag-and-drop or picker, scores each one's threat level up front (flagging how many carry GPS), processes them sequentially with a progress bar and per-thumbnail status, and packages the results as a ZIP using JSZip, with each file renamed to '<name>-clean<ext>'. Everything runs in the browser via the FileReader and canvas APIs, so images are never uploaded. Note that re-encoding to a lossy format (or PNG-to-JPEG) is a true pixel re-compression, not a metadata-only edit, so it costs one generation of image quality even though it guarantees a metadata-free result.