File metadata, privacy and what transformations really preserve

Files can contain information that is not obvious in the visible page or image. Metadata may identify software, dates, authors, locations or document history. Whether metadata survives depends on the transformation: copying original objects tends to preserve more, while rebuilding or re-encoding often discards more.

Last technical review: 2026-09-19

Key takeaways

Visible content and metadata are different data surfaces.
Image re-encoding commonly drops EXIF and GPS unless they are copied deliberately.
A PDF rebuild can remove original metadata by not reusing source objects, but it may also remove useful document structure.
A privacy statement should separately explain file processing, analytics, diagnostics and retained identifiers.

What counts as metadata

Image metadata can include camera information, capture dates, orientation, software fields and GPS coordinates. PDF metadata can include document information dictionaries, XMP packets, producer information and other document-level structures.

Metadata is not always sensitive, but it can reveal context that the visible page or image does not show.

Copying versus rebuilding

A transformation that copies source containers and objects may preserve metadata unintentionally. A transformation that decodes and re-encodes an image or rebuilds a PDF from new page representations often leaves source metadata behind unless it is deliberately reintroduced.

The second approach can improve privacy for some workflows but may also remove useful color profiles, accessibility data or document semantics. The correct behavior depends on the tool purpose and must be stated explicitly.

Telemetry is a separate privacy question

A tool can keep file bytes local and still collect bounded website analytics such as a tool identifier, browser class or referral source. Those two facts are not contradictory, but they must not be blurred together.

Strong privacy architecture prevents document bytes, filenames, extracted text, form values and request bodies from leaking into diagnostics or analytics while still allowing safe operational measurements.

How ZAMAZY communicates preservation

ZAMAZY focused tools define what they preserve and what they intentionally do not preserve. The redaction rebuild does not preserve original PDF metadata or interactive structures, while image re-encoding does not claim to preserve EXIF or GPS metadata.

The analytics and diagnostics contracts are separately allow-listed so local file processing is not presented as a claim that the entire website has zero measurement.

Example and decision guide

A photo can have ordinary-looking pixels but carry location metadata; a PDF can have no visible author name while retaining document information. Conversely, a file with no source metadata can still show a private address in its pixels. Metadata removal and removing visible sensitive content are separate objectives.

Choose the operation by the property you need. ZAMAZY JPEG/WebP re-encoding does not copy original metadata containers; the supported PNG optimizer preserves supported chunks and is not a metadata scrubber. The redaction rebuild uses new page images instead of original PDF objects. Do not transfer a claim about one operation to another simply because both produce a smaller or new file.

When this approach does not fit

Treat ICC color information and orientation as functional data, not automatically disposable clutter. Re-encoding can change color behavior, and a general encoder description is not a forensic sanitization certificate. Keep a suitable original for work that needs provenance or print fidelity. Removing file metadata also does not erase website traffic records or copies held elsewhere.

Editorial illustration, not a benchmark. Addition reviewed on: 2026-09-19

Sources and evidence boundaries

These sources explain formats and methods. They do not certify ZAMAZY; its supported facts and limits are stated on each tool page.

Google: WebP Container Specification — WebP modes, alpha, animation and metadata containers.
W3C: PNG Specification, Third Edition — PNG pixels, alpha and chunks; not a claim that every PNG variant is accepted.
MDN: HTMLCanvasElement.toBlob() — Encoding format and quality parameters; no certified metadata-removal claim.
Adobe: How to redact a PDF — Removal of selected and hidden information; Adobe is not certifying ZAMAZY.

Frequently asked questions

Can a photo reveal location even when the location is not visible?

Yes. Some images contain GPS metadata. Whether it is present depends on the source and whether later software preserved or removed that metadata.

Does rebuilding a PDF remove all possible hidden information?

A clean rebuild that does not copy original objects removes many original structures, but the guarantee must be defined by explicit verification. No generic “rebuild” label should replace actual checks.

Can a site use analytics and still process files locally?

Yes. The important boundary is that analytics events remain separate from the selected file contents and do not include filenames, extracted text or document bytes.