Add PNG Metadata
Write authorship, copyright and provenance directly into a PNG. Text metadata lives in tEXt chunks alongside the image data — this adds or replaces them without touching a single pixel, so the file you get back is byte-identical in its image content and only differs in what it says about itself.
Drop PNG files here
Other formats are converted to PNG first
How to add metadata to a PNG
- Drop the files. A batch all gets the same metadata, which is the point when you are stamping a set of exports.
- Fill in the keywords. Delete any line you do not want — a blank value still writes an empty chunk, which is rarely what you mean.
- Add your own keywords on new lines. Anything is allowed as a keyword; the registered ones are just conventions readers know to look for.
- Tick strip to clear whatever was already there before writing. Leave it off and existing chunks survive, except any whose keyword you are rewriting.
- Write and download. Verify with the chunk viewer, or
exiftool/pnginfoon the command line.
The registered keywords
The PNG specification names eleven conventional keywords: Title, Author,
Description, Copyright, Creation Time, Software,
Disclaimer, Warning, Source, Comment. Using these means
other tools display your values in the right place. Custom keywords are legal and widely used — generative
image tools write their prompt and seed into custom tEXt chunks, for example — and readers that
do not recognise them simply show them as-is.
Keywords must be 1–79 characters of printable Latin-1 with no leading, trailing or repeated spaces; anything
that would produce an invalid chunk is skipped rather than written. Values here are also Latin-1, which is what
tEXt permits. For text outside that range — CJK, emoji, most accented scripts beyond western
European — the format has iTXt, which is UTF-8; this page writes tEXt only, so
characters outside Latin-1 will not round-trip.
Metadata is not a watermark
Anyone can read it, anyone can strip it, and most platforms strip it for you. Social networks, chat apps and
CDN image pipelines routinely re-encode uploads and discard every ancillary chunk in the process. Treat
tEXt as documentation for files you control — asset pipelines, archives, design handoffs — not as
a rights-enforcement mechanism. For a visible mark that survives re-encoding, use
Watermark.
The reverse operation is Remove metadata, which is the one to reach for before publishing — a PNG exported from a design tool often carries the software version, and screenshots can carry more than you expect.
FAQ
Does this re-encode the image?
Not if the input is already a PNG — the chunk stream is parsed, edited and reassembled, so the compressed image data is copied through untouched. A JPG or WebP input has to be decoded and re-encoded to PNG first, and the status line says so.
Where do the new chunks go?
Immediately after IHDR, before the image data. That is valid for tEXt and means a reader gets the metadata without decoding the image.
What happens to a keyword that already exists?
The old chunk is dropped and yours replaces it, so you do not end up with two conflicting Author values. Other keywords are left alone unless you tick strip.
Can I add EXIF data instead?
PNG can carry an eXIf chunk, but support for reading it is inconsistent and it is not what most PNG tooling looks at. tEXt is the format's native answer and is far more widely understood.