PNG MIME Type

What does image/png mean?

The image/png media type identifies a Portable Network Graphics datastream. PNG stores raster images with lossless compression, so decoding reconstructs the encoded sample values without the generational damage introduced by a lossy codec. It supports grayscale, indexed-color, and truecolor images, with optional transparency and sample depths from 1 to 16 bits.¹

The current registration has no required or optional media-type parameters. Its registered extension is .png, and every PNG datastream begins with these eight bytes:¹²

89 50 4E 47 0D 0A 1A 0A

That signature is followed by chunks. A conforming stream starts with IHDR, contains one or more IDAT chunks, and ends with IEND. Each chunk carries a type and a cyclic redundancy check, which helps decoders detect damaged data.²

How PNG became a web standard

PNG began as a collaborative replacement for GIF. The first proposal, then called Portable Bitmap Format, appeared in January 1995, and the group adopted the PNG name later that month.³ W3C published PNG 1.0 as a Recommendation on October 1, 1996.⁴

The design emphasized lossless storage, portability, progressive display, integrity checking, and freedom from patent restrictions. Those goals made PNG a natural fit for web graphics, screenshots, diagrams, and images that need transparency.²

The first W3C Recommendation suggested image/x-png as an interim media type because registration was not yet complete.⁴ That historical spelling may still appear in old systems, but new content should use the registered image/png type.

What PNG preserves

PNG applies a reversible filter to each scanline and then compresses the filtered data with the zlib Deflate method. Filtering changes how the samples are represented for compression, not their decoded values.²

The format can represent five image types:

PNG stores alpha separately from color rather than premultiplying the color samples. Fully transparent pixels can therefore retain RGB values that are invisible in an ordinary viewer.²

Color-management chunks can describe gamma, chromaticity, sRGB, ICC profiles, wide color gamut, and HDR signaling. Text can appear in tEXt, zTXt, or iTXt chunks. Since ancillary chunks may be ignored by a decoder or stripped by an editor, do not use them as the only copy of critical information.²

How to open an image/png file

Chrome, Edge, Firefox, Opera, and Safari display PNG directly. MDN lists the same browser support for both PNG and APNG.⁵ Desktop image tools also support it broadly:

• Adobe Photoshop can export PNG and can preserve 16-bit PNG data in supported workflows.⁶
• Apple Preview can open images and export them as PNG.⁷
• Most operating-system photo viewers can open a .png without an additional codec.

If a file named .png will not open, verify its first eight bytes rather than trusting the extension. Renaming a JPEG, WebP, or SVG file does not convert its contents.

Support in Poly

Poly provides format-specific support for PNG. It recognizes both the image/png identifier and .png filenames, decodes the image, and creates lossless WebP thumbnails at several sizes. The original PNG can open directly in the image viewer.

During indexing, Poly extracts pixel dimensions and readable EXIF and PNG text tags. It also creates a visual embedding from a generated thumbnail, so PNGs can be found by filename, metadata, visual similarity, and search by color. Words that exist only inside the pixels are not OCR'd, so a screenshot is not searchable by its visible text.

The Poly agent normally receives the largest generated WebP thumbnail whose longest edge is at most 1536 pixels. If processing has not finished, it can fall back to the original image/png bytes when the file is no larger than 4 MiB. This lets the agent reason about the visible image, but it is not a byte-level review of every chunk, original bit depth, or metadata value.

PNG, JPEG, WebP, and AVIF

No single image format is best for every job. The useful distinction is what must survive and where the file must work.

MDN recommends PNG when precise reproduction or transparency matters, while noting that WebP and AVIF can compress more efficiently.⁵ A photograph saved as PNG can be much larger than a visually similar lossy JPEG, WebP, or AVIF. Conversely, JPEG can create ringing and block artifacts around text, UI elements, and hard edges.

Conversion should follow the intended use:

• Convert PNG to JPEG for a smaller photographic copy when transparency and exact pixels are unnecessary. Composite transparency onto an explicit background first.
• Convert PNG to WebP or AVIF for web delivery, then compare the decoded result and keep a fallback if your audience requires it.
• Convert a lossy source to PNG to avoid further loss during later saves. The conversion does not restore information already discarded by the source codec.
• Keep PNG for a lossless screenshot, diagram, texture, or image that needs widely supported alpha transparency.

Is APNG the same as PNG?

Animated PNG is an extension of PNG that adds acTL, fcTL, and fdAT chunks. It remains backward compatible: a decoder that does not understand animation can ignore those ancillary chunks and display the static image.²

The current specification registers animated content as image/apng, normally with the .apng extension. Older animated files are also commonly stored as .png, and browsers support that practice.²⁵ The two types share the PNG signature, so detecting animation requires parsing the chunks rather than checking only the first eight bytes.

Poly does not currently define a separate image/apng type or infer .apng filenames. A file using the ordinary .png extension can enter the PNG pipeline, but Poly's thumbnails, visual embedding, and agent view represent a static rendered image rather than the complete animation.