What is HEIC? What's inside a HEIC file

The answer in one paragraph

A HEIC file is an image stored in the High Efficiency Image File Format (HEIF) with its pixels compressed by HEVC (High Efficiency Video Coding). HEIF is a container: it holds a primary image and, optionally, a thumbnail, metadata such as Exif, tiles and extra images. HEVC is the compression inside it. Apple's High Efficiency camera setting (opens in a new tab) writes HEIC on iPhone 7 and later; choosing Most Compatible writes JPEG instead.

HEIF vs HEIC vs HEVC vs AVIF

These four names live on different layers, which is why they are so often mixed up. Two layers describe the file (the container and the extension and brand that name it) and one describes the pixels (the codec).

The filename is a label; the brand is in the file

IANA's registration says there is no magic number: the way to recognise a HEIF file is the ftyp box at the start, which lists a major brand and compatible brands. Nothing stops a file's name and its brand from disagreeing, because the name is only a label. Here are the first 28 bytes of the HEIC we generated for this article, decoded.

This is why a viewer can support HEIF and still fail on a HEIC: reading the container is one job and decoding HEVC is another. The HEIC viewer and the tools below read the brand, not the filename.

A short, sourced history

Two dates for HEIF look contradictory and are not. The Library of Congress format description says MPEG declared the specification technically complete at its 112th meeting in June 2015, and Apple's engineers, in their WWDC 2017 session, say version one became an ISO standard in June 2015. ISO's catalogue, however, lists the first edition of ISO/IEC 23008-12 as published on 12 December 2017. We report both and do not try to reconcile them further.

HEVC came first (ITU-T lists its first edition as April 2013), and HEIF was designed to store its still pictures. The Library of Congress says Apple and Nokia were heavily involved in HEIF's development; Nokia Technologies publishes an open HEIF implementation (opens in a new tab) on GitHub. Apple made it practical for everyone: Apple's support page says you need iOS 11 or macOS High Sierra or later (opens in a new tab) to view or edit HEIF media, and iPhone 7 and later can capture it.

What is physically inside a HEIC

A HEIC is a sequence of boxes, a length and a four-letter type followed by contents, as defined by the ISO base media file format (the same box structure MP4 video uses). Rather than draw this from memory, we generated a HEIC and ran this site's own bounded parser on it, the same code behind the HEIC structure viewer.

The shape is the point: everything that tells a program how to read the file is in meta, at the start, and costs about a kilobyte. A reader can parse meta, find the primary item or the thumbnail, and jump straight to its bytes without decoding anything else. Apple's WWDC 2017 session makes this argument for putting thumbnails early in the file: it allows displaying a large number of HEIF images "without the need to parse the whole file".

Items: what the parser found

A big photo is a grid of independent tiles

Our 4032 × 3024 image is stored as twelve tiles because we asked for 1024-pixel tiles. Without that setting, the same image was a single hvc1 item. So tiling is an encoder choice, not a law of the format. Apple's WWDC 2017 session says iOS 11 encodes images using 512 × 512 pixel tiles positioned in a grid, and lists the benefits: parallel decoding, lower memory use when resizing and fast cropping. By that rule a 4032 × 3024 iPhone photo would be 8 × 6 tiles. We have not opened an iPhone file, so that count is our arithmetic, not an observation.

Everything else is an item pointing at an item

• Thumbnail. A separate small image, linked with thmb. Apple's session describes a 320 × 240 HEVC thumbnail, four times the area of a common 160 × 120 JPEG thumbnail. The thumbnail extractor and the thumbnail extractor cover pulling it out.
• Exif and XMP. Metadata blocks stored as their own items, linked to the image with cdsc.
• Auxiliary images. An image that complements the main one and is not normally shown. Apple's session names an alpha plane or a depth map as examples, and says Apple stores depth as an auxiliary image coded with HEVC's monochrome profile. Our second file shows the alpha case: an auxl reference to the primary image.
• HDR gain maps. Apple's WWDC 2024 session describes an SDR base image plus a gain map, and says a Core Image save includes the gain map as an auxiliary image along with the base SDR image. The file holds ingredients and a recipe, not a finished HDR picture, in Apple's words.
• Extra full-size images. A HEIC can hold several pictures with one marked primary. A macOS dynamic wallpaper is one: a stack of frames plus an XMP note saying when to show each; see HEIC dynamic wallpaper.
• Live Photos are not inside the HEIC. Apple's 2016 session describes a Live Photo as two assets, a still image file and a QuickTime movie, tied together by a shared UUID. That session predates HEIC (the still was then a JPEG); we have not opened a Live Photo file, so treat this as the design Apple described, not an inspection.

Is HEIC lossless?

The format can be; the files your phone writes are not. Nokia's HEIF site and the Library of Congress both say HEIF supports lossy and lossless storage, and HEVC includes a lossless mode. We found no Apple documentation of lossless HEIC capture, and Apple states its saving as about 2x against JPEG "at the same visual quality", which is a lossy-coding comparison, not an exact-pixels one.

We tested what "lossless" means in practice. We encoded one image four ways, decoded each, and compared every pixel with the source:

• The default lossless setting was not pixel-identical (largest error 65 of 255). The likely cause is 4:2:0 chroma: the colour channels are stored at quarter resolution, so detail is already gone before lossless coding starts.
• With 4:4:4 chroma the error shrank to 1, and with the identity matrix (matrix coefficients 0, so no RGB-to-YCbCr conversion) the result was bit-exact. That file was 1.90 MB, close to the 2.06 MB PNG of the same pixels.
• So a tool that says "lossless HEIC" is worth checking. Chroma layout and colour conversion decide whether it is.

Bit depth and colour

JPEG's lossy modes carry 8 bits per channel (Library of Congress (opens in a new tab)), which is 256 levels per colour. HEVC's original Main Still Picture profile is also 8-bit, 4:2:0 and one picture (Sullivan et al. (opens in a new tab)); Apple's session says that is the profile iOS 11 used for images. The heix brand covers the Main 10 and range-extension profiles, which add more bits per sample.

We made a 10-bit file to see what that looks like to the parser: its major brand was heix instead of heic, and the pixi property reported 10, 10, 10 bits per channel (1,024 levels). Apple's WWDC 2024 session says the most modern way to save HDR is 10-bit HEIF with a PQ colour space. Colour is declared by a colr property, either as numeric colour codes (nclx) or an embedded ICC profile; in our files it was nclx with full-range flag set. We did not verify which of those forms Apple's cameras write, or how Display P3 is signalled, so we make no claim about them here. The HEIC HDR guide goes into bit depth, PQ and gain maps, and Can a HEIC file contain a virus? covers safety.

Why a HEIC is smaller than a JPEG

Apple's engineers put the saving at "an average of 2X compression compared to JPEG containing the same visual quality". It is an average, not a guarantee: it depends on content, the encoder and the quality target, and the HEVC authors' roughly 50% goal was set for video. Three design differences, from the HEVC authors' own overview, explain most of it:

1. Prediction. Instead of coding every block from scratch, HEVC predicts a block from neighbouring pixels that were already decoded and stores only the difference. Its intra prediction offers 33 angular directions plus planar and DC modes.
2. Bigger, more flexible blocks. HEVC divides a picture into coding tree blocks of up to 64 × 64 samples and uses transforms from 4 × 4 to 32 × 32. JPEG's lossy modes work on fixed 8 × 8 blocks. Flat areas can be coded as a few large blocks; detail gets small ones.
3. Better entropy coding. The last stage packs the numbers into bits. HEVC uses context-adaptive binary arithmetic coding (CABAC), which adapts to the data as it goes.

The cost is work: HEVC encoders and decoders are far more complex than JPEG's, so hardware support matters. For a photo-by-photo size comparison with caveats, see HEIC vs JPG. We give no size ratio from our synthetic test image, because a generated picture does not behave like a photograph.

Patents and licensing

The HEIF container is an ISO standard. HEVC, the codec, is covered by patents licensed through pools. Access Advance (opens in a new tab) says its HEVC Advance pool licenses patents essential to HEVC on fair, reasonable and non-discriminatory terms. Via LA (opens in a new tab) runs a second HEVC/VVC licence, says it extends to devices implementing the technology, and adds that each user must determine its own licensing needs. Both are about companies shipping encoders and decoders; neither page describes anything a person viewing their own photo has to do.

AVIF exists partly as an alternative. The Alliance for Open Media's patent licence (opens in a new tab) grants a perpetual, worldwide, non-exclusive, royalty-free licence for its specifications. We did not verify what individual vendors pay or how Windows and Apple license HEVC, so we say nothing about that. See the Windows guide for the practical side.

Who can open it

Apple's support page says iOS 11 or later and macOS High Sierra or later can view, edit and duplicate HEIF media. Elsewhere it varies by operating system, app and browser, and it changes. These pages track the details: browser support, opening HEIC on Windows, HEIC on iPhone and converting on a Mac, HEIC on Linux and HEIC in Photoshop, Lightroom and GIMP. If a file will not open, start with HEIC will not open.

When to use HEIC, and when not to

What to do with a HEIC file

• Look at it: open it in the HEIC viewer, which decodes in your browser.
• See how it is built: the HEIC structure viewer draws the box tree and item table for your own file.
• Read what it says about itself: HEIC inspector and HEIC file info show metadata and properties.
• Get a JPG: HEIC to JPG converts on your device; nothing is uploaded.
• Pull out the embedded preview: the thumbnail extractor.