HEIF: More imaging, fewer bytes | Nokia.com

How a flexible image container is expanding from advanced photography and HDR to geospatial imagery and lightweight web assets

Every day, billions of images are captured, edited, shared and displayed across smartphones, cameras, web applications and cloud services. As image quality improves through high dynamic range (HDR) photography, computational imaging and richer metadata, the file format around the compressed image matters almost as much as the codec itself.

To address this need, Nokia and other MPEG contributors developed the High Efficiency Image File Format (HEIF), first standardized in 2015. HEIF has since evolved from a format for efficiently stored images into a versatile framework for advanced photography, image collections, HDR, annotations, geospatial imagery and more [1][9]. Recent work adds support for Versatile Video Coding (VVC/H.266) and introduces the Low-overhead Image File Format, for small assets such as icons, thumbnails and stickers.

Why HEIF is more than another image format

Traditional image formats usually bind the image codec and the file structure closely together. HEIF takes a more flexible approach. It is a structural format: a container that defines how images, metadata and relationships are organized, while allowing the image data itself to be compressed with different coding technologies.

In simple terms, the codec compresses the pixels, while HEIF packages the resulting image and the information needed to use it. This separation allows the container to adopt newer codecs without redesigning the surrounding imaging workflow.

Built on the ISO Base Media File Format (ISOBMFF, ISO/IEC 14496-12), HEIF can store single images, image collections and image sequences. It can also associate the main image with thumbnails, alpha planes, depth maps, metadata and other supporting content.

What HEIF enables

Advanced photography and computational imaging

Multiple captures for exposure, white balance, focus, flash or depth-of-field bracketing
Auxiliary images, including alpha planes for transparency and depth maps
Camera and capture metadata that can support computational photography, 3D reconstruction and AR or XR workflows

Image collections and sequences

Burst captures and animated image sequences
Panoramas, albums, image collections and slideshows with transition effects
A main image together with thumbnails and related images in one file

Non-destructive image operations

HEIF can describe common presentation operations without changing or recompressing the underlying coded image. These include 90-, 180- and 270-degree rotation, cropping, image overlays and image grids. Avoiding unnecessary recompression helps preserve quality and can reduce processing.

Choice of compression technology

Because HEIF is codec-flexible, it can store images and image sequences encoded with several standards, including:

Advanced Video Coding (AVC/H.264)
High Efficiency Video Coding (HEVC/H.265)
Versatile Video Coding (VVC/H.266)
MPEG-5 Essential Video Coding (EVC)
JPEG

What the HEIF third edition brings

The third edition of HEIF [2] extends the format beyond storage and organization, enabling richer delivery and presentation experiences.

Progressive rendering

An image can be refined gradually while it is downloaded, allowing an early representation to appear before all image data has arrived. This is useful for web delivery, cloud photo services and large image repositories.

Camera parameters and region annotations

HEIF can carry camera intrinsic and extrinsic parameters as well as annotations linked to regions of an image. These features can support computational imaging, 3D reconstruction, machine-assisted analysis and interactive content.

Renderable text and fonts

Text and its related fonts can be carried separately from the underlying image. This makes it possible to create captioned or meme-like images, localized graphics and annotated content without recompressing the base image whenever the text changes.

HDR gain-map imaging

An amendment to the third edition adds tone-map derivation for HDR gain-map imaging [3]. A base image, a gain-map image and tone-mapping metadata can be stored together so that the same content can be rendered appropriately on displays with different capabilities. The gain-map metadata can follow ISO 21496-1.

HEIF for maps and very large images

Further extensions support overview images, tiled images and multi-resolution image pyramids. A viewer can request only the region and resolution it needs instead of decoding or transferring an entire very large image.

This is particularly relevant to digital maps, satellite imagery, navigation systems and cloud-based geospatial services. Combining tiling with efficient codecs such as HEVC or VVC supports scalable storage and distribution of large visual datasets.

Low-overhead HEIF: reducing overhead for small web images

Rich container features are valuable for complex media, but conventional HEIF metadata can be proportionally large when the image itself is tiny. Icons, stickers, emojis, thumbnails and interface assets may contain only a small compressed payload, so several hundred bytes of container metadata can significantly affect the total file size.

The Low-overhead HEIF [4], addresses this case by replacing the conventional hierarchical HEIF metadata structure with a compact linear representation. It is designed to preserve the essential capabilities needed for a small primary image and its related data while minimizing container overhead.

What Low-overhead HEIF can carry

A single primary image
Auxiliary images, including an alpha image for transparency and a gain map for HDR
ICC color profiles, EXIF and XMP metadata
Image dimensions, color information and decoder configuration
A compact representation of codec configuration data

Low-overhead HEIF also defines how a reader can reconstruct a conventional HEIF-compliant metadata structure. This allows existing HEIF processing code to be reused behind a lightweight preprocessing step, reducing the integration burden for established implementations.

Measured benefits for small images

Nokia evaluated low-overhead HEIF with compact VVC configuration against conventional HEIF with regular VVC configuration using images from the JPEG AI test dataset [8]. The relative benefit is largest at the smallest resolutions, where metadata represents a larger share of the complete file.

Image resolution	Bitrate reduction with low-overhead HEIF
8 x 8	67.32%
16 x 16	56.63%
32 x 32	38.50%
64 x 64	20.21%
128 x 128	8.53%
256 x 256	3.12%
Average across tested sizes	32.39%

Table 1. Relative bitrate reduction achieved by low-overhead HEIF with compact VVC configuration.

Nokia’s public release of HEIF source code

Nokia publicly released HEIF source code as a reference implementation to help developers explore the format, test interoperability and build applications. The repository includes ISOBMFF box parsing and writing, HEIF reader and writer libraries, and Java APIs. Recent additions cover VVC-coded images and low-overhead HEIF with compact VVC configuration.

The reference implementation is available in the Nokia HEIF GitHub repository [5]. A companion HEIF conformance repository [6] provides candidate files for testing established and newly standardized features. Nokia also maintains an HEIF example site [7] with practical examples such as thumbnails, image grids, overlays, sequences, HEVC- and VVC-coded images, and low-overhead HEIF files.

Looking ahead

HEIF has evolved from an efficient way to package HEVC-coded images into a broader imaging platform. Its flexible container model supports advanced photography, image collections, non-destructive operations, HDR, annotations, large tiled imagery and several current and future coding technologies.

VVC support strengthens HEIF for next-generation image compression, while low-overhead HEIF extends its value to small assets where every byte of metadata matters. Together, these developments allow the same standards-based ecosystem to address both feature-rich imaging and lightweight web delivery.

References

K. Kammachi-Sreedhar, Miska M. Hannuksela, Emre B. Aksu, “Updated overview of the High Efficiency Image File Format” , JVET 44^th Meeting, Hangzou, China, October 2026
ISO/IEC 23008-12:2025, Information technology – High efficiency coding and media delivery in heterogeneous environments – Part 12: Image File Format, third edition, July 2025.
ISO/IEC 23008-12:2025 Amendment 1, Support for tone map derivation and other technologies, October 2025. See also ISO 21496-1.
ISO/IEC 23008-12:2025 Amendment 2, Low-overhead image file format, June 2026.
Nokia HEIF source code repository.
Nokia HEIF conformance file repository.
Nokia HEIF examples.
ISO/IEC JTC 1/SC 29/WG 1 and ITU-T SG16, JPEG AI Common Training and Test Conditions.
Miska M. Hannuksela, “High five to 10 years of HEIF”, Nokia Blogs, 30 May 2023