AVIF
AVIF (AV1 Image File Format) is an open, royalty-free image file format specification for storing images or image sequences compressed with the AV1 video codec within the High Efficiency Image File Format (HEIF), which is based on the ISO Base Media File Format (ISOBMFF).[1] It enables the encapsulation of AV1 intra-frame coded content to achieve efficient compression while supporting a range of image types, including static images, animations, and multi-layered compositions.[2] Developed by the Alliance for Open Media (AOMedia), a cross-industry consortium of technology companies, AVIF builds on the royalty-free AV1 codec to address limitations in legacy image formats like JPEG by providing superior compression efficiency and visual quality.[2] The format's initial specification, version 1.0.0, was finalized and publicly released in February 2019, with subsequent updates in versions 1.1.0 and 1.2.0 introducing enhancements such as improved support for high dynamic range (HDR) imagery and progressive decoding.[1] AVIF draws from established standards like HEIF and the Multimedia Application Format (MIAF) to ensure interoperability across devices and applications.[1] Key features of AVIF include support for both lossless and lossy compression, wide color gamut (WCG), transparency via alpha channels, and HDR content, making it suitable for modern web and mobile applications.[2] It handles monochrome and multi-channel images across various bit depths and color spaces defined in AV1, as well as image sequences for animations.[1] Compared to traditional formats, AVIF delivers over 50% file size reduction relative to JPEG and more than 30% savings over WebP at equivalent quality levels, enabling faster loading times and reduced bandwidth usage.[2] AVIF has seen growing adoption since its release, with native support in major web browsers including Chrome (from version 85), Firefox (from version 93), and Safari (from version 16), as well as operating systems like Android 12 and iOS 16.[2] Reference implementations, such as the libavif library, facilitate encoding and decoding, and the format is integrated into content delivery networks and image processing tools for large-scale deployment.[2] As an open standard, AVIF promotes long-term sustainability without licensing restrictions, positioning it as a next-generation solution for image coding in streaming, web, and archival contexts.[2]History and Development
Origins and Development
The Alliance for Open Media (AOMedia) was founded on September 1, 2015, by seven leading technology companies—Amazon, Cisco, Google, Intel, Microsoft, Mozilla, and Netflix—to create open-source, royalty-free media compression technologies aimed at improving efficiency for internet-based video and image delivery. Following the release of the AV1 video codec specification in March 2018, AOMedia extended its efforts to still images by developing the AV1 Image File Format (AVIF), which adapts AV1's intra-frame encoding tools for static image storage within the High Efficiency Image File Format (HEIF) container. The initial version of the AVIF specification was published in September 2018, marking the format's formal inception as a collaborative project involving contributions from AOMedia members such as Google, Netflix, and others.[3] AVIF's development was driven by the need for a royalty-free image format that overcomes the compression limitations of JPEG, which struggles with high-resolution content and modern bandwidth constraints, while avoiding the licensing complexities of proprietary alternatives like HEIF, which depends on the royalty-bearing HEVC codec.[4] By repurposing AV1's advanced coding techniques—originally designed for video—for single-frame images, AOMedia sought to enable smaller file sizes and broader feature support, such as high dynamic range and wide color gamut, in an open ecosystem accessible to all developers. This approach aligned with AOMedia's broader mission to foster innovation without intellectual property barriers, positioning AVIF as a foundational technology for next-generation web imagery. Early development included prototypes tested by AOMedia members, with Netflix playing a key role in practical validation. In December 2018, Netflix released the first public AVIF sample images, showcasing the format's potential through encoded examples derived from AV1 tools.[5] Building on this, Netflix open-sourced a comprehensive framework in February 2020 for evaluating AVIF's performance alongside other codecs, using standardized metrics to demonstrate its viability in real-world applications like user interface graphics.[4] These efforts highlighted AVIF's integration of video codec advancements into image workflows, paving the way for broader adoption within open-source communities.[6]Standardization and Versions
The Alliance for Open Media (AOMedia) finalized and published the initial version of the AVIF specification, version 1.0.0, on February 19, 2019, establishing the foundational open-source framework for the format based on the AV1 video codec for static images.[1] This release represented a pivotal milestone, enabling early adoption and implementation by developers while aligning with AOMedia's mission to advance royalty-free media compression technologies. Subsequently, AVIF achieved international standardization as an extension to the High Efficiency Image File Format (HEIF) under ISO/IEC 23000-22:2019, published in July 2020, which incorporated AVIF as a defined profile within the HEIF ecosystem.[3] This integration leverages the ISO Base Media File Format (ISOBMFF) as the underlying container structure, ensuring compatibility with existing multimedia standards and facilitating interoperability across devices and platforms.[7] In April 2022, AOMedia released version 1.1.0 of the specification on April 15, enhancing AVIF's capabilities with support for High Dynamic Range (HDR) imaging, including wide color gamut representation and gain maps to enable dynamic tone mapping for backward compatibility with Standard Dynamic Range (SDR) displays. These updates addressed growing demands for advanced visual fidelity in modern displays, building on the core compression efficiency of earlier versions while maintaining the format's open and extensible nature.[7] In November 2025, AOMedia released version 1.2.0 of the specification on November 3, introducing support for progressive image decoding through layered images and sample transform derived image items, which enable higher precision per sample (e.g., 16-bit) by combining multiple AV1 image items. These enhancements further improve decoding efficiency and flexibility for advanced image processing.[8]Technical Specifications
Container Format
AVIF employs the High Efficiency Image File Format (HEIF) as its container, which builds upon the ISO Base Media File Format (ISOBMFF) to organize and store image items, associated metadata, and optional image sequences in a modular box-based structure.[9] This design allows for efficient encapsulation of visual content while supporting extensibility for future enhancements.[9] The HEIF container ensures compatibility with broader media ecosystems, as ISOBMFF serves as the foundational framework for formats like MP4.[3] AVIF files are identified by the .avif extension and the MIME type image/avif.[10] The core file structure commences with the 'ftyp' box, which declares the file's major brand—such as 'mif1' for the media image file format, 'avif' for AVIF still images, or 'avis' for AVIF sequences—and lists compatible brands to ensure interoperability.[9] Subsequent boxes include the 'meta' box, which houses declarative image properties and references to items, and the 'ipco' box within the item properties container, detailing attributes like color spaces, pixel aspect ratios, and transformation matrices for individual image items.[9] This container supports the inclusion of multiple images within a single file, enabling features like image grids or sequences derived from video frames.[9] Furthermore, AVIF accommodates rich metadata integration, such as EXIF tags for capturing device information and exposure settings, ICC profiles for precise color reproduction across devices, and XMP packets for extensible metadata including editing workflows and rights management.[3]Coding and Compression
AVIF relies on the AV1 video codec to compress still images by treating each image as a single intra-coded frame, effectively utilizing AV1's keyframe encoding capabilities without temporal dependencies.[9][4] This approach encapsulates the compressed image data within the HEIF container format, enabling efficient storage of high-quality visuals.[9] The core compression process in AV1, as applied to AVIF, follows a block-based hybrid coding framework that includes recursive partitioning of the image into square or rectangular blocks, ranging from 4x4 to 128x128 pixels, to adapt to varying content complexity.[11][12] For intra-frame prediction—essential for still images—AV1 employs multiple directional modes (up to 56 angular directions plus smooth and DC modes) to estimate pixel values from neighboring blocks within the same frame, reducing spatial redundancy.[11][12] The prediction residual is then transformed using techniques such as discrete cosine transform (DCT) for larger blocks or asymmetric discrete sine transform (ADST) for edge-heavy areas, concentrating energy into fewer coefficients for compaction.[11][12] Quantization follows the transform stage, scaling coefficients with a quantization parameter to discard less perceptible details in lossy mode, while setting the parameter to zero enables true lossless compression by preserving all data without alteration.[12][13] Finally, entropy coding with adaptive arithmetic (ans) efficiently represents the quantized data and side information, adapting to local statistics for further bitrate reduction.[11] Although inter-frame prediction modes exist in AV1, they are not utilized in AVIF still images, limiting prediction to intra modes only.[9][12] AVIF supports resolutions up to 8K (8192 pixels in width or height for baseline profiles), accommodating large images while maintaining compatibility.[9] However, AV1's encoding complexity—stemming from extensive mode decision searches, multiple transform options, and rate-distortion optimization—results in high computational demands, often requiring significantly more processing time than legacy formats like JPEG.[11][12]Profiles and Compatibility
AVIF defines profiles to categorize bitstreams based on the capabilities of the underlying AV1 codec, ensuring consistent decoding performance across devices. These profiles impose specific constraints on AV1 profiles, levels, and other parameters, facilitating interoperability in still images and short sequences.[9] The Baseline Profile, denoted by the four-character code 'MA1B', relies on the AV1 Main Profile at levels up to 5.1. This configuration targets basic still images, restricting the maximum luma dimensions to 8192 pixels in width and 4352 pixels in height, with a total luma sample count of up to 8,912,896 pixels to align with typical hardware decoding limits. The maximum bit rate is capped at 100 Mbit/s, promoting efficient processing on consumer-grade devices without advanced features. For image sequences under this profile, the maximum frame rate is constrained such that the total luma samples per second do not exceed 268,435,456 (e.g., up to approximately 32 fps at 4K resolution).[3][9][14] The Advanced Profile, identified by 'MA1A', incorporates the AV1 High Profile at levels up to 6.0, accommodating more sophisticated applications like high-resolution stills and sequences with enhanced color depth. It permits larger dimensions of up to 16384 pixels in width and 8704 pixels in height, with a total luma sample count of up to 35,651,584 pixels, and a higher bit rate limit of 400 Mbit/s, enabling support for demanding content while maintaining compatibility with capable decoders. Sequence frame rates can reach higher values, with the samples per second limit increased to 483,183,820 (e.g., up to approximately 58 fps at 4K resolution).[3][9][14] To ensure broad interoperability, AVIF files are structured as conformant HEIF containers, incorporating the 'mif1' brand for backward compatibility with existing HEIF decoders that can ignore AV1-specific elements. Compliant files must declare compatible brands including 'avif' (for AVIF identification), 'mif1' (HEIF compatibility), 'miaf' (media image file format), and the profile-specific 'MA1B' or 'MA1A'. These brands, listed in the file's major and compatible brand boxes, allow decoders to recognize and process AVIF content within HEIF ecosystems. Levels further enforce operational bounds, preventing excessive resource demands and promoting consistent behavior across implementations.[9]Features and Capabilities
Compression and Quality
AVIF demonstrates significant compression efficiency, typically achieving file size reductions of around 50% compared to JPEG while preserving equivalent perceptual quality. This performance is driven by the underlying AV1 codec's advanced intra-frame prediction methods, including a wide array of directional modes and smooth interpolators, combined with flexible transform coding that adapts to image content. These techniques excel in handling the complex textures and gradients common in photographic images, enabling superior bitrate efficiency over legacy formats.[4][15][16] In lossy scenarios, AVIF provides 20-30% better compression than WebP at comparable quality levels, resulting in smaller files without noticeable degradation in visual fidelity. For lossless compression, AVIF's mode can yield improved ratios over PNG for specific content types, such as certain UI graphics and synthetic images, where its entropy coding optimizations reduce redundancy more effectively than PNG's deflate-based approach.[17][18] Quality assessments using metrics like Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM) highlight AVIF's strengths, with benchmarks indicating consistent gains in these scores relative to JPEG across diverse datasets.[4]Supported Image Properties
AVIF accommodates bit depths of 8, 10, and 12 bits per channel, providing flexibility for standard dynamic range (SDR) images as well as enhanced precision for advanced applications. Version 1.2.0 further extends this to 16 bits or more per channel using the Sample Transform Derived Image Item ('sato').[19][8] This range aligns with the AV1 codec's capabilities, where 8-bit suits typical web imagery, while 10- and 12-bit depths reduce banding in gradients and support professional workflows requiring subtle tonal variations.[2] For high dynamic range (HDR) content, AVIF leverages transfer functions such as Perceptual Quantizer (PQ) or Hybrid Log-Gamma (HLG), paired with BT.2020 color primaries to deliver vivid highlights, deep shadows, and expanded contrast.[7] Starting with version 1.2.0 in October 2025, the format also incorporates HDR gain maps, which enable backward compatibility by allowing SDR displays to render a base image while HDR-capable devices apply tonal adjustments for enhanced detail.[8] These features make AVIF suitable for modern displays, preserving perceptual quality across varying brightness levels. In terms of color representation, AVIF supports wide color gamut spaces like BT.2020, which covers a broader spectrum than sRGB or BT.709, enabling more accurate reproduction of saturated hues in photography and graphics.[2] It also permits embedding of ICC profiles within the HEIF container to ensure consistent color management across devices and software.[20] Additionally, an alpha channel is natively supported for per-pixel transparency, facilitating compositing without artifacts common in older formats. Beyond static images, AVIF handles animation by treating sequences as short video clips encoded with AV1, supporting variable frame rates for smooth motion in web and app contexts.[21] This approach leverages the container's ability to store multiple timed images, offering efficient playback for icons, UI elements, or brief loops. AVIF further includes scalable coding through AV1's multi-layer structure, which allows embedding lower-resolution thumbnails or progressive refinements within a single file for faster loading of previews.[4] Orientation and rotation are managed via built-in transformative properties (irot for rotation and imir for mirroring), supplemented by EXIF metadata storage to preserve capture details like sensor orientation without altering pixel data.[22]Comparisons with Other Formats
Versus JPEG and PNG
AVIF provides substantial efficiency gains over JPEG for photographic content, producing files that are typically around 50% smaller at equivalent visual quality levels due to its advanced AV1-based compression.[20] This reduction is particularly beneficial for web delivery, where bandwidth constraints are common, and AVIF exhibits superior artifact handling in areas like gradients and smooth transitions, avoiding the blocking effects inherent in JPEG's discrete cosine transform method.[4] Furthermore, while JPEG—standardized in 1992 by the Joint Photographic Experts Group—remains ubiquitous for lossy compression of photos, it lacks native support for transparency or high dynamic range (HDR) imaging, features that AVIF incorporates seamlessly.[23] Compared to PNG, AVIF delivers smaller file sizes even in lossless mode for photographic images, where PNG's deflate-based compression proves inefficient for continuous-tone content, often resulting in files several times larger.[24] AVIF retains PNG's strengths in transparency support while adding animation capabilities—extending beyond PNG's static nature (with animated variants like APNG being non-standard)—without incurring PNG's bloated sizes for complex, high-detail images.[20] PNG, formalized as a W3C recommendation in 1996 as a patent-free alternative to GIF, excels in scenarios requiring exact lossless reproduction of simple graphics, icons, or line art, but its limitations make it less suitable for bandwidth-sensitive photographic applications.[25] In practice, AVIF is increasingly favored over JPEG for web-based photographs, enabling faster load times and reduced data usage on modern networks ill-suited to JPEG's dated compression from the pre-broadband era.[4] PNG, however, persists as the go-to for crisp, transparent elements like logos or UI icons where AVIF's computational demands might represent overkill and compatibility is paramount.[20]Versus WebP and HEIF
AVIF offers notable advantages over WebP in compression efficiency for lossy photographic images, achieving file sizes approximately 10-20% smaller at comparable quality levels, as demonstrated in benchmarks using distortion metrics like DSSIM on natural content.[26] This edge stems from AVIF's reliance on the AV1 video codec, which employs more advanced prediction and transform techniques compared to WebP's VP8/VP9 foundations developed by Google.[27] In terms of quality preservation, AVIF outperforms WebP particularly at low bitrates, where it maintains superior structural similarity for natural images, with SSIM scores around 0.93 versus WebP's 0.89 in controlled tests on diverse datasets.[28] Additionally, AVIF provides better support for high dynamic range (HDR) imaging through native 10-bit color depth and HDR metadata handling, enabling richer tonal ranges that WebP lacks in its standard implementation.[20] However, WebP benefits from faster encoding and decoding speeds, making it more suitable for real-time applications, while its longer history contributes to broader and more mature adoption across web ecosystems.[29] In comparison to HEIF (High Efficiency Image Format), AVIF serves as a royalty-free alternative by utilizing the same ISOBMFF-based container but pairing it with the open-source AV1 codec instead of the patented HEVC (High Efficiency Video Coding).[30] This design avoids the licensing fees associated with HEVC, which can impose significant costs on implementers, while delivering similar compression efficiency—benchmarks indicate AVIF achieves comparable or slightly better file size reductions for equivalent perceptual quality on still images.[31] For instance, in evaluations of natural scenes, AVIF and HEIF exhibit tied performance in metrics like PSNR and SSIM at matched bitrates, but AVIF's open licensing fosters broader support from diverse platforms and developers without proprietary restrictions.[32] HEIF, while efficient, remains more Apple-centric due to its integration with iOS ecosystems and HEVC's prevalence in mobile devices, limiting its cross-platform openness compared to AVIF's emphasis on royalty-free accessibility.[27]Software and Platform Support
Web Browsers
Google Chrome introduced full support for the AVIF image format in version 85, released in August 2020, enabling decoding and display of AVIF files within web pages.[33] Microsoft Edge, based on the Chromium engine, followed with AVIF support starting in version 121, released in January 2024, aligning its capabilities with Chrome for consistent rendering across Chromium-based browsers.[33] Subsequent updates in Chrome and Edge have extended support to include high dynamic range (HDR) AVIF images, allowing for enhanced color and brightness reproduction on compatible displays. Mozilla Firefox added AVIF support in version 93, launched in October 2021, after an experimental phase that allowed enabling the feature via configuration flags in earlier releases.[33] This implementation initially covered static AVIF images, with support for animated sequences added in version 113 (April 2023), focusing on efficient still and animated image rendering for web use.[33] Apple's Safari browser implemented AVIF support beginning with version 16 in September 2022, coinciding with the release of macOS Ventura and iOS 16, though initial versions offered partial functionality for certain features.[33] Full support, including broader compatibility with AVIF variants, was solidified in Safari 16.4 and later, retroactively extending to older macOS versions like Monterey and Big Sur. By November 2025, AVIF enjoys approximately 94% global browser support among users of major web browsers, driven by adoption in Chrome, Firefox, Safari, and Edge, which collectively dominate market share.[34] Developers are advised to implement fallback mechanisms, such as the HTML<picture> element, to serve alternative formats like WebP or JPEG to the remaining unsupported browsers, ensuring graceful degradation.
A notable limitation of AVIF in web browsers is the absence of progressive loading, where images cannot display incrementally as data streams in; instead, the full file must be downloaded and decoded before any visual output appears.[35] This requires complete decoding prior to rendering, potentially impacting perceived performance on slower connections compared to formats like JPEG that support progressive refinement.[36]