Can AVIF Support Dynamic HDR Like SMPTE ST 2094?

The AV1 Image File Format (AVIF) can encapsulate dynamic High Dynamic Range (HDR) metadata, including formats specified under SMPTE ST 2094-10. Because AVIF is structurally derived from the ISO Base Media File Format (ISOBMFF) and wraps native AV1 bitstreams, it inherits flexible mechanisms for transporting color metadata. Dynamic metadata can be carried either embedded directly within the AV1 payload via Open Bitstream Units (OBUs) or mapped as metadata properties within the ISOBMFF container itself.

The Underlying Architecture: ISOBMFF and AV1

AVIF achieves image compression by packaging an AV1 still frame within an ISOBMFF container (standardized under ISO/IEC 23000-12 / HEIF). While static HDR metadata—such as SMPTE ST 2086 mastering display data and CTA 861.3 content light levels—is commonly parsed via standard container boxes like mdcv and clli, dynamic metadata requires a framework capable of handling scene-by-scene or frame-by-frame color adjustments.

SMPTE ST 2094-10 defines dynamic metadata application for color volume transform (originally associated with Dolby Vision). AVIF accommodates this dynamic paradigm through two primary pathways:

1. Bitstream-Level Encapsulation via Metadata OBUs

The native and standard-compliant method to carry SMPTE ST 2094-10 within an AVIF is inside the AV1 bitstream itself. The AV1 format architecture includes distinct packet structures called Open Bitstream Units (OBUs).

Within AV1, metadata is signaled via METADATA_TYPE_ITUT_T35. Dynamic HDR metadata profiles, including SMPTE ST 2094-10 and SMPTE ST 2094-40 (HDR10+), define standardized payloads registered under ITU-T Recommendation T.35 country codes and provider codes. When an encoder processes an image, it injects these T.35 metadata OBUs alongside the sequence headers and tile data. Because an AVIF container stores raw AV1 image data, it natively preserves these metadata OBUs untouched.

2. Container-Level Encapsulation via ISOBMFF Properties

ISOBMFF allows metadata to be associated with an image item using item properties. In standard HEIF workflows, proprietary or registered boxes—such as Dolby Vision configuration boxes (dvcC or dvvC)—can accompany an item.

While common consumer AVIF implementations favor placing payload-critical metadata into the OBU stream, container-level association allows a parser to detect the presence of dynamic color volume transformations without fully parsing the compressed image bitstream first.

Parsing and Ecosystem Compatibility

Although the AVIF container format and the AV1 bitstream specification structurally allow SMPTE ST 2094-10 dynamic metadata, end-to-end support relies on the playback pipeline:

• Demuxing and Parsing: The reader must parse either the outer ISOBMFF metadata properties or pass the full bitstream to an AV1 decoder capable of extracting METADATA_TYPE_ITUT_T35.
• Tone Mapping: The image rendering pipeline must interpret the ST 2094-10 parameters (such as minimum, mid, and peak brightness curves) and apply the target transformation relative to the user's specific display capabilities.

In summary, AVIF fully supports the encapsulation of SMPTE ST 2094-10 dynamic HDR metadata, using the robust packaging capabilities of ISOBMFF and AV1's native metadata OBU framework.