AV1 Dynamic HDR: HDR10+ and Dolby Vision Support

This article provides a technical overview of how the open-source AV1 video codec incorporates dynamic High Dynamic Range (HDR) metadata, specifically focusing on HDR10+ and Dolby Vision. While legacy static HDR (HDR10) applies a single tone-mapping curve across an entire video, dynamic formats adjust brightness and color frame-by-frame or scene-by-scene. The following sections explain the bitstream architecture, standardization protocols, and container mechanisms AV1 uses to transport and synchronize this dynamic metadata.

The Architecture: Open Bitstream Units (OBUs)

AV1 organizes compressed video data into modular payloads called Open Bitstream Units (OBUs). Unlike older codecs that rely heavily on Supplemental Enhancement Information (SEI) messaging, AV1 handles non-video data through dedicated OBU_METADATA structures.

Dynamic HDR metadata does not alter the underlying compressed video samples (typically 10-bit YUV 4:2:0 encoded using the BT.2020 color space and the Perceptual Quantizer, or SMPTE ST 2084). Instead, the metadata resides alongside frame data within these metadata OBUs. When an AV1 decoder processes the elementary stream, it extracts the metadata OBU synchronized with a specific temporal unit (a frame or set of frames) and passes the parameters directly to the display processor for tone mapping.

How AV1 Carries HDR10+ Metadata

HDR10+ relies on the open standard SMPTE ST 2094-40, which defines dynamic metadata containing color saturation and peak brightness parameters per scene.

In AV1, HDR10+ metadata is packaged using ITU-T Recommendation T.35:

OBU Identification: The bitstream defines an OBU_METADATA with the metadata type set to METADATA_TYPE_ITUT_T35.
Country and Provider Codes: The payload includes the specific country code (0xB5 for the United States) and terminal provider codes assigned to the HDR10+ standard.
Payload Injection: The actual SMPTE ST 2094-40 payload is placed into the T.35 data field.

Decoders that support HDR10+ parse this T.35 data to adjust luminance targets dynamically. Decoders or displays that do not support HDR10+ simply ignore the T.35 metadata OBU and fall back to baseline static HDR10 playback using the base color metadata found in the AV1 sequence header.

How AV1 Carries Dolby Vision Metadata

Dolby Vision utilizes proprietary Reference Processing Unit (RPU) dynamic metadata, which controls dynamic tone mapping, extended color gamuts, and target display adaptation.

AV1 accommodates Dolby Vision primarily through Dolby Vision Profile 10:

Bitstream Integration: Similar to HDR10+, Dolby Vision RPUs can be encapsulated in ITU-T T.35 metadata OBUs within the AV1 elementary stream.
Single-Layer Delivery: Profile 10 is a single-layer implementation. It relies on the standard AV1 10-bit Main Profile stream using the PQ transfer function (SMPTE ST 2084), while the RPU carries the proprietary dynamic mastering data.
Container Signaling: At the container level (such as MP4/ISOBMFF), the presence of Dolby Vision is signaled using specific sample entry extensions (e.g., the dvcC or dvvC configuration boxes paired with standard AV1 av01 sample descriptors). This informs the demuxer and media pipeline to prepare the Dolby Vision processing engine before decoding begins.

If a playback chain cannot process the Dolby Vision RPU, the system discards the metadata and maps the image according to standard HDR10 static characteristics, preventing playback errors or distorted colors.

Container Synchronization

While metadata OBUs allow AV1 elementary bitstreams to remain self-contained, media streaming protocols and file formats also coordinate this delivery:

MP4 (ISOBMFF): The ISO Base Media File Format defines specific visual sample entries to signal AV1 streams carrying dynamic HDR metadata, ensuring demuxers send metadata packages to display pipelines in sync with frame presentation timestamps (PTS).
Matroska (MKV) / WebM: Metadata OBUs pass through transparently inside the AV1 block payloads. MKV element tags can provide high-level signaling, but the synchronized dynamic data stays intact within the bitstream itself.

By implementing standardized ITU-T T.35 carriers within modular OBUs, AV1 decouples video decoding from display post-processing, allowing seamless, backwards-compatible support for both HDR10+ and Dolby Vision.