Static HDR Metadata in AV1 Containers

This article provides an overview of the static High Dynamic Range (HDR) metadata formats supported for direct transport within AV1 bitstreams and associated container formats, such as ISOBMFF (MP4) and Matroska (MKV). It details how standard static metadata—specifically SMPTE ST 2086 and CTA-861.3, which together form the foundation of HDR10—is packaged, signaled, and preserved across both the elementary video stream and container structures to ensure accurate video rendering.

Core Static HDR Metadata Formats

The primary static HDR metadata formats transported directly in AV1 workflows are the standards that comprise baseline HDR10:

SMPTE ST 2086 (Mastering Display Color Volume - MDCV):
Defines the color primaries (Red, Green, Blue), white point chromaticity coordinates, and the nominal minimum and maximum luminance of the display used to master the content.
CTA-861.3 / CTA-861-G (Content Light Level Information - CLLI):
Defines the maximum content light level (MaxCLL), indicating the brightest individual pixel in the program, and the maximum frame-average light level (MaxFALL), indicating the highest average brightness of any single frame.

Bitstream-Level Transport: Open Bitstream Units (OBUs)

AV1 allows static HDR metadata to be embedded directly into the video elementary stream using dedicated Metadata Open Bitstream Units (OBUs). This ensures the metadata remains coupled with the video frames regardless of the outer container:

METADATA_TYPE_HDR_MDCV (Type 1): Encapsulates SMPTE ST 2086 data. It specifies the chromaticity coordinates \((x, y)\) of the color primaries and white point, along with the display's mastering luminance range.
METADATA_TYPE_HDR_CLL (Type 2): Encapsulates CTA-861.3 data, conveying the 16-bit integer values for MaxCLL and MaxFALL in candelas per square meter (\(\text{cd/m}^2\) or nits).

Container-Level Transport: ISOBMFF (MP4)

When AV1 is encapsulated in MP4 or other ISO Base Media File Format (ISOBMFF) files, static HDR metadata can be carried in standardized container boxes at the visual sample entry level (inside av01 or the track's sample description):

mdcv (Mastering Display Colour Volume Box): Carries the SMPTE ST 2086 parameters directly at the container level to allow demuxers to identify display requirements without parsing the bitstream.
clli (Content Light Level Information Box): Carries the CTA-861.3 values (max_content_light_level and max_pic_average_light_level).

Container-Level Transport: Matroska (MKV) and WebM

In Matroska and WebM containers, static HDR metadata is carried within the track's Video element under the Colour sub-element:

MasteringMetadata: Translates SMPTE ST 2086 parameters into individual Matroska elements, including PrimaryRChromaticityX/Y, WhitePointChromaticityX/Y, LuminanceMax, and LuminanceMin.
MaxCLL and MaxFALL: Dedicated integer elements directly under Colour to store CTA-861.3 content light levels.

Summary

Direct static HDR metadata carriage in AV1 environments is restricted to SMPTE ST 2086 (MDCV) and CTA-861.3 (CLLI). These formats can exist redundantly or independently at both the elementary bitstream level (via AV1 Metadata OBUs) and the container level (via ISOBMFF boxes or Matroska elements) to guarantee broad playback device interoperability.