AVIF Content Light Level vs Mastering Metadata

High Dynamic Range (HDR) AVIF images rely on auxiliary metadata to ensure accurate tone mapping across displays with varying brightness and color capabilities. This metadata is primarily split into two distinct elements: Content Light Level (CLL) information and Mastering Display Color Volume (MDCV) metadata. While both guide the display engine on how to adapt the image, mastering metadata defines the physical capabilities of the monitor on which the image was originally authored, whereas Content Light Level info measures the luminance values present within the actual pixel data of the image itself.

Mastering Display Color Volume (MDCV)

Mastering metadata, stored in the mdcv box within an AVIF file (based on the SMPTE ST 2086 specification), characterizes the reference display used during the color grading and mastering process. It does not measure the image content; instead, it establishes the authoring environment's boundaries.

MDCV metadata contains the following fixed parameters:

Color Primaries: The CIE 1931 \((x, y)\) chromaticity coordinates for the red, green, and blue primaries of the reference monitor.
White Point: The CIE 1931 \((x, y)\) coordinates for the reference monitor's white point (typically D65).
Maximum Luminance: The peak brightness the mastering display could physically produce, measured in candelas per square meter (\(\text{cd/m}^2\) or nits), such as 1,000 or 4,000 nits.
Minimum Luminance: The black level floor the mastering display could reproduce (for example, 0.0005 nits for an OLED or 0.05 nits for an LCD).

Content Light Level Information (CLL)

Content Light Level information, stored in the clli box in AVIF (derived from CTA-861.3), provides statistical luminance data derived directly from the encoded image pixels. It conveys the real-world brightness profile of the picture.

CLL metadata consists of two values:

MaxCLL (Maximum Content Light Level): The maximum light level of any single pixel component (R, G, or B) across the entire image, expressed in nits. It identifies the single brightest highlight in the image.
MaxFALL (Maximum Frame Average Light Level): The average luminance of all pixels across the full image frame, also expressed in nits. For a still AVIF image, this represents the overall brightness of the scene.

Key Differences

Feature	Mastering Display Metadata (`mdcv`)	Content Light Level Info (`clli`)
Origin of Data	The physical hardware used by the colorist/creator.	The decoded pixel array of the AVIF image.
Color Information	Defines RGB chromaticity coordinates and the white point.	None; strictly measures luminance.
Measured Units	Coordinates \((x, y)\) and Luminance (\(L_{max}, L_{min}\)) in nits.	MaxCLL and MaxFALL in nits.
Display Role	Informs the target display what color space volume the content was judged in.	Informs the target display of the brightest highlight and average scene brightness.
Tone Mapping Effect	Prevents color shifts by indicating where the original monitor clipped or preserved color gamut boundaries.	Dictates how the target display compresses or tone-maps highlights to prevent clipping bright details.

How Tone Mappers Use Both

When an AVIF decoder and display system render an HDR image, they evaluate both sets of metadata:

Gamut Mapping via MDCV: If an image was mastered on a display with a DCI-P3 color space inside a BT.2020 container, the MDCV tells the target screen not to expect colors beyond P3, avoiding inaccurate gamut expansion.
Luminance Scaling via CLL: If an image is tagged with an MDCV maximum of 4,000 nits, but the clli reports a MaxCLL of only 800 nits, a 1,000-nit consumer display knows it does not need to compress the highlights. It can display the image natively without aggressive tone mapping, preserving the creator's intended contrast and shadow detail.