Static vs Dynamic HDR Metadata Static vs Dynamic HDR Metadata

How static metadata works

HDR10 uses static metadata: one unchanging set of data applied to the entire program rather than adjusted per scene. That static metadata includes three things: color volume information, the average light level of the brightest frame in the program, and the maximum light level of the brightest pixel anywhere in the program. Because it is a single, program-wide data set, static metadata generates one tone-mapping curve that is used for every scene, even though maximum light levels vary widely from scene to scene _{-- a skyscraper scene with 4,000-nit highlights and a dim interior scene get the same curve}. Like any static-metadata system, this is a one-size-fits-all approach that limits flexibility, which is the specific limitation HDR10+ was created to address.

How dynamic metadata works

Dynamic metadata systems -- HDR10+ and Dolby Vision -- attach additional data on a scene-by-scene or frame-by-frame basis rather than once for the whole program. HDR10+ enables consumer devices to optimize tone mapping scene-by-scene and frame-by-frame by indicating the maximum light level for each individual scene, letting the display build a tone-mapping curve that tracks the creative intent more closely, particularly in low-light scenes where more shadow detail becomes visible. HDR10+ mastering goes further than just a light-level number: it gathers "luminance distribution" statistics describing all grayscale levels in a scene, including the exact pixel count across nine different grayscale ranges, and uses that to construct a customized "guided curve" tone-mapping curve per scene, which is included in the metadata. Despite each frame containing millions of pixels, this dynamic metadata adds only a few hundred bits per frame, keeping it cheap to encode and distribute.

Dolby Vision uses dynamic metadata to map content on a shot-by-shot or scene-by-scene basis to the specific target TV or playback device, using "trim" controls together with metadata generated by the colorist to preserve artistic intent from the color suite to the consumer's screen. Dolby's trims include parameters such as Lift, Gamma, Gain, Saturation, and Chroma Weight, and each set of trims is specific to a target display class, defined by that display's nits and colorspace -- meaning the same dynamic metadata stream can drive different tone-mapping outcomes depending on which display class it's being mapped to. Dolby's HDR research contributed to two SMPTE standards: ST 2084, the PQ luminance curve used by HDR displays, and ST 2094, which governs dynamic metadata.

The underlying standards

Static metadata corresponds to SMPTE ST 2086 (mastering display color volume), together with two related light-level values: MaxFALL (maximum frame-average light level) and MaxCLL (maximum content light level). Dynamic metadata corresponds to the SMPTE ST 2094 family of standards. Dolby Vision content typically carries both: static metadata (SMPTE ST 2086) for baseline mastering-display colorimetry alongside dynamic metadata (SMPTE ST 2094-10, Dolby's own format) for each scene or frame. Dynamic metadata layers on top of a static baseline rather than replacing it.

How this plays out on a display

Every consumer display does its own internal processing to adjust the highest PQ luminance levels down to fit within that display's actual brightness capability -- a step called tone mapping. With static metadata, the display's tone mapping gets the information it needs to retain maximum highlight detail while using the display's full specific brightness, but only as a single program-wide value. With dynamic metadata, the display instead gets per-scene guidance: for Dolby Vision, when the consumer display has lower color volume than the mastering display, the content is adjusted to that consumer display's actual capability based on the dynamic metadata rather than a single fixed curve. The mechanism is well documented; independent lab measurement of exactly how much visible difference this makes on any specific lower-brightness display was not part of the sourcing for this entry.

Backward compatibility runs in HDR10+'s favor by design: as part of the HDR10 ecosystem, HDR10+ content works on HDR10-only devices, which simply ignore the dynamic metadata and fall back to whatever static metadata is present, while HDR10 content plays as intended on HDR10+ devices.

Common confusions

Dolby Vision does not replace HDR10 -- it is typically delivered alongside static metadata (SMPTE ST 2086) plus its own dynamic metadata (SMPTE ST 2094-10) layered on top, and grading a title for Dolby Vision produces all needed deliverables -- SDR Rec. 709, generic HDR10, and Dolby Vision -- from a single hero grade rather than separate grading passes. Having dynamic metadata present does not by itself guarantee a better picture: both HDR10+'s and Dolby Vision's dynamic metadata still depend on the quality of the source mastering and on how well a given display's tone-mapping engine actually uses that metadata. And the two dynamic systems are not identical in what they measure -- HDR10+ metadata is built from grayscale/luminance-distribution statistics computed per scene into a guided tone-mapping curve, while Dolby Vision's dynamic metadata takes the form of colorist-set or generated "trims" (Lift, Gamma, Gain, Saturation, Chroma Weight) mapped to specific target-display classes.