Dolby Vision is a proprietary HDR video format that encodes images with the PQ transfer function and carries dynamic metadata describing each shot or frame, so that every compatible display can tone map the content to its own capabilities. In HDR video, it sits alongside HDR10, HDR10+ and HLG as one of the main ways to deliver high dynamic range to cinemas, televisions, phones and streaming platforms.
The format was introduced by Dolby Laboratories in 2014 and is defined by a family of profiles and levels rather than a single codec. Its central idea is that the creative intent of the grade should survive on displays that range from a few hundred nits to several thousand nits of peak luminance.
How Dolby Vision carries creative intent
A Dolby Vision master is graded on a high-brightness reference monitor, usually at 1000 or 4000 nits, in a wide-gamut container such as Rec. 2020 or P3. During grading, analysis software measures every shot and stores its minimum, average and maximum luminance as Level 1 metadata.
The colorist then performs trim passes for lower target displays, typically 100 nits for SDR and intermediate targets such as 600 or 1000 nits. These trims are stored as Level 2 and Level 8 metadata and tell the playback device how to adjust lift, gain, saturation and mid-tone contrast when its peak luminance is lower than the mastering display.
At playback, the display manager in the television or set-top box reads the metadata and interpolates between the stored trims for its own peak brightness. This is the essential difference from static HDR: tone mapping is guided shot by shot instead of being guessed from a single set of values for the whole program.
| Metadata level | Content | Purpose |
|---|---|---|
| Level 0 | Mastering display and container | Describes the grading environment |
| Level 1 | Per-shot min, average, max luminance | Automatic analysis for tone mapping |
| Level 2 | Trim controls per target display | Creative adjustments, legacy CM v2.9 |
| Level 6 | MaxCLL and MaxFALL | Fallback values shared with HDR10 |
| Level 8 | Extended trims, CM v4.0 | Finer control including mid-tones and saturation |
| Level 11 | Content type hints | Lets displays choose suitable picture modes |
Profiles, bit depth and signal structure
Dolby Vision is delivered in several profiles, each defining codec, layer structure and compatibility. The profile decides whether a stream can also play as HDR10, SDR or HLG on devices that do not support Dolby Vision.
Early profiles used a dual-layer design: a base layer plus an enhancement layer that extended precision to 12 bits. Most streaming today uses single-layer profiles with 10-bit HEVC or AV1, where the metadata travels alongside the video as SEI messages or in the container.
| Profile | Codec and layers | Cross-compatibility | Typical use |
|---|---|---|---|
| Profile 5 | HEVC 10-bit, single layer, IPTPQc2 | None, Dolby Vision only | Streaming services |
| Profile 7 | HEVC, dual layer, up to 12-bit | HDR10 base layer | Ultra HD Blu-ray discs |
| Profile 8.1 | HEVC 10-bit, single layer | HDR10 fallback | Streaming and broadcast |
| Profile 8.4 | HEVC 10-bit, single layer | HLG fallback | Smartphone HDR capture |
| Profile 10 | AV1 10-bit, single layer | Variants with HDR10 or SDR | Newer streaming pipelines |
Profile 5 uses a proprietary IPTPQc2 color representation instead of standard Y’CbCr, which improves hue linearity but means that a non-supporting player shows strongly wrong colors, often green and purple. Profile 8.1 avoids this problem because its base layer is ordinary HDR10.
Dolby Vision compared with other HDR formats
The HDR formats differ mainly in transfer function, metadata type and licensing. Dolby Vision and HDR10+ both use dynamic metadata over PQ, HDR10 relies on static metadata, and HLG avoids metadata entirely by using a scene-referred, backward-compatible curve.
| Format | Transfer function | Metadata | Bit depth | Licensing |
|---|---|---|---|---|
| Dolby Vision | PQ (also HLG in profile 8.4) | Dynamic, per shot or frame | 10 or 12-bit | Proprietary, licensed |
| HDR10 | PQ | Static, MaxCLL and MaxFALL | 10-bit | Open standard |
| HDR10+ | PQ | Dynamic, per scene | 10-bit | Royalty-free with certification |
| HLG | Hybrid log-gamma | None | 10-bit | Open, broadcast oriented |
In practice, the quality advantage of dynamic metadata is largest on mid-range televisions with peak luminance of roughly 400 to 800 nits, where tone mapping has to compress highlights heavily. On a display that already reaches the mastering brightness, the formats look much more alike.
Producing and delivering Dolby Vision
A professional Dolby Vision workflow requires licensed grading tools, a reference HDR monitor and a delivery package that validates correctly. The usual deliverable is a 12-bit or 16-bit image sequence or ProRes master with a separate XML metadata file.
| Decision | Common choice | Reason |
|---|---|---|
| Mastering display | 1000-nit P3-D65 in Rec. 2020 container | Widely available reference monitors |
| Content mapping version | CM v4.0 | Better trims and mid-tone control than v2.9 |
| Trim targets | 100-nit SDR plus one HDR target | Covers most consumer displays |
| Streaming profile | Profile 8.1 or 5 | Balance of compatibility and efficiency |
| Quality control | Metadata validation and target playback | Catches shot cuts with wrong analysis |
Because analysis is stored per shot, the edit decision list must match the final cut exactly. A misaligned shot boundary applies the wrong metadata to neighboring frames and can cause visible brightness pumping on consumer displays.
Limitations and common misconceptions
Dolby Vision improves consistency, but it cannot add brightness or color a display does not have. Its success depends on correct mastering, a well-implemented display manager and a playback chain that passes the metadata intact.
- A Dolby Vision logo does not guarantee a bright image; a 350-nit television still tone maps heavily.
- Dynamic metadata is not automatic grading; the trims reflect choices that a colorist made or approved.
- HDMI chain problems such as a soundbar or capture device without support can silently drop the format to HDR10 or SDR.
- Low-latency Dolby Vision used by some players and game consoles moves tone mapping to the source device, so results vary by device.
In everyday language
Dolby Vision is a way of packing extra instructions with a movie so that every screen shows it as close as possible to what the filmmakers intended. Bright televisions and dim phones cannot show the same range of light, so something has to adapt the picture.
Instead of letting each screen guess, Dolby Vision gives it a scene-by-scene recipe: here the room is dark, keep the shadows; here the sun is bright, protect the highlights. The result is that the same film looks consistent on very different screens.
Seen in the real world
A post-production house finishes a feature film for a streaming release. The colorist grades in a Rec. 2020 container on a 1000-nit reference monitor, runs Dolby Vision analysis on every shot and then performs a 100-nit SDR trim and a 600-nit trim for typical living-room televisions.
The facility delivers a 16-bit image sequence with a CM v4.0 metadata XML. The streaming service encodes it as profile 8.1 so that viewers without Dolby Vision still receive a proper HDR10 picture. Viewers with a 700-nit television see night scenes with intact shadow detail and sunlit exteriors without clipped skies, because their set follows the colorist’s trims instead of a generic curve.