How ReCAP Handles Dolby Vision And HDR Metadata

High-dynamic-range video carries more information than a brighter picture. Alongside the pixels, a file or live stream may contain colour primaries, transfer characteristics, mastering-display details, maximum light levels and, in Dolby Vision, dynamic metadata that can vary from scene to scene. A video analysis platform must preserve, interpret and validate these signals while it performs other tasks such as face recognition, logo detection and duplicate-content matching.

ReCAP’s role is to treat HDR metadata as part of the media asset rather than as an optional label. Its processing pipeline can inspect the container, codec and stream descriptors, identify the HDR format, and associate those findings with technical and semantic metadata. This creates a more reliable record for broadcasters, production teams and media asset managers.

That approach matters in Australia, where viewers move between free-to-air broadcasts, sports streams, catch-up services and subscription platforms on televisions with very different HDR capabilities. A master prepared in Sydney may be reviewed in Perth, delivered over the National Broadband Network and later repurposed for mobile playback. Consistent analysis helps prevent format confusion across that chain.

Video characteristic Dolby Vision HDR10 HLG
Metadata model Static base metadata plus optional dynamic metadata Static metadata, usually carried with the stream Designed for broadcast compatibility, generally without mastering-display metadata
Transfer function PQ-based PQ-based Hybrid Log-Gamma
Typical use Premium streaming, UHD discs and selected broadcast workflows UHD delivery and consumer distribution Live television and broadcast contribution
Processing priority Preserve profile, level, colour data and dynamic metadata Validate mastering and signal-range information Confirm HLG signalling and avoid incorrect SDR conversion
Common risk Dynamic metadata is stripped or mismatched HDR is labelled correctly but displayed with poor tone mapping HLG is treated as ordinary SDR or converted unnecessarily

What the pipeline needs to identify

The first stage is stream inspection. ReCAP can examine the media container and video bitstream to determine whether the content is SDR, HDR10, HLG or a Dolby Vision profile. It can also look for indicators such as BT.2020 colour primaries, PQ or HLG transfer characteristics, mastering-display colour volume and content-light-level values. These fields provide the technical context required to interpret the image correctly.

Dolby Vision adds another layer. Depending on the profile and delivery format, the video may contain a compatible HDR10 base layer accompanied by dynamic metadata, or it may use a more specialised arrangement. The dynamic information can describe scene-level or frame-level guidance for tone mapping. An analysis system therefore needs to distinguish the base video from the Dolby Vision metadata instead of treating every HDR signal as interchangeable.

Metadata extraction should preserve the original values and record how they were found. A missing field is different from a field that states a value of zero, and an unrecognised Dolby Vision profile is different from an SDR file. ReCAP’s broader metadata model can represent these distinctions so that downstream users can filter assets, investigate anomalies and avoid making automated decisions from incomplete information.

Reading HDR without changing the picture

A central principle is to separate analysis from transformation. ReCAP can inspect the source signal without immediately converting it to SDR, applying a new tone map or rewriting the mastering metadata. This is important because a conversion may improve compatibility while also changing highlights, shadow detail and colour relationships.

For computer-vision tasks, the platform may create a controlled analysis rendition. For example, a face detector or logo recogniser can work on a normalised preview while the original Dolby Vision or HDR10 asset remains untouched. The rendition should carry a clear record of its colour conversion, transfer function and scaling decisions. That makes it possible to understand whether a missed logo resulted from the content itself or from the preview-generation process.

Tone mapping requires particular care. A bright yellow sign, stadium floodlight or sunlit beach can clip when PQ content is forced into an SDR range. Conversely, a dark scene can lose detail if the conversion is too conservative. ReCAP can use the source metadata to select an appropriate processing path and flag cases where the declared HDR characteristics do not match the measured image statistics.

This separation is useful in Australian workflows involving high-contrast material, from an Australian Open night match to outdoor news footage recorded in strong Queensland sunlight. Analysts can review a practical proxy while retaining the source-quality evidence needed for mastering, compliance and later distribution.

Detecting metadata errors and format drift

HDR metadata is only valuable when it agrees with the video. A file can declare BT.2020 primaries while containing an effectively SDR image, or identify PQ transfer characteristics while carrying levels that suggest an incorrect range conversion. ReCAP can compare declared properties with sampled frames and signal statistics to identify such inconsistencies.

The system can also monitor metadata over time in a live feed. If a broadcast changes from SDR to HLG during an event, or if a contribution stream unexpectedly switches profile, the transition can be recorded as an event. Sudden changes in frame rate, resolution, colour space or transfer function may indicate an encoder fault, an incorrect source route or an operational handover.

Dolby Vision metadata presents an additional validation challenge because dynamic instructions may be absent from some segments, duplicated, malformed or stripped during transcoding. ReCAP can report whether the expected metadata is present and whether it remains aligned with the corresponding video frames. It should preserve the distinction between “not detected”, “not expected for this profile” and “expected but missing”.

These checks support media asset management as well as live operations. An Australian streaming catalogue may contain several versions of the same programme for connected televisions, mobile applications and regional delivery. Reliable HDR records make it easier to select the right master and to identify a bad transcode before it reaches viewers.

Connecting HDR analysis with visual intelligence

HDR processing becomes more useful when it is combined with content understanding. ReCAP is designed to extract metadata such as faces, logos, scenes and duplicated material, so its HDR-aware workflow can provide the visual-analysis services with consistent inputs. A detector should receive a properly interpreted frame rather than an image whose highlights or colour channels have been accidentally distorted.

Logo recognition illustrates the issue. A network watermark may sit in a bright corner of an HDR frame, while sponsor branding appears on a dark stadium board. If a preview clips the bright area or crushes the shadows, the recognition result may be unreliable. A colour-managed analysis rendition, with confidence scores and a link back to the original timecode, gives operators a stronger basis for review.

The same logic applies to face analysis. HDR itself does not identify a person, but extreme luminance differences can affect facial visibility and detector performance. ReCAP can retain the original frame characteristics, create a consistent working image and attach technical context to each detection. Where biometric or personal information is involved, the workflow should also enforce access controls, retention rules and purpose limitations.

For Australian organisations, that governance may intersect with the Privacy Act 1988 and the Australian Privacy Principles, especially when face data is used beyond simple editorial indexing. A broadcaster or archive operator should document why the analysis is performed, who can access the results and how long derived biometric information is retained. HDR metadata does not remove those obligations; it helps make the processing environment more auditable.

Handling live feeds and broadcast-quality evidence

In live production, the analysis system must keep pace with the feed. ReCAP can inspect incoming segments or frames, update technical metadata and raise alerts without requiring the entire programme to be processed first. Low-latency checks can focus on format changes, missing HDR signalling, corrupted packets and visible quality issues, while deeper analysis can run after the event.

Compression is especially important when HDR content travels through contribution links, playout systems and online encoders. Banding in a sunset, block errors around fast movement and ringing near text can become more noticeable when high dynamic range is compressed aggressively. ReCAP’s work on compression artifact detection shows how quality monitoring can complement HDR metadata validation.

The platform can correlate a quality alert with the relevant HDR properties. For example, an operator may see that a spike in blockiness occurred immediately after a Dolby Vision stream was converted to a different delivery profile. That relationship is more informative than a generic “poor quality” warning because it points towards a particular encoder, transcode or signal-path decision.

This is relevant to Australian live broadcasting, where major sporting events can serve viewers in Melbourne and Sydney while regional audiences use different distribution paths. A feed that looks acceptable on a production monitor may show clipping, banding or incorrect tone mapping on a consumer television. Automated evidence gives engineering teams a way to compare signals across the chain rather than relying only on subjective viewing.

Supporting archives, versions and editorial decisions

A well-structured HDR record helps an archive distinguish between an original camera master, a graded mezzanine file, a Dolby Vision distribution version and an SDR derivative. ReCAP can associate these technical variants with shared content identifiers, timecodes and perceptual fingerprints. Duplicate detection can then recognise related versions without assuming that different resolutions or colour formats are unrelated programmes.

This matters when a media library contains promotional clips, news packages and sponsored segments. A casino promotion, for instance, may appear in several edits with different logos, voiceovers or legal disclaimers; an example of this type of branded media is discussed in the casino video example. HDR-aware comparison helps distinguish a genuine duplicate from a new version created through grading, cropping or delivery conversion.

Editors and rights teams can use the extracted information to find assets by format, brightness characteristics, detected brands or distribution suitability. A request for “HLG sports footage suitable for broadcast” can be answered more reliably when the archive stores verified transfer characteristics rather than relying on filenames. Similarly, a Dolby Vision asset can be excluded from a pipeline that accepts only HDR10 or SDR.

Metadata should remain understandable to humans as well as machines. Technical fields such as mastering luminance and Dolby Vision profile are valuable, but operators also need clear statuses such as “HDR metadata consistent”, “dynamic metadata unavailable” or “conversion required”. ReCAP’s purpose is strongest when its analysis supports practical decisions in production rooms, broadcast control centres and asset-management systems.

Practical recommendations for Australian media workflows

Organisations deploying HDR analysis should design the process around preservation, verification and controlled rendering. The following practices reduce avoidable errors:

A production team should also test the complete delivery chain. The file may be technically correct at ingest but lose dynamic metadata during editing, packaging or playback. Testing across professional monitors, consumer HDR televisions, mobile screens and ordinary SDR displays can reveal whether the metadata survives and whether the fallback image remains acceptable.

The commercial context is changing quickly. Australian viewers increasingly use connected televisions and streaming applications, yet many households still receive free-to-air services and mixed-format content. A practical ReCAP workflow therefore needs to support graceful conversion rather than assume that every endpoint understands Dolby Vision. The most useful output is a traceable decision record showing what the source contained, what processing occurred and which version was delivered.

HDR metadata is best understood as operational evidence. It tells ReCAP how the image was authored, how it should be interpreted and where a transformation may have taken place. When that evidence is joined with quality monitoring, duplicate detection and visual recognition, media teams gain a more complete picture of each asset.

The key point to remember is that Dolby Vision and other HDR formats are not just display settings: their metadata must travel with the video, be checked against the image and remain visible throughout the production and distribution workflow.