Dolby Vision & derived deliverables

ExpertDuration ~15 min video + hands-onTools DaVinci Resolve Studio, HDR display recommended

Why this lesson

In 3.6 you graded HDR and derived an SDR trim as a second file. Dolby Vision does the same job a smarter way: it grades once in HDR, then carries a small bundle of metadata that tells every downstream display — a 4000-nit reference, a 600-nit TV, a 100-nit SDR set — how to map your grade to its capability, on the fly, with no extra masters. Authoring the trims conceptually (L1, L2) and QC’ing the result is a core finishing skill and a paid-course moat. The explainer carries the concepts; the Dolby video carries the QC pass you’ll actually run.

Watch

Watch for: The official QC workflow, which doubles as a tour of the pieces. Metafier validates the Dolby Vision XML (the 'validate with lift' check for stray letterbox lift); MediaInfo confirms the file's color primaries and that the transfer characteristic is PQ; then in Resolve the Dolby Vision palette's 'Tone Map using' menu (Custom / Embedded / Original metadata), the target-display dropdown, and the tone-mapping preview let you QC the HDR and the derived SDR against each other. Watch how the SDR is checked by switching the target to 100-nit BT.709, not by opening a separate file.

The explainer

Dolby Vision is dynamic metadata over a single HDR master. HDR10 (from 3.6) ships one static MaxCLL/MaxFALL for the whole program, so a modest display has to guess how to tone-map two hours of content from two numbers. Dolby Vision instead analyzes and stores metadata per shot, so each scene maps optimally: a dark scene and a blazing exterior get different mapping instructions. The HDR master doesn’t change — the metadata rides alongside it in an XML sidecar (or interleaved in the file), and the playback device does the mapping. This is why Blackmagic calls the SDR output a derived master: there’s no second render, just the HDR file plus metadata, mapped live on the viewer’s device.

L1 — automatic analysis. The base layer is L1: Dolby’s automatic per-shot measurement of three luminance values — the minimum, the mid (average), and the maximum brightness in each shot. You generate it by running an analysis in Resolve’s Dolby Vision palette, which walks the timeline and writes L1 for every shot. L1 alone gives a competent automatic mapping to any target display; it’s the floor, and for a lot of content it’s enough.

L2 — the colorist’s trims. Where you earn your keep is L2 trims. An L2 trim is a manual adjustment authored for a specific target display — most importantly the 100-nit Rec.709 SDR target, but also intermediate HDR targets like 600 nits. You tell the tone-mapper “when you map my 1000-nit HDR grade down to SDR, lift this shadow, pull this specular back, add a touch of saturation here.” These are the lift/gamma/gain/saturation (and more) controls in the Dolby Vision palette, stored as metadata against that target. Crucially, a trim is not a re-grade and not a new file — it’s guidance the mapper obeys. Author good L2 trims for the SDR target and your SDR “derived master” looks intentional instead of like a machine crushed your highlights. (Higher levels exist — L3, L5, L8 for finer control — but L1 analysis + L2 SDR trims is the working core.)

The CMU — what actually does the mapping. The CMU (Content Mapping Unit) is the engine that takes the HDR master plus the L1/L2 metadata and produces the tone-mapped output for a chosen target. Historically this was a dedicated Dolby hardware box in the suite, driving a second reference monitor so the colorist could see the SDR trim live. Modern Resolve does it internally — the “tone-mapping preview” in the QC video is the CMU running in software — but the term persists in specs and older workflows, and knowing it means “the mapping engine” keeps you fluent when you meet it.

The XML and QC. The whole thing is carried by a Dolby Vision XML: shot boundaries, L1 analysis, and every L2 trim, per target. QC has three parts, exactly as the video shows. First, validate the metadata with Metafier — a clean XML with no stray lifts or out-of-range values. Second, check the file metadata with MediaInfo — color primaries present, transfer characteristic reads PQ, format matches the spec. Third, visual QC both renderings: the HDR at its mastering target, and the derived SDR by setting the target display to 100-nit BT.709 and eyeballing that the trims hold — nothing clipping, nothing lost, metadata events synced to the cuts. You can also spot-check an intermediate target (say 600 nits) to see how a mid-tier TV will fare.

Why this is the professional default for premium HDR. One graded master, one metadata file, correct rendering on every class of display, and an SDR version that inherits your creative intent without a separate deliverable to grade, track and QC. That efficiency — plus the guarantee of intent-preserving downmaps — is why Dolby Vision is the standard on premium streaming, and why authoring L2 trims is a skill facilities pay for.

Do it

DCTL-free but Studio-only: Dolby Vision authoring needs Resolve Studio and, to finish for real, an HDR display. Without one, run the workflow with tone-mapping preview and treat the SDR judgement as approximate (per 3.6’s honesty).

1. On your HDR-graded short from 3.6, open the Dolby Vision palette and run a full L1 analysis across the timeline.
2. Set the target display to 100-nit BT.709 and enable tone-mapping preview — this is the derived SDR, mapped live by the software CMU.
3. Author L2 trims on three shots: rescue a shadow that fell dark, pull a specular back under clipping, nudge a color. Confirm the HDR is untouched — you’re guiding the map, not re-grading.
4. Export the Dolby Vision XML, then QC it: validate with Metafier if available (or Resolve’s own checks), confirm PQ in the file metadata, and visually QC HDR vs derived SDR against each other.

HDR quick reference — PQ/HLG, nits, MaxCLL/MaxFALL, Rec.2100, Dolby Vision L1/L2/CMUhdr-quick-reference.pdf123 KBOriginal course material — free to useLevel 3 workbook — every Do it exercise, 3.1–3.10, plus the capstone (printable)level-3-workbook.pdf764 KBOriginal course material — free to use

Terms introduced

Terms introduced in this lesson

• Dolby Vision
• L1 / L2 trims (trim pass)
• CMU (Content Mapping Unit)

Each term links to the glossary, where it is defined by the way you just used it.

Check

Check yourself

1. How is Dolby Vision different from static HDR10?

   It only works in SDRIt has better resolutionIt carries dynamic, per-shot metadata (not one fixed MaxCLL/FALL for the whole program), so the display tone-maps each scene optimally to its own capabilityIt doesn’t use nits

2. What is the L1 metadata in Dolby Vision?

   The colorist’s manual creative adjustmentThe automatic per-shot analysis — the min, mid (average) and max luminance Dolby measures from each shot — that drives the base mappingThe subtitle trackThe audio loudness

3. What is an L2 trim?

   A type of LUTA second video file you exportA camera settingThe colorist’s manual adjustment (lift/gamma/gain/sat and more) for a specific target display — e.g. how the HDR grade should look when mapped to 100-nit Rec.709 SDR

4. What does the CMU (Content Mapping Unit) do?

   It edits the soundIt applies the Dolby Vision mapping — taking the HDR master plus the L1/L2 metadata and producing the tone-mapped output for a given target; historically a hardware box, now done inside ResolveIt stores the LUTIt conforms the timeline

Check answers

You can move on when you can… explain Dolby Vision as dynamic per-shot metadata over one HDR master, distinguish L1 analysis from L2 trims, say what the CMU and the XML are, author an SDR trim conceptually, and run the three-part QC (Metafier / MediaInfo / visual).

Mark this lesson complete

Go deeper

Blackmagic / Cullen Kelly — Dolby Vision as a derived master: the segment that frames DV as the reverse of an SDR-to-HDR trim — take an HDR grade, enable Dolby Vision, analyze a frame, and watch the correct SDR appear via best-fit mapping, all as metadata rather than a second file.

Segment: 27:04–end — enabling Dolby Vision, target-display config, the derived-SDR conceptwatch full video

Watch for: Enabling Dolby Vision, the target-display dropdown, 'analyze frame,' and the point that the HDR→SDR mapping travels as metadata with the master and happens on the playback device — no second physical file.

Read

• Dolby — Dolby Vision for Content Creators / professional docs (professional.dolby.com): the authoritative reference for the metadata levels (L1–L8), the mastering workflow, and the CMU.
• Netflix Partner Help — Dolby Vision delivery (partnerhelp.netflixstudios.com): the delivery-side spec and QC expectations for a real DV master.

The paid path: hands-on Dolby Vision authoring — building L2 trims across a whole show on a licensed setup — is a classic Mixing Light moat (their HDR/Dolby Vision series). This page gives you the concepts and the QC pass; the paid training plus a Dolby license is where you learn to trim a season.

Next up: 3.8 · Deliverables & QC — packaging masters as IMF/DCP and QC’ing them against a streamer spec.