Preserve relationships. Not simply cut the peak.

Range mapping is needed when image representation and output conditions differ: HDR to SDR, PQ on a lower-capability display, or multiple outputs from one working master. The current image state matters more than the word HDR in a project name.

Identify input units first. Scene-linear values, normalised PQ and cd/m² are different scales. This lab accepts synthetic display luminance from 0 to 4000 cd/m²; its curve cannot be applied directly to camera log.

Hard clipping preserves values below a limit and makes all higher values identical. It is easy to verify but removes highlight differences; lowering Gain afterwards cannot recover them.

Touching the limit is not always an error: a deliberately blown emitter may be intentional. The problem is accidental clipping replacing a controlled rendering and erasing useful gradation. The lab compares this loss against other trade-offs.

A shoulder gradually reduces slope in highlights. Inputs remain ordered but move closer together: preserving order does not preserve contrast. A shoulder starting too early can flatten bright surfaces.

The lab’s exponential curve follows a straight line up to K and smoothly approaches P, with a continuous join slope. This is a verifiable teaching model, not a Resolve or standards implementation. It exposes parameter-to-value relationships but has no spatial or colour adaptation.

A global mapping applies the same rule to identical inputs everywhere. Local methods may use spatial context to preserve detail, at the cost of possible halos, texture changes and temporal instability.

A pleasing still does not validate a video conversion. Inspect motion, edits, newly appearing highlights and subtitles. All three methods on this page are global; a pixel’s surroundings do not change its mapping rule.

The selected white is an anchor, but later stages can alter it. If scaled white enters the shoulder, actual output is lower than the pre-compression setting. Showing only the setting can conceal this consequence.

The lab reports both W and the actual output for input 203. A lower target peak may require a lower white to retain highlight room. This is a task-dependent trade-off, not a rule to keep 203 cd/m² in every SDR output.

A PQ display adaptation can use an EETF that accounts for source and destination black and white. This is a defined stage in a signal chain, not a modification of ST 2084 constants that still claims to be the original EOTF.

The current BT.2408-9 worked EETF example is in Annex 5. This page’s exponential shoulder does not reproduce its spline or black mapping. The reference supports comparison; conformance requires the complete specified method and test conditions.

HDR to SDR often changes white-to-peak relationships, contrast, viewing conditions and available colour volume. A good neutral curve does not guarantee convincing saturated emitters. Evaluate tone and colour together.

An “HDR to SDR” preset is not final approval. Review faces, logos, neon, gradients and adjacent shots. BT.2446 describes different methods; the source state, target and acceptable trade-offs guide selection.

Creative grading establishes character; output rendering adapts it to a destination. They interact: strong pre-shoulder contrast changes which values get compressed. The technical stage is not visually neutral packaging.

Separating responsibilities still helps change destinations without repairing every earlier node. Do not conceal a duplicate conversion with a creative curve. Correct the pipeline first, then assess intent.

For a nonlinear function, f(kL) usually differs from kf(L). Gain before mapping shifts values relative to the shoulder; gain after mapping scales an already compressed result and may exceed the target again. Equal Gain settings at different stages do different jobs.

A second mapping receives an already rendered image, not the original HDR. Further conversions can be legitimate, but should be intentional. Label signal states at each boundary.

In Resolve, determine whether project colour management or a chosen conversion node owns output mapping. Its input and output definitions must be explicit. The task is to process the current representation correctly, not an assumed source.

One responsible output stage does not forbid several CSTs across a project. Input conversion, element processing and final output can be distinct. Avoid unintended duplication of the same operation rather than blindly forbidding multiple transforms.

Values below the ceiling are not enough. Check ordering, important highlight distinctions, white placement, shadows and coloured highlights. Heavy compression may preserve mathematical order while making differences nearly invisible.

Review motion and edits. Keep reference outputs, comparison conditions and reasons for trade-offs across formats. The lab explains global-curve consequences, not the appearance of your complete production.