Why It Looks Like That
Article 13

HDR Tone Mapping: How a Television Fits the Signal to the Screen

HDR tone mapping is part of display mapping, the controlled adaptation of luminance and color when a source exceeds the target television or viewing conditions.

By · Published · Updated

Direct answer

In brief

HDR tone mapping is one part of display mapping. It adapts PQ luminance and color that a target television cannot reproduce directly, using clipping, compression, color-volume mapping, metadata, image analysis, or combinations of these tools. No single curve is correct for every source, display, room, or goal. HLG uses a different relative system relationship, and HGiG coordinates game rendering with display behavior rather than simply turning all television mapping off.

An HDR signal can describe luminance and color that a television cannot reproduce directly.

The television must then:

That process is often called tone mapping.

A broader term is display mapping.

Tone mapping usually emphasizes the relationship between encoded luminance and displayed luminance.

Display mapping includes the larger task of adapting the source image to the target display and viewing environment.

That can require a color-volume transform, not just one brightness curve.

The main practical answer is:

Tone mapping is not a repair for a defective HDR signal. It is an adaptation used when the target display, viewing condition, or delivery goal differs from the conditions represented by the source.

The quality of that adaptation matters.

So do the source, panel, room, settings, and calibration.

Start with PQ

PQ is the HDR system in which the target mismatch is easiest to explain.

SMPTE ST 2084 defines an absolute reference EOTF.

Under that EOTF, a normalized PQ signal value maps to a defined display luminance between 0 and 10,000 cd/m². [4]

Suppose a source contains a PQ highlight corresponding to 3,500 cd/m².

Suppose the target television can reproduce only 800 cd/m² for that image area, duration, mode, and color.

The television cannot reproduce the reference luminance directly.

It needs to choose a target output.

“Can reproduce 800 cd/m²” needs conditions

A television does not have one universal peak number.

The available output can depend on:

The HGiG recommendations reflect this difference by defining separate full-frame and smaller-area tone-map luminance parameters. [7]

A target mapper therefore needs more information than one advertised peak-brightness number.

HLG is different

Not every HDR signal encodes an absolute display luminance.

HLG uses a relative scene-light signal.

Recommendation ITU-R BT.2100-3 defines an HLG display relationship that depends on:

The same HLG signal can therefore produce different absolute luminance on displays with different nominal peaks.

This is an intended part of the HLG system.

A manufacturer may still add processing, but the basic adaptation is not the same PQ mismatch described above.

The rest of this article concentrates mainly on PQ display mapping.

Side-by-side system diagram. PQ uses an absolute reference EOTF, which can create a mismatch when the target cannot reproduce the encoded color volume, leading to display mapping or clipping. HLG uses a relative scene-light signal whose displayed result depends on nominal peak, black level, system gamma, and viewing conditions.
PQ and HLG can both require careful display behavior, but they do not begin from the same signal-to-light architecture.

Tone mapping is part of display mapping

Recommendation ITU-R BT.2390-12 explains that PQ content may be optimized for a reference monitor and reference environment.

Real displays may have:

In those cases the display characteristic needs to change through display mapping. [1]

This is broader than preserving highlight detail.

The mapping may also address:

Reference display does not mean every pixel used the display maximum

Keep these quantities separate.

Mastering-display maximum

The configured maximum luminance of the display used during mastering.

Content maximum

The brightest encoded or measured content value under the selected definition.

MaxCLL

A static content-light metadata field intended to represent the maximum content light level.

MaxFALL

The maximum frame-average light level.

HDR Reference White

A nominal reference signal level. Under current ITU operational guidance it corresponds to 203 cd/m² on a PQ reference display. [2][3]

Target-display maximum

The luminance the receiving display can produce under the relevant size, duration, mode, color, and environmental conditions.

A 4,000 cd/m² mastering display does not prove that the program contains a 4,000 cd/m² highlight.

A target television's 1,000 cd/m² small-window result does not prove it can reproduce a 1,000 cd/m² full-screen image.

The target display has a color volume

Luminance is one axis.

Color also matters.

A display may reproduce an achromatic white highlight at one luminance while reaching much less luminance for a highly saturated red, green, or blue.

The source can exceed the target in:

A mapping that handles only white luminance can still produce clipped, shifted, or desaturated colors.

Conceptual three-dimensional diagram showing a larger source color volume extending beyond a smaller target television volume. The target has a lower peak, a higher black floor, and reduced saturated-color reach, so several source points lie outside the target.
A display can exceed or fall short of the source in several dimensions, so a luminance-only answer can be incomplete.

Option 1: hard clipping

Hard clipping sets every value above a threshold to the threshold.

If a target clips at 800 cd/m²:

This preserves the relationship below the clipping point.

It removes luminance distinctions above it.

What clipping can lose

A highlight can contain:

If all of that structure lies above the target threshold, hard clipping can flatten it.

A polished surface can become a uniform patch.

A lamp can lose internal detail.

A cloud can lose bright texture.

This is the familiar reason for using a roll-off.

Conceptual comparison of ordered source highlight steps and a clipped target output. Values below a target threshold remain distinct, while all values above the threshold become the same output level.
Hard clipping can preserve values below a chosen threshold while flattening highlight structure above it.

Hard clipping is not always an error

Clipping can be deliberate.

Report ITU-R BT.2408-9 says that hard clipping may be appropriate in some conversion circumstances.

The same report notes that reference-display color conversion may prefer hard clipping for some out-of-gamut values. [2]

A display or workflow may choose clipping when:

The correct question is not:

Does the display clip anything?

It is:

What is clipped, where is the threshold, and what source relationships are lost or preserved?

Option 2: uniform scaling

Another simple mapping multiplies all luminance values by the same factor.

If a 4,000 cd/m² source is scaled into an 1,000 cd/m² target, a factor of one-quarter would map:

The ordering remains.

The entire image becomes lower in absolute luminance.

What uniform scaling protects

Uniform scaling can preserve ratios.

It avoids one hard threshold that collapses all high values.

It can be useful when a complete image must be fit into a smaller range with a predictable proportional relationship.

What uniform scaling changes

A display that could reproduce many midtone values directly now places them lower.

That can change:

Uniform scaling is therefore not a free solution.

It is also not universally wrong.

It can be part of a deliberate transform or environment adjustment.

Conceptual diagram showing four ordered source luminance values multiplied by one constant factor smaller than one. All target values are reduced by the same proportion, so ordering is retained but midtones and highlights both move lower.
Uniform scaling preserves order, but it also lowers diffuse white, midtones, and highlights together.

Option 3: nonlinear mapping

A nonlinear mapping can preserve one region more closely and compress another.

For PQ display mapping, a common conceptual structure is:

Report ITU-R BT.2408-9 describes an example EETF with a central 1:1 region and toe and knee controls for adapting PQ signals to displays with different black and white luminance. [2]

EETF means electrical-electrical transfer function.

It maps one encoded signal relationship into another before the target EOTF produces light.

The 1:1 region

A 1:1 region leaves selected source values unchanged in the display-light relationship.

This can protect:

The exact protected region depends on the mapping goal.

The knee

The knee is where highlight compression begins.

An earlier knee creates more room for a gradual roll-off.

It also alters more source values.

A later knee protects more of the source 1:1.

It leaves less output range for the brightest values.

The shoulder

The shoulder compresses the high end toward the target maximum.

A strong shoulder can preserve the presence of many highlight levels while reducing their spacing.

A weak shoulder can keep highlights brighter but may reach the clipping limit sooner.

Neither is automatically more accurate.

The toe

A toe changes the low end.

It may be used to adapt to:

A lifted toe can make dark detail more visible.

It can also reduce the intended depth of black.

Conceptual plot of target output against source input. The curve shows a toe near black, a central one-to-one region, a knee where highlight compression begins, and a shoulder approaching the target maximum. The curve is not a required consumer implementation.
A nonlinear mapping can protect a central relationship and compress the regions that do not fit the target.

There is no lossless mapping into a smaller color volume

If the source contains colors and luminance outside the target color volume, every mapping makes a compromise.

Possible priorities include:

A target display cannot preserve every source coordinate when those coordinates do not fit.

Tone mapping should be judged against a stated priority and condition.

Four conceptual HDR mapping cards show different priorities for a smaller target: protect midtones, protect highlight detail, protect highlight intensity, or protect saturated color. A note states that no option preserves every source coordinate outside the target volume.
Mapping choices are compromises, so comparisons should state which source relationship each mode prioritizes.

Tone mapping is not enough by itself

Suppose a mapper reduces the luminance of a bright saturated color.

Depending on the processing space and method, the result may:

A complete mapping may need a saturation or gamut adjustment.

SMPTE ST 2094-40 defines dynamic metadata for a color-volume transform and includes tone and color-saturation mapping information. [5]

The correct broader phrase is often:

tone and color-volume mapping

Comparison diagram. The left panel shows a one-dimensional luminance curve that changes target light output. The right panel shows a three-dimensional target volume where luminance, saturation, and hue relationships are managed together.
A luminance curve can be necessary, but a complete target adaptation may also need saturation and hue management.

Global mapping

A global mapper applies one relationship across the image at a given time.

Advantages can include:

Limitations can include:

Local mapping

A local mapper can vary the transform by image region. [15]

It may attempt to preserve local contrast in:

Local mapping is not the same as local dimming.

Local tone mapping

An image-processing transform that varies spatially.

Local dimming

A display-light control that varies backlight or emissive output by area.

The two can interact.

They are not the same mechanism.

Local mapping tradeoffs

A local method may create:

It can also make difficult scenes easier to see on a limited display.

The result must be evaluated on the tested implementation.

Two conceptual screen drawings compare global and local tone mapping. The global panel applies one curve to the full frame. The local panel divides the frame into regions that can receive different mapping. A footer states that local tone mapping is image processing, not local dimming hardware.
Local mapping can protect selected regions, but it also introduces spatial and temporal tradeoffs that a global curve avoids.

Temporal mapping

A dynamic mapper can change with time.

If it reacts too quickly, the viewer may see:

If it reacts too slowly, the mapping may lag behind the scene.

Good temporal processing balances:

No one response time is correct for every source.

Conceptual timeline showing a dark scene, a bright cut, and a return to dark. One output response changes rapidly and overshoots, creating visible pumping. A second response changes more smoothly but lags the cut. The plot is not a product measurement.
Rapid adaptation can reveal pumping, while slower adaptation can lag cuts and flashes.

Metadata can guide the mapping

Metadata can describe the mastering display, content light levels, scene characteristics, or creative transforms.

It can help the target system choose a mapping suited to the source and display.

Metadata do not increase the physical peak or color volume of the television.

They do not replace the mapping algorithm.

Static metadata

The HDR10 media profile includes:

These values can help identify the source context.

They remain limited.

Static metadata do not define one curve

A television can use HDR10 static metadata while also performing:

The static nature of the metadata does not make the display processing static.

Flow diagram showing one fixed HDR10 metadata set entering a television together with changing image analysis, panel state, and picture mode. These inputs feed a display mapper that can change over time even though the delivered metadata remain static.
Static metadata describe the asset, but a television may still adapt its processing to the image, mode, and panel state.

Static metadata may be incomplete

A field can be:

A robust display needs a fallback.

That is one reason real-time image analysis exists.

Dynamic metadata

Dynamic metadata can vary with the content.

The SMPTE ST 2094 suite is intended for color-volume transforms when the target display has a smaller color volume than the mastering display.

The metadata can vary scene-by-scene or image-by-image. [6]

That gives the target more content-dependent information.

Flow diagram showing video essence and dynamic metadata entering a tone and color-volume mapper together with target-display capability. The mapper produces displayed light. A warning says metadata cannot increase the display peak or color volume.
Dynamic metadata can guide a compatible mapper, but the final result still depends on the target display and its implementation.

HDR10+

HDR10+ uses SMPTE ST 2094-40 Application #4 metadata.

The system can provide:

The display still has physical limits and an implementation.

The metadata do not guarantee one identical output across products.

Dolby Vision

Dolby Vision content mapping can use:

Dolby's workflow allows the creator to inspect and adjust mapped targets.

This provides more creative guidance than unreviewed automatic mapping alone.

It does not remove every consumer variable.

Dolby Vision is not one consumer mapper

Dolby Vision behavior can depend on:

Two Dolby Vision televisions do not become identical.

The system can reduce uncontrolled variation.

It does not eliminate it.

Dynamic metadata are not an automatic quality score

Dynamic metadata can help when:

HDR10 can perform well when:

Compare the actual result.

Do not rank it from the badge alone.

Why televisions differ

A television's displayed HDR result can depend on:

Two products with similar advertised peak luminance may still produce different images.

Price is not a measurement

A higher-priced television may include more capable hardware or processing.

That does not establish the behavior of a specific mode.

A lower-priced television may track the selected target well in some conditions.

Use model-specific measurements.

Do not substitute price category for evidence.

What a reviewer needs to record

A meaningful tone-mapping result should identify:

A single photograph or one advertised number is not enough.

Dynamic Tone Mapping controls

Many televisions offer a setting with a dynamic-sounding name.

Possible labels include:

These labels are not standardized.

What a dynamic mode may do

A mode can:

One brand's On is not another brand's On.

Accuracy versus visibility

A reference-oriented mode may attempt to preserve the applicable target until the display reaches its limit.

A brighter mode may intentionally alter the image so it is easier to see in a lit room.

Those are different goals.

A viewer can prefer the brighter presentation.

The preference does not prove closer reference reproduction.

Should Dynamic Tone Mapping be enabled?

There is no universal answer.

Use this decision:

Choose the reference-oriented mode when:

Consider a brighter or adaptive mode when:

Measure or compare the exact mode.

Do not assume the label describes the result.

“Off” may still map

A television called Tone Mapping Off may still:

The word Off may mean:

Only documentation and testing can determine the behavior.

Peak-brightness controls

A setting called:

can affect the output available to the mapper.

The control may also change:

The correct setting is model-specific.

Highest is not automatically reference

For some HDR modes, the documented reference configuration may use the highest panel-drive setting.

For others, the mode configures it automatically.

A manually changed value may cause the measured curve to deviate.

Use the exact manual and measurement workflow.

Do not copy a peak setting from another model or screen size.

Room-light adaptation

ITU-R BT.2408-9 notes that manufacturers may provide modified brightness and display characteristics for PQ monitoring in environments brighter than the reference surround. [2]

This is practical.

It is also a change from the reference viewing condition.

Room-light adaptation can:

The correct choice depends on whether the goal is:

HGiG and games

Games are different from fixed video.

A game engine can know:

This makes source-side tone mapping possible.

What HGiG recommends

The HDR Gaming Interest Group Version 1.1 document recommends a coordinated system. [7]

The game uses a parameterized tone mapper.

The display or system provides information related to the display's tone-mapping range.

The game renders critical objects within defined HDR ranges.

The display preserves gradation in those ranges in Game mode.

Primary and extended HDR ranges

HGiG defines:

The distinction reflects the fact that display capability can depend on image area.

It is more precise than using one peak number.

HGiG is not just hard clipping

A common explanation says HGiG turns tone mapping off and makes the television hard clip.

That can resemble one manufacturer's implementation.

It is not the complete recommendation.

The HGiG system still includes:

The goal is to avoid uncontrolled overlapping mappings while preserving gameplay-critical detail.

HGiG Version 1.1 best-practice flow. Display range information and system calibration provide MinTML, MaxFFTML, and MaxTML information to a game parameterized tone mapper. The game uses Primary and Extended HDR ranges, and a Game-mode display is intended to preserve gradation in those ranges. The exact product menu is not standardized.
HGiG is a coordinated source-and-display workflow, not a promise that every television simply disables all mapping.

Double mapping

A game may tone map into an assumed target.

The television may then apply another mapping.

Those transforms can interact.

Possible results include:

This is the problem coordinated calibration tries to reduce.

Double mapping is not guaranteed to look bad in every game.

It makes the result harder to predict.

A menu named HGiG is a product feature

The mode name does not prove:

Test the exact television and game.

Calibrating a game system

A defensible setup should:

  1. Put the television in the intended Game mode.
  2. Select the intended tone-mapping behavior before calibration.
  3. Disable unrelated adaptive controls for diagnosis.
  4. Run the system or game calibration exactly as documented.
  5. Verify that the game uses the calibration information.
  6. Check full-frame and small highlight behavior.
  7. Record the television mode and firmware.

Do not use universal click counts from another television.

Dolby Vision picture modes

Some televisions offer modes with names such as:

The names and behavior vary.

A dark-room label does not prove reference EOTF tracking.

A bright-room label does not prove the best adaptation.

Use the manufacturer's exact documentation and measurements.

How to compare two tone-mapping modes

Use both patterns and real images.

Pattern 1: PQ EOTF sweep

Shows:

Pattern 2: small highlight steps

Shows:

Pattern 3: large bright area

Shows:

Pattern 4: saturated color ramps

Shows:

Pattern 5: near-black steps

Shows:

Real scene sequence

Use scenes with:

Patterns explain the curve.

Scenes expose image and temporal behavior.

Six test tiles for evaluating HDR mapping: PQ EOTF sweep, highlight steps, full-frame bright area, saturated-color ramps, near-black steps, and a dark-to-bright cut. A footer says patterns reveal static relationships while real scenes reveal local and temporal behavior.
A defensible evaluation combines transfer-function patterns, area and color tests, near-black checks, and temporal scene changes.

Change one setting at a time

A valid comparison keeps constant:

Change only the mapping control.

Otherwise the result cannot be attributed to tone mapping alone.

What calibration can verify

Calibration and controlled measurement can evaluate:

What calibration cannot recover

Calibration cannot:

It can reduce measurable display error under stated conditions.

Tone mapping is not the whole television

The title calls tone mapping the secret main character.

That is an analogy.

Tone mapping can be a major reason HDR differs across displays.

The result also depends on:

The analogy stops when it suggests processing can overcome any physical limitation.

It cannot.

Where this leaves the Foundations arc

The earlier Foundations articles established:

Tone mapping connects those concepts to the target display.

For PQ, the source can contain absolute reference luminance and color outside the target display's capabilities.

Display mapping decides what relationships to preserve and what must change.

For HLG, the reference display relationship already adapts to peak, black, and viewing conditions.

Additional manufacturer processing can still alter the result.

There is no universal map that preserves everything.

There is a controlled set of compromises.

When the mapping works well:

When it works poorly:

The final question is not:

Did the television use tone mapping?

The useful questions are:

What source and target color volumes were involved?

Which relationships did the mapping preserve?

Which values were compressed, clipped, lifted, or shifted?

Did the measured result approach the intended target under the stated viewing conditions?

That is the bridge between display science and calibration in practice.

Sources and standards

Primary HDR display-mapping standards, operational guidance, gaming recommendations, and format-owner documentation used in this article. Last verified July 31, 2026.

  1. International Telecommunication Union Radiocommunication Sector. High Dynamic Range Television for Production and International Programme Exchange. Report ITU-R BT.2390-12, March 2025.
    https://www.itu.int/dms_pub/itu-r/opb/rep/R-REP-BT.2390-12-2025-PDF-E.pdf

    ↩

  2. International Telecommunication Union Radiocommunication Sector. Guidelines for operational practices in high dynamic range television production. Report ITU-R BT.2408-9, March 2026.
    https://www.itu.int/pub/R-REP-BT.2408-9-2026

    ↩↩↩↩

  3. International Telecommunication Union Radiocommunication Sector. Image Parameter Values for High Dynamic Range Television for Use in Production and International Programme Exchange. Recommendation ITU-R BT.2100-3, February 2025.
    https://www.itu.int/dms_pubrec/itu-r/rec/bt/R-REC-BT.2100-3-202502-I%21%21PDF-E.pdf

    ↩↩

  4. Society of Motion Picture and Television Engineers. High Dynamic Range Electro-Optical Transfer Function of Mastering Reference Displays. SMPTE ST 2084:2014, reaffirmed 2017.
    https://pub.smpte.org/pub/st2084/st2084-2014.pdf

    ↩

  5. Society of Motion Picture and Television Engineers. Dynamic Metadata for Color Volume Transform - Application #4. SMPTE ST 2094-40:2020, approved April 9, 2020.
    https://pub.smpte.org/pub/st2094-40/st2094-40-2020.pdf

    ↩↩

  6. Society of Motion Picture and Television Engineers. The ST 2094 Standards Suite for Dynamic Metadata. January 12, 2017.
    https://www.smpte.org/standards/webcasts/st2094_standards_dynamic_metadata_video

    ↩

  7. HDR Gaming Interest Group. For a Better HDR Gaming Experience: Best Practice Recommendations for Game HDR Creation. Version 1.1, July 11, 2019.
    https://www.hgig.org/doc/ForBetterHDRGaming.pdf

    ↩↩

  8. Dolby Laboratories. Dolby Vision for Content Creators.
    https://professional.dolby.com/content-creation/dolby-vision-for-content-creators/

    ↩

  9. Dolby Laboratories. Dolby Vision Metadata Levels. Updated April 1, 2025.
    https://professionalsupport.dolby.com/s/article/Dolby-Vision-Metadata-Levels

    ↩

  10. Dolby Laboratories. Dolby Vision Color Grading Best Practices Guide. Version 4, December 16, 2019.
    https://professional.dolby.com/siteassets/pdfs/dolby_vision_best-practices_colorgrading_v4.pdf

    ↩

  11. HDR10+ Technologies, LLC. Understanding the HDR10 Ecosystem. Updated April 2025.
    https://hdr10plus.org/wp-content/uploads/2025/04/HDR10_Ecosystem_April_2025.pdf

    ↩

  12. HDR10+ Technologies, LLC. HDR10+ System Whitepaper. September 4, 2019.
    https://hdr10plus.org/wp-content/uploads/2023/11/HDR10_WhitePaper.pdf

    ↩

  13. Consumer Technology Association Video Division. 4K Ultra High-Definition TV and High Dynamic Range Compatible Display Consumer Definitions.
    https://cdn.cta.tech/cta/media/media/membership/pdfs/video-technology-consumer-definitions.pdf

    ↩

  14. International Telecommunication Union Telecommunication Standardization Sector. Information Technology - Generic Coding of Moving Pictures and Associated Audio Information: Systems. Recommendation ITU-T H.222.0 v10, April 2025. The current edition is access restricted; the MaxCLL and MaxFALL field definitions were checked against the publicly accessible v9 edition from August 2023.
    https://www.itu.int/rec/T-REC-H.222.0

    ↩

  15. HDR10+ Technologies, LLC. HDR10+ Advanced White Paper. February 6, 2026.
    https://hdr10plus.org/wp-content/uploads/2026/02/HDR10_ADVANCED_WP_020526.pdf

    ↩
Related reading: HDR TV Setup: HDR10, HDR10+, HLG, and Games Move from tone-mapping theory to practical HDR setup and verification. Next step: Setting the Stage Begin the practical calibration arc by controlling reflections, surround light, viewing distance, wall color, and repeatable room conditions.