What Nits Mean in HDR, and Why PQ Uses Absolute Luminance

Television reviews often lead with one number:

• 600 nits
• 1,000 nits
• 2,000 nits
• 4,000 nits

The number sounds simple. It is the measured luminance of the screen under stated test conditions.

What it means for the picture is more complicated.

First, the formal unit is candela per square metre, written cd/m². The display industry commonly calls one cd/m² a nit, although CIE terminology marks nit as a deprecated name. ^[1][2][3]

Second, one peak-luminance number does not describe the whole television. It does not tell you how long that output lasts, how large the bright area can be, how accurate the EOTF is, how bright saturated colors become, how the display handles highlights above its capability, or what happens in a bright room.

Third, not every HDR system uses luminance in the same way.

PQ uses an absolute display-luminance relationship. HLG uses a relative scene-light signal whose display result depends on the target display and viewing conditions. ^[7][8][9]

So the useful answer is:

Nits tell you how much luminance a display or image region produces under defined conditions. They are essential to understanding PQ-based HDR, but they are only one part of HDR performance.

What the unit measures

Luminance is a photometric quantity.

It describes light traveling in a specified direction from an emitting or reflecting surface, weighted using a standardized model of human spectral sensitivity. Its SI unit is the candela per square metre. ^[2][3]

A television measurement of 1,000 cd/m² means that the measured screen area has a luminance of 1,000 candelas per square metre in the measurement direction.

The unit does not directly tell you:

• total light leaving the television
• electrical power consumption
• how bright the image feels to every viewer
• how much light reaches the eye after room reflections
• whether the color is accurate
• whether the luminance is sustained
• whether the whole screen can reach that value

Those are different questions.

Why the candela is not a candle measurement anymore

The candela is one of the seven SI base units.

Its modern definition does not rely on comparing a source with an ordinary candle flame. It fixes the luminous efficacy of monochromatic radiation at a frequency of 540 × 10¹² hertz to 683 lumens per watt. ^[3][4]

This provides a physical basis for photometric measurement while retaining weighting related to standardized human visual sensitivity.

The candela is a unit of luminous intensity in a direction.

The candela per square metre is a derived unit of luminance.

For television readers, the formal chain is:

• candela: luminous intensity
• square metre: area
• candela per square metre: luminance of a surface in a direction

Is a nit the same as cd/m²?

Numerically, yes.

One nit equals one candela per square metre.

Terminologically, the SI and CIE preference is cd/m². CIE S 017:2020 lists nit as deprecated and says that the symbol nt is no longer in use. ^[1]

Television marketing, reviews, and everyday discussion still use the word because it is short and familiar.

This article uses cd/m² when discussing standards and measurements. It uses nit when explaining the language readers see in reviews.

Luminance is not the same as perceived brightness

People often use brightness as an everyday substitute for luminance.

That is understandable, but the terms are not identical.

Luminance is a measurable photometric quantity.

Brightness is a visual perception influenced by:

• the luminance of the image
• surrounding light
• adaptation
• image size
• contrast
• color
• viewing distance
• duration
• the rest of the scene

A small 1,000 cd/m² highlight in a dark image can look intense. A large field at the same measured luminance may feel uncomfortable or may trigger display limiting. A 100 cd/m² white patch can look bright in a dark room and subdued near a sunlit window.

The measurement remains useful. The viewer's experience depends on context.

SDR was relative, not unmeasured

It is common to say:

SDR did not use nits.

That needs correction.

SDR signals are relative because the normalized signal values do not assign one fixed cd/m² output to every code value by themselves.

However, SDR production and evaluation still use luminance measurements.

Recommendation ITU-R BT.2035-0 specifies that HDTV reference white should correspond to 100 cd/m² and reference black should be below 0.01 cd/m² in its defined reference environment. It also points to BT.1886 for the display EOTF. ^[5]

Recommendation ITU-R BT.1886-0 uses a normalized input from reference black to reference white and gives 100 cd/m² as its reference white setting. ^[6]

The accurate distinction is:

• SDR signal values: relative between defined black and white points
• SDR reference display setup: measured in cd/m² under stated conditions

What SDR reference white means

Reference white is not the same as the maximum physical output a television can produce.

In the BT.2035 reference environment, reference white is set to 100 cd/m².

A consumer television can often produce more than that in SDR. It may be adjusted brighter for a living room. That does not change the relative meaning of the SDR signal, but it changes the display scale and the viewing result.

Two televisions can show the same SDR program at different white luminances.

Both may preserve the relative tone relationships reasonably well. They are not producing identical light at the viewer.

This is one reason SDR can be adapted to different rooms.

It is also why a display set far above the production reference can look brighter without necessarily being closer to the reference presentation.

SDR white was never simply "maximum beam"

The original article described SDR white as whatever the display could produce at maximum output.

That is too broad.

Professional reference displays are adjusted to defined values. The signal and display have headroom, black-level behavior, EOTF requirements, and reference conditions.

A consumer mode may use more of the panel's capability, but that is an implementation choice.

The phrase relative system does not mean no target.

It means the signal-to-light relationship is normalized to reference black and reference white rather than assigning one absolute cd/m² value to every signal level.

What PQ changed

SMPTE ST 2084 defines the electro-optical transfer function commonly called PQ, or Perceptual Quantizer.

PQ maps nonlinear signal values to an absolute display-luminance scale.

The standard's ideal range extends from nominal black to 10,000 cd/m². The EOTF is based on human visual contrast sensitivity so that a limited number of digital code values can represent a very wide luminance range efficiently. ^[7]

ITU-R BT.2100-3 incorporates PQ as one of its two HDR television systems. Its reference PQ EOTF maps the nonlinear PQ signal into display light in cd/m². ^[8]

This is the real move to absolute luminance.

A particular PQ value has a defined target display luminance on the reference EOTF.

That does not mean every television can produce the target directly.

Absolute does not mean the original scene is reproduced literally

A PQ signal can be display-referred, but the production process still includes rendering decisions.

The scene may contain sunlight, reflections, lamps, and shadows far outside the production display's range.

The camera exposure, reference OOTF, grading, visual effects, and artistic adjustments map the captured scene into the available HDR presentation.

ITU-R BT.2390-12 explains that the production process imposes rendering intent rather than simply sending linear scene light directly to the viewer. ^[9]

PQ provides an absolute display-luminance relationship after those choices.

It does not remove creative grading.

Why the PQ range ends at 10,000 cd/m²

SMPTE ST 2084 defines an idealized absolute range up to 10,000 cd/m². ^[7]

That number is the upper limit of the encoding system.

It is not:

• a requirement that every display reach 10,000 cd/m²
• a statement that every program contains 10,000 cd/m² pixels
• a recommended full-screen white level
• a normal average picture luminance
• a promise that 10,000 cd/m² is available at every chromaticity

ST 2084 specifically notes that the EOTF is intended to enable increased luminance range, not to make every image brighter overall. It also describes the ideal EOTF as a nominal target because real displays can differ due to output limitations and non-reference viewing environments. ^[7]

A 10,000 cd/m² code space leaves room for the future

A large encoding range allows:

• current programs to use only the luminance they need
• future displays to reproduce more of the encoded range
• highlight information to remain distinct above today's common display limits
• conversion and tone-mapping systems to work from a defined scale

Unused range is not wasted in the same way it would be under a linear code distribution. PQ allocates code values according to a perceptually designed nonlinear function.

The presence of the range does not prove that content uses it.

HDR reference white is not peak white

A common misunderstanding is that HDR white must be the brightest level in the signal.

ITU-R BT.2390-12 states that reference white is neither the maximum white level of the signal nor the maximum white level of the display. Luminance above reference white is available for highlights. ^[9]

ITU-R BT.2100-3 defines HDR Reference White as the nominal signal produced by a 100 percent reflectance white card, resulting in a nominal 203 cd/m² on:

• a PQ reference display
• an HLG display with nominal peak luminance of 1,000 cd/m² ^[8]

This creates room above diffuse white for:

• specular reflections
• emissive light sources
• glints
• bright windows
• fire
• lamps
• visual effects

HDR is not supposed to turn every white surface into a maximum-intensity highlight.

Peak display luminance is not full-screen white

BT.2100-3 specifies a reference HDR display peak luminance of at least 1,000 cd/m² for critical viewing, with a note that this does not imply that the level must be achieved for full-screen white. It is intended for small-area highlights. ^[8]

This distinction is essential when reading television measurements.

A display may produce:

• high luminance in a small window
• lower luminance in a larger window
• still lower luminance full-screen
• a brief peak that declines over time
• different luminance depending on scene history

A review should state the conditions.

Why HLG is different

HLG is the second HDR television system defined in BT.2100-3.

Its signal begins as a relative scene-light representation. The display EOTF uses:

• nominal peak luminance
• black luminance
• system gamma
• viewing conditions ^[8]

BT.2100 adjusts the HLG system gamma for displays with different nominal peak luminance. It also permits adjustment for brighter background and surround conditions.

This is fundamentally different from PQ's fixed absolute EOTF.

The corrected summary is:

• PQ: absolute display-luminance coding
• HLG: relative scene-light coding with display and environment adaptation

Both can produce HDR pictures.

Only one uses the absolute architecture described by ST 2084.

Why the distinction matters at home

A PQ television generally needs to decide how to handle encoded luminance above its capability.

An HLG television uses the display's nominal peak luminance and system relationship to render the relative signal.

The two systems can therefore respond differently to:

• a lower-peak display
• a higher-peak display
• a brighter room
• a changed contrast control
• production conversion
• transmission workflows

Calling all HDR "absolute brightness" hides this design difference.

What a mastering-display peak means

A mastering display has:

• a maximum usable luminance
• a minimum luminance
• a color gamut
• a color volume
• uniformity and stability limits
• pattern-size behavior
• signal-handling limits

The production team grades within the chosen environment and display capability.

A mastering display rated or operated at 1,000 cd/m² can show highlights up to that working limit under its specified conditions.

That does not mean the final program necessarily contains a pixel at exactly 1,000 cd/m².

The content may peak below the display limit.

The display limit may also differ from the content statistics or metadata.

Four luminance ideas that should not be merged

1. Mastering-display peak luminance

The maximum or nominal peak capability used in the mastering environment.

2. Content peak luminance

The highest luminance represented by the finished image data under the applicable interpretation.

3. HDR reference white

A nominal diffuse-white level, commonly 203 cd/m² in the BT.2100 reference cases.

4. Target-display peak luminance

The peak the viewer's display can produce under stated conditions.

These values can be different.

A television's mapping system may use some combination of them, plus metadata or assumptions, to create the final picture.

Why "mastered at 1,000 nits" can be misleading

The phrase can mean several things:

• the mastering display had a 1,000 cd/m² peak capability
• the workflow used 1,000 cd/m² as a nominal mastering maximum
• metadata reports a 1,000 cd/m² mastering display
• the actual content reaches close to 1,000 cd/m²
• a reviewer is using shorthand without distinguishing the values

Those statements are not interchangeable.

Publication wording should identify which one is known.

When only the mastering-display capability is known, the safe wording is:

The program was mastered using a display with a stated or signaled peak capability of 1,000 cd/m².

Do not silently turn that into:

The brightest pixel is 1,000 cd/m².

What happens when the display cannot follow PQ directly

A real display may not reproduce the full PQ EOTF because of:

• limited peak luminance
• limited black level
• limited color volume
• power or thermal limits
• room light
• panel behavior
• viewing comfort
• manufacturer processing

The display can apply an electro-optical transfer function adjustment or another display-mapping process.

ITU-R operational guidance describes mapping a PQ source or mastering range into the target display's black and white luminance. When the mastering limits are unknown, the complete 0 to 10,000 cd/m² PQ range can be used as an assumption. ^[10]

The goal is not always to preserve every luminance value one-to-one.

That may be physically impossible.

The goal is to reproduce the program within the target's capabilities while managing highlight separation, overall brightness, contrast, color, and creative relationships.

Tone mapping is not automatically an error

Tone mapping compresses or redistributes luminance when the source range exceeds the target range. ^[11]

Possible approaches include:

• hard clipping
• gradual roll-off
• scene-dependent mapping
• frame-dependent mapping
• metadata-guided mapping
• hue-preserving color-volume mapping
• luminance-priority mapping
• preserving midtones while compressing highlights
• adjusting the whole image to protect highlight detail

Every approach has tradeoffs.

Hard clipping can preserve most of the image unchanged but merge the brightest values.

A broad roll-off can preserve highlight detail but darken or flatten more of the image.

Dynamic systems can adapt to content but may alter the grade differently from a fixed reference display.

Tone mapping should be evaluated by:

• the source
• the target display
• the selected mode
• metadata or signal assumptions
• highlight detail
• midtone accuracy
• color preservation
• scene consistency
• viewing environment

Peak luminance does not equal tone-mapping quality

A higher peak can reduce the amount of compression needed.

It does not guarantee better mapping.

A lower-peak display with careful processing may preserve highlight relationships more gracefully than a brighter display with poor clipping or unstable scene-by-scene behavior.

Peak capability and mapping quality are related but separate.

What a television peak measurement needs to tell you

A useful peak-luminance measurement should identify: ^[12]

• exact model
• screen size
• region
• firmware
• signal format
• picture mode
• white point or color-temperature preset
• tone-mapping setting
• test-pattern size
• test-pattern type
• measurement duration
• sustained versus transient result
• meter
• meter profiling
• stabilization method
• date

Without those conditions, the number is difficult to compare.

Why a 10 percent window is used

A window pattern places white or another measured color in part of the screen while the rest remains darker.

A 10 percent window covers 10 percent of the picture area.

Small windows are useful because HDR highlights are often local rather than full-screen. They also reveal the display's ability to concentrate luminance in a limited area.

The result should not be interpreted as full-screen capability.

A display that reaches 2,000 cd/m² in a small window may produce much less with a large bright field.

Why duration matters

Some displays can produce a high output briefly and then reduce it because of:

• thermal management
• power management
• panel protection
• automatic brightness limiting
• static-image detection
• local-dimming behavior

A reviewer may report:

• initial peak
• peak after stabilization
• sustained output
• full-screen output
• real-scene highlight measurement

Those values answer different questions.

White luminance and color luminance are different

Peak-white measurement normally uses an achromatic signal.

Saturated colors do not necessarily reach the same luminance.

A display's red, green, and blue channels have different luminous contributions and physical limits. Some display architectures add a white component, use color filters, or change saturation as output rises.

A television can therefore have:

• very high white peak
• lower saturated-color luminance
• reduced saturation near peak white
• a color volume smaller than the white number suggests

This is why color volume matters.

From gamut to color volume

A two-dimensional chromaticity diagram plots color coordinates without luminance.

A dim red and a bright red can occupy the same chromaticity point.

Color volume adds luminance to the description.

It asks which combinations of:

• hue
• saturation or chroma
• luminance

the display can reproduce.

ITU-R BT.2390 identifies color volume as one of HDR's important image dimensions and explains that HDR and wide color gamut increase the range that must be represented and mapped. ^[9]

Why color volume has an irregular shape

A display's color volume is not a simple box.

At low luminance, black level and near-black behavior limit the available colors.

At higher luminance, primary and panel limits restrict saturated colors.

Near peak white, some displays move colors toward white or reduce saturation.

The shape depends on:

• display primaries
• white luminance
• primary luminance
• black level
• panel architecture
• processing
• tone mapping
• color-volume mapping
• measurement method

Percent color-volume figures need a method

A claim such as 80 percent BT.2020 color volume is incomplete unless it identifies:

• the reference volume
• the color-difference space
• the luminance range
• the sampling method
• whether out-of-gamut overshoot is counted
• the display mode
• the tone-mapping state
• the pattern type
• the measurement equipment

Two reviews can publish different percentages without either being dishonest if they use different methods.

Do not directly compare the percentages until the methods are compatible.

HDR is not simply brighter SDR

HDR can improve several things at once:

• highlight separation
• shadow detail
• local contrast
• diffuse-white and highlight relationships
• color volume
• indoor and outdoor detail within one scene
• realism of specular and emissive objects

SMPTE ST 2084 states that PQ is intended to increase luminance range, not simply raise the overall luminance of every image. ^[7]

A well-made HDR image can keep much of the scene at comfortable luminance while reserving the extended range for selected parts of the picture.

How to read a television review

When a review says that a television reaches 1,500 nits, ask:

1. Was that SDR or HDR?
2. Which HDR system?
3. Which picture mode?
4. Which tone-mapping setting?
5. What window size?
6. Was it a brief peak or sustained?
7. What was full-screen luminance?
8. Did white balance remain accurate?
9. What happened to saturated colors?
10. How did the display track the PQ EOTF?
11. Did highlights clip or roll off?
12. Was local dimming active?
13. Were energy-saving controls disabled?
14. Was the result measured on one sample?

A review that answers those questions is more useful than a larger number without context.

What the Peak Brightness control may do

Some televisions include controls named:

• Peak Brightness
• HDR Tone Mapping
• Dynamic Tone Mapping
• Contrast Enhancer
• HDR Optimizer
• Active Tone Mapping
• Static Tone Mapping
• Brightness Preferred
• Gradation Preferred

The names are not standardized.

A control may change:

• panel drive
• EOTF tracking
• highlight roll-off
• metadata use
• average picture luminance
• local dimming
• color volume
• energy behavior

Do not assume that a setting called Peak Brightness merely raises the display's hardware maximum.

Use the exact model manual and measurements.

What calibration can and cannot do

Calibration can reduce measurable error in:

• PQ EOTF tracking
• HLG system setup
• reference white
• grayscale
• color management
• clipping thresholds
• mode consistency
• luminance targets

Calibration cannot:

• make a 700 cd/m² panel produce 2,000 cd/m²
• create color volume outside the panel's capability
• recover clipped source information
• eliminate all automatic brightness behavior
• make a consumer television identical to a mastering display
• guarantee one tone-mapping result for every scene
• turn incorrect metadata into known production intent

The accurate claim is:

Calibration can bring the display closer to the applicable target within the display's capabilities and the stated measurement conditions.

Where this leaves us

The television industry talks about nits because luminance matters.

The formal unit is candela per square metre.

SDR signal values are relative, but SDR reference displays still use measured luminance, commonly 100 cd/m² reference white in the BT.2035 environment.

PQ changed the architecture by assigning absolute display-luminance targets across a 0 to 10,000 cd/m² encoding range.

HLG is different. It uses a relative scene-light signal and adapts the displayed result to the display and environment.

HDR reference white is not peak white.

A mastering-display peak is not automatically the content peak.

A television's small-window peak is not its full-screen or sustained output.

Tone mapping is the necessary process of adapting one luminance or color volume to another when direct reproduction is not possible.

Peak luminance matters because it can preserve more highlight intensity and separation.

It does not tell the whole story.

The better question is not only:

How many nits can this television reach?

It is:

Under what conditions, for how long, with what color accuracy, over what area, and how well does the television map the HDR program into its actual luminance and color volume?