How Photography Works | Tone Curves — How Brightness Relationships Are Rewritten After Capture

Two pixels can differ by the same amount in the sensor data and end up looking very different on the screen.

One small difference in the shadows may be expanded until texture appears. The same numerical difference in the highlights may be compressed until both values look almost identical. A photograph is not only captured. It is mapped.

A tone curve is a rule for turning captured brightness values into rendered brightness values.

This article continues the canonical How Photography Works | Every Photograph Leaves Something Out knowledge map. Editing Is the Second Exposure owns the broad post-capture stage. Tone curves isolate one of its deepest mechanisms: how the relationship among brightness values is deliberately rewritten for a display, print or aesthetic purpose.

Quick Read

A camera sensor can record a wide range of scene luminance, often in a relatively linear or near-linear measurement domain before rendering. Displays and prints cannot reproduce that range directly, and human vision does not respond linearly to physical light. Tone curves therefore map input values to output values. Steepen part of the curve and local contrast increases there; flatten it and differences are compressed. Lift the black point and deep shadows become grey. Lower highlights and bright regions retain more visible structure. Tone mapping can make a photograph look dramatic, soft, flat, luminous or compressed without changing the original scene that was photographed.

captured luminance → tone mapping → display value → perceived contrast

The Sensor and the Display Speak Different Brightness Languages

A sensor measures incoming light within its operating limits. A display emits light within a much smaller practical range, while paper reflects ambient light and has a different black and white boundary again.

The recorded scene therefore has to be translated. A tone curve is one part of that translation.

Linear Capture Does Not Look “Natural” by Itself

Raw sensor data viewed with a simple linear mapping often looks dark and flat to human observers because our perception and common display systems do not interpret physical luminance linearly. Rendering pipelines apply nonlinear transforms so middle tones become visually useful.

This is one reason a RAW file does not contain one finished appearance. It contains measurements that require interpretation.

The Curve Has an Input Axis and an Output Axis

Imagine input brightness running from black to white along the horizontal axis and output brightness running from dark to bright vertically. A straight diagonal means each input value maps proportionally to its corresponding output value. Bend the line and the relationship changes.

Move one point upward and that tonal region becomes brighter. Move it downward and it becomes darker. Change the slope and local contrast changes.

Slope Is Local Contrast

Where the curve is steep, a small difference in input brightness produces a larger difference in output brightness. Texture and separation become stronger. Where the curve is flat, many input values are squeezed into a narrow output range and appear more similar.

This is the mathematical heart of tone-curve editing. Contrast is not one global switch; it can be redistributed across the tonal range.

The S-Curve Redistributes Contrast

A classic S-shaped curve darkens some shadows, brightens some highlights and steepens the midtone region. The image often appears punchier because midtone separation increases while the ends of the range compress toward black and white.

But there is no free contrast. Expanding one region inside a limited output range usually compresses another. Tone mapping is allocation.

Lifting the Black Point Changes the Meaning of Black

If the darkest recorded values are mapped to a grey output instead of pure black, shadows become softer and the image acquires a faded or matte appearance. Detail may remain visible, but the photograph no longer uses the display’s deepest black.

This is not the same as recovering shadow detail from underexposure. It is a rendering choice about where the output black boundary sits.

Lowering Highlights Can Preserve Separation

Bright recorded values can be compressed into a lower output range so cloud texture, skin highlight transitions or reflective detail remain visible. This can make a high-dynamic-range capture fit a conventional display more gracefully.

But if the sensor data already clipped, no curve can restore values that were never distinguished. Highlights End Abruptly owns that acquisition boundary.

Shadow Recovery and Shadow Noise Are Linked

Lift weak shadow values and both real signal and uncertainty become more visible. The operation can reveal texture, but it can also expose colour speckles, banding and denoising artefacts.

This is why Article 29, Noise, sits immediately before tone curves. Rendering cannot create certainty that capture did not provide.

Global Curves Change the Whole Image

A global tone curve applies one mapping rule across the frame. A pixel with a given brightness receives the same tonal transform wherever it appears, subject to colour-space and channel interactions.

This preserves a simple relationship but can be too blunt when one part of the photograph needs different treatment from another.

Local Contrast Uses Spatial Context

Modern editing tools can increase contrast around edges or within local neighbourhoods rather than applying one global mapping. Clarity, dehaze and local tone mapping often work partly in this spatial domain.

The image can gain apparent depth and texture, but excessive local contrast can create halos, crunchy surfaces and unnatural transitions.

Tone Curves Can Be Applied Per Colour Channel

Instead of remapping only overall brightness, an editor can alter red, green or blue channel relationships separately. Raising red in shadows, for example, changes colour as well as luminance relationships.

This shows why tone and colour are not fully separable in RGB editing. White Balance adjusts broad channel relationships for illumination; channel curves can reshape those relationships across different brightness regions.

Gamma Is a Tone Relationship, Not an Exposure Setting

Gamma-like encoding reshapes the relationship between stored numerical values and displayed luminance. It helps allocate code values in ways better matched to perception and display behaviour.

Changing gamma-like tone response is not the same as changing how many photons were captured. One belongs to representation; the other belongs to exposure.

Film Had Tone Curves Too

Photographic film and paper have characteristic response curves that map exposure into density and print brightness. The familiar shoulder and toe of film response compress highlights and shadows differently from many digital workflows.

Digital tone curves therefore did not invent nonlinear photographic rendering. They make the mapping more explicit and adjustable.

Camera Picture Styles Are Pre-Built Rendering Decisions

Standard, vivid, neutral, portrait and film-simulation modes often differ partly through tone curves, colour transforms and sharpening. Two JPEGs from the same sensor exposure can therefore look substantially different without any change in the world or lens.

The appearance belongs to the camera pipeline, not solely to the scene.

HDR Requires Tone Mapping to Become Viewable

A high-dynamic-range scene or multi-exposure composite may contain a luminance span that cannot fit directly into a conventional output. Tone mapping compresses the range while trying to preserve useful local differences.

Handled gently, the result can resemble ordinary vision with retained highlight and shadow detail. Handled aggressively, it can make every region equally emphatic and destroy the hierarchy created by natural light.

Tone Curves Change Visual Hierarchy

Brighten a face while compressing the background and attention moves toward the person. Deepen a sky while lifting foreground texture and the landscape becomes more balanced. Crush the shadows and objects merge into silhouette.

Tone mapping therefore changes not only technical contrast but what the viewer notices first.

The Same Curve Can Behave Differently on Different Images

An S-curve applied to a low-contrast fog scene may create welcome separation. The same curve on harsh midday light may clip or exaggerate already strong contrast. A preset is a fixed transformation meeting a variable input.

This is why editing recipes cannot be separated from the histogram and scene distribution they act upon.

Tone Mapping Can Hide Context

Darkening a background until details vanish can make a subject cleaner. Raising every shadow can reveal previously invisible context. Neither operation changes what was physically in the scene, but both change what survives into the visible representation.

This connects tone curves back to Context Collapse. Information can be removed by tonal rendering as well as by cropping.

Scientific Images Need Declared Tone Mapping

When brightness represents measured intensity, nonlinear remapping can alter apparent differences among regions. A scientifically valid display may still use contrast enhancement, but the mapping should be documented when quantitative interpretation matters.

Scientific Photography Is Measurement With a Camera therefore separates measurement data from the visualization used to make patterns visible.

A Histogram Is Not the Tone Curve

A histogram describes how many pixels occupy different tonal values in a particular representation. A tone curve is the mapping that can move those values into different output regions.

Apply a curve and the histogram changes because the distribution has been remapped. One describes; the other transforms.

The Tone-Curve Audit

1. Input domain: RAW-linear, camera-rendered, log-like or already display-referred?
2. Black point: where do the darkest values map?
3. White point: where do the brightest values map?
4. Slope: which tonal regions gain or lose local contrast?
5. Midtones: are faces and main subjects being brightened or darkened?
6. Highlights: is bright detail compressed or clipped?
7. Shadows: is weak signal being lifted into visible noise?
8. Local processing: are halos or unnatural microcontrast appearing?
9. Colour: are channel-specific curves changing hue and saturation?
10. Purpose: expressive rendering, display compression or quantitative visualization?

Photography Laboratory 1: Build an S-Curve Slowly

Use one well-exposed image and apply a gentle S-curve, then a stronger one. Compare midtone separation, shadow compression and highlight compression. Notice where the image stops looking more dimensional and starts losing subtlety.

Photography Laboratory 2: Lift the Black Point

Raise the darkest output value until black becomes grey. Compare perceived depth, mood and the visibility of shadow structure.

Photography Laboratory 3: Same Exposure, Three Renderings

From one RAW file, make a high-contrast rendering, a flat rendering and a gentle naturalistic rendering. Ask which scene details each version makes easiest to notice.

For Primary Readers

Take one photograph and make a bright version, dark version and high-contrast version. Ask: Did the scene change, or did the way we showed the scene change?

For Secondary Readers

Plot simple input and output values on graph paper. Show how a steeper section of the curve increases differences and a flatter section compresses them.

For Advanced Readers

Analyse tone mapping as a nonlinear transfer function between scene-referred or sensor-referred values and output-referred luminance. Examine derivative slope as local contrast gain, interaction with quantisation and gamut, and the distinction between global operators and spatially varying local tone mapping.

Common Misconceptions

• “Tone curves change exposure.” They remap recorded values; exposure determines capture before rendering.
• “More contrast means more detail.” Contrast can make existing differences easier to see while compressing or hiding others.
• “Lifting shadows recovers certainty.” It may reveal recorded signal, but it also reveals noise and processing limits.
• “A preset curve works the same on every photograph.” Its effect depends on the input tonal distribution.
• “Scientific images should never use tone mapping.” Visualization may require it, but the mapping must not be mistaken for raw measurement.

Frequently Asked Questions

What does an S-curve do?

It typically increases contrast through part of the midrange while compressing some shadow and highlight values.

Can a tone curve recover clipped highlights?

No. It can remap bright values that were captured, but values already clipped to the sensor maximum cannot be separated again by a curve alone.

Why do RAW files look different in different software?

Different software applies different demosaicing, colour transforms, tone curves and default rendering choices to the same underlying measurements.

Final Thought: Brightness Is Rewritten Into a Displayable Story

The camera records a range. The final photograph decides how that range will be distributed so a viewer can see it on a finite medium.

A tone curve does not change what light reached the sensor. It changes which brightness relationships the finished photograph asks us to notice.