The Gold Standard of Visual Arts | Separating Out-of-Gamut Clipping from Ordinary Desaturation

A colour coordinate has meaning only inside its colour space. Record transfer function, primaries, white point and encoding where relevant, then name the destination. The same numerical triplet can denote different colours in sRGB, Display-P3 and other spaces.

Start with a claim ledger. Write the reported quantity, numerator, denominator, observational unit, time window and transformation rule in separate fields. Mark every value as observed, stipulated, estimated or assumed. This prevents a familiar label from carrying a different basis into the calculation. Keep the raw record beside each derived value so another reader can reproduce the step and locate any disagreement.

A fictional source label says Display-P3 but an export process silently interprets the numbers as sRGB. The visible change occurs before any legitimate gamut mapping, so calling it clipping would misdiagnose the fault.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to repeat the example with the correct source tag and compare the resulting destination coordinates. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Next chapter

A destination gamut is the set of colours representable under the specified model or device conditions. It is not a universal boundary around good colour. Device-independent coordinates can describe colours beyond a particular RGB cube, while a printer profile represents a different shaped set.

Draw the mechanism before interpreting the number. Show where material, information or energy enters, what transformation occurs and where a loss or bias can arise. Then ask which observation would distinguish two plausible explanations. A diagram is useful only when every arrow has a stated condition and an evidence route.

Plot three invented colours against one declared destination. Two lie inside and one lies outside. Only the third requires a mapping step, even if an observer personally finds all three intense.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to change the destination from sRGB to a wider space and reclassify the same source coordinates. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Simple clipping limits each destination RGB channel independently. It is easy to calculate but can discard how far a source lay beyond a boundary and can distort relationships among channels. Preserve the pre-clipped values.

Use two checks that fail differently. Reconstruct an input from outputs, calculate a limiting case, or solve the same example from the original definitions. Agreement matters most when the checks do not repeat the same assumption in different notation. Preserve the first result in the audit trail if a later check changes it.

For linear RGB (1.18, 0.42, -0.06), clipping to [0,1] gives (1.00, 0.42, 0.00). The red excess 0.18 and negative blue -0.06 vanish; the result cannot reveal either original magnitude.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to compare (1.02,0.42,-0.01) with the main source and show that both can map near the same boundary colour. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Desaturation normally describes a chosen movement toward a less colourful result. It can be applied to an in-gamut colour and need not touch a channel boundary. Gamut mapping may use chroma reduction, but purpose and trigger still matter.

Keep the comparison fair. Hold the population, interval, boundary, protocol and unit constant before treating a difference as evidence of change. If one condition differs, report the difference as a changed context and either standardise it transparently or narrow the claim.

An in-gamut orange is reduced from a stipulated chroma 0.18 to 0.12 while lightness and hue are held fixed. No channel exceeded the destination range, so the change is deliberate desaturation rather than boundary repair.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to apply the same chroma reduction to both an in-gamut and out-of-gamut source and record which operation was required. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Per-channel clipping can alter hue and lightness because RGB coordinates are not perceptual attributes. A result that looks less colourful may also have shifted in another dimension. Compare in a suitable perceptual representation and by viewing.

Separate a measurement from the decision built on it. A value can be calculated correctly while remaining too uncertain, too local or too method-dependent for the intended action. Report the result, the conditions that make it interpretable and the next observation needed for the real decision.

Two fictional conversions reduce a source chroma by similar amounts. Version A changes hue by one degree; version B changes it by twelve degrees. Calling both equally desaturated hides the more consequential hue change.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to hold chroma constant while changing lightness, then test whether viewers still describe the change as desaturation. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

A single patch can hide clipping. In a smooth gradient, several distinct source values may all map to the same endpoint, creating a flat region or contour. Inspect spatial structure rather than only one sample.

Treat missing and excluded observations as data. Record why they are absent, whether absence is related to the process being studied and which denominator they leave. A clean table produced by silently dropping difficult cases is less trustworthy than an incomplete table whose limits are explicit.

Five source reds rise from 0.92 to 1.20. Direct clipping maps the last four values above 1.00 to the same red endpoint, while a documented compression preserves ordered differences.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to reverse the gradient and verify that the output ordering remains monotonic. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

A fair comparison starts from identical source pixels and the same destination. Record whether the method clips channels, reduces chroma, changes lightness, compresses the whole gamut or uses a profile rendering intent.

Preserve sequence and identity. Two datasets with the same totals can imply different mechanisms when events occur in a different order or belong to different units. Keep timestamps, specimen or cohort identities and processing stages until the reader question no longer depends on them.

Method C clips the main orange at the boundary. Method D reduces its chroma until it is in gamut. The correct report describes both transformations and asks which better preserves the artwork’s relationships.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to add a second neighbouring colour and judge the pair rather than the focal patch alone. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Some mapping algorithms leave in-gamut colours unchanged; others compress a wider region. Verify the chosen method instead of assuming every conversion protects all safe colours.

Test a deliberately changed case. Change one assumption while holding the rest fixed, calculate again and explain why the conclusion moves or stays stable. Transfer is demonstrated when the reasoning survives a new context, not when the original numbers are merely substituted into the same line of arithmetic.

A neutral grey changes slightly in one export even though it was in gamut. That points to a broader transform, profile conversion or tone adjustment, not proof that grey itself was clipped.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to test a set of neutral and low-chroma control patches through the same pipeline. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

A file may encode a wide-gamut colour that a current display cannot show. A screenshot, browser path or operating system may also convert it. Separate file validity, software colour management and physical display gamut.

State the boundary of inference in the answer itself. Name the system, population, representation or interval to which the result applies, then name one tempting stronger claim that the evidence cannot support. This turns a caveat into part of the reasoning rather than an afterthought.

The same tagged file appears vivid on one managed display and flatter on another. Without measuring or profiling the displays, the difference cannot be assigned solely to the artwork.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to view the file through a controlled soft-proof and an unmanaged viewer, documenting the pipeline difference. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

An eyedropper reports coordinates in some sampling context. It does not automatically reveal the source profile, post-conversion space, display output or perceptual difference. Save the document settings and conversion history.

Use uncertainty as structure. Identify which input controls the conclusion, vary it over a defensible range and show whether the decision changes. A long list of generic caveats is weaker than one sensitivity calculation tied to the actual mechanism.

A sampled value reads 255,107,0 in an 8-bit export. That boundary red channel is consistent with clipping but does not prove it; the artist could have authored that exact in-gamut value.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to trace the same pixel back to a layered source and compare pre-export coordinates. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Bit-depth quantisation rounds representable values to discrete codes. It can produce banding even when all colours are in gamut. Gamut clipping concerns the boundary of representable colours, so the two failures need different evidence.

Finish with an evidence package. Include raw observations, definitions, exclusions, processing version, calculations, validation result and decision rule. A future reviewer should be able to reproduce the result and understand what would require the conclusion to be revised.

A shallow in-gamut gradient shows steps after conversion to a low-bit-depth format. No channel exceeds its permitted range; increasing precision helps, while changing gamut mapping alone may not.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to hold bit depth high and force an out-of-gamut endpoint to isolate the other mechanism. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Profile-based output can use relative colorimetric, perceptual or other intents with behaviour dependent on profiles and implementations. Record profiles and settings rather than reducing every result to clipping versus saturation.

Start with a claim ledger. Write the reported quantity, numerator, denominator, observational unit, time window and transformation rule in separate fields. Mark every value as observed, stipulated, estimated or assumed. This prevents a familiar label from carrying a different basis into the calculation. Keep the raw record beside each derived value so another reader can reproduce the step and locate any disagreement.

Two exports use the same source file but different destination profiles. Their boundary colours differ because the devices describe different gamuts. A visual disagreement is expected and auditable.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to keep the profile fixed and change only the rendering intent. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

A specified colour may fall outside a destination gamut. The task becomes choosing and documenting an acceptable reproduction, not claiming the impossible source coordinate was printed exactly.

Draw the mechanism before interpreting the number. Show where material, information or energy enters, what transformation occurs and where a loss or bias can arise. Then ask which observation would distinguish two plausible explanations. A diagram is useful only when every arrow has a stated condition and an evidence route.

A fictional brand cyan fits a wide-gamut display but not a proofed print condition. The designer compares nearest boundary, chroma-reduced and alternate-ink versions against an approved reference.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to change the paper profile while keeping the source specification fixed. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Photographs contain gradients, textures and relationships. Mapping one pixel independently can create seams or flatten highlight detail. Inspect representative regions and the whole image at intended viewing size.

Use two checks that fail differently. Reconstruct an input from outputs, calculate a limiting case, or solve the same example from the original definitions. Agreement matters most when the checks do not repeat the same assumption in different notation. Preserve the first result in the audit trail if a later check changes it.

A sunset export keeps the central orange but clips surrounding reds into one plateau. The patch match looks acceptable while the gradient loses depth.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to test a local crop and the full image under identical viewing conditions. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

An artist may desaturate to alter emphasis, mood or unity even when the source fits every intended output. Preserve that intention as an edit step distinct from later output conversion.

Keep the comparison fair. Hold the population, interval, boundary, protocol and unit constant before treating a difference as evidence of change. If one condition differs, report the difference as a changed context and either standardise it transparently or narrow the claim.

A painterly digital scene is deliberately muted by 20% before an sRGB export. A second small gamut adjustment occurs at export. Combining both into one label loses authorship and reproducibility.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to export the original and deliberately muted versions through the same destination transform. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

A soft proof simulates an output condition on another display. Its value depends on profiles, calibration, viewing conditions and whether the display can represent the simulated colours. It is evidence, not the physical print.

Separate a measurement from the decision built on it. A value can be calculated correctly while remaining too uncertain, too local or too method-dependent for the intended action. Report the result, the conditions that make it interpretable and the next observation needed for the real decision.

A proof flags two flowers as out of gamut and shows reduced separation. The later print confirms the direction but not every magnitude because paper, illumination and calibration also matter.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to repeat under a second approved viewing illuminant and describe what remains stable. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Colour changes can alter distinctions used for reading or interpretation. Gamut mapping does not guarantee sufficient contrast or preserve category separation. Test the actual communication goal after conversion.

Treat missing and excluded observations as data. Record why they are absent, whether absence is related to the process being studied and which denominator they leave. A clean table produced by silently dropping difficult cases is less trustworthy than an incomplete table whose limits are explicit.

Three map categories are distinct in the source but two become nearly identical after clipping. Adding labels or patterns preserves the task without claiming the original colours survived.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to convert a chart and an artwork with the same palette, then apply different decision criteria. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Digital gamut mapping does not predict a paint mixture. Pigment spectra, opacity, substrate and lighting create a different mechanism. Use the linked paint-mixture guide when material recipes are the reader’s question.

Preserve sequence and identity. Two datasets with the same totals can imply different mechanisms when events occur in a different order or belong to different units. Keep timestamps, specimen or cohort identities and processing stages until the reader question no longer depends on them.

A clipped digital orange and a physical orange swatch share a screen reading under one capture. That coincidence does not establish a mixing recipe or equality under another light.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to change illumination and compare the photographed and physical results again. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Statements such as “the colour was clipped because it looks dull” skip the source, destination and transform. Repair them by naming the gamut test, saved coordinates and competing explanations.

Test a deliberately changed case. Change one assumption while holding the rest fixed, calculate again and explain why the conclusion moves or stays stable. Transfer is demonstrated when the reasoning survives a new context, not when the original numbers are merely substituted into the same line of arithmetic.

A student compares two JPEGs and calls the flatter one clipped. Metadata shows it was intentionally desaturated before export; both files are fully within sRGB. The original claim is rejected.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to construct a genuinely clipped file that looks vivid and explain why appearance alone misleads. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Store the source file, embedded profiles, software and version, destination profile, rendering settings, bit depth, test patches, output date and viewing conditions. Preserve screenshots only as secondary evidence.

State the boundary of inference in the answer itself. Name the system, population, representation or interval to which the result applies, then name one tempting stronger claim that the evidence cannot support. This turns a caveat into part of the reasoning rather than an afterthought.

A reviewer recreates the conversion from a saved wide-gamut file and obtains the same boundary values. When a later software version differs, both outputs remain traceable.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to remove the profile tag from a copy and document the ambiguity that follows. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

The best mapping depends on the task: preserve a logo relation, retain a photographic gradient, match an approved proof, or keep chart categories distinct. Translate the artistic goal into observable checks.

Use uncertainty as structure. Identify which input controls the conclusion, vary it over a defensible range and show whether the decision changes. A long list of generic caveats is weaker than one sensitivity calculation tied to the actual mechanism.

For a fictional poster, the chosen method must keep three oranges ordered and maintain text contrast. A slightly lower maximum chroma is accepted because it protects those relationships.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to switch the task from poster legibility to expressive single-colour impact and reconsider the decision rule. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Next chapter

Use the W3C CSS Color Module Level 4 for current web-colour definitions and gamut-mapping concepts. Continue with grayscale contrast, paint mixtures and matching reflectance records for adjacent mechanisms.

Finish with an evidence package. Include raw observations, definitions, exclusions, processing version, calculations, validation result and decision rule. A future reviewer should be able to reproduce the result and understand what would require the conclusion to be revised.

The final transfer is to explain one changed colour with a complete source-to-destination chain, then identify a second plausible mechanism and the evidence that separates them.

Before moving on, identify the observation that would most directly challenge the current explanation. If no feasible observation exists, narrow the claim instead of hiding the ambiguity.

A changed-context transfer check is to repeat the audit for a print profile and state why its result cannot be inferred from the screen conversion alone. State what changes, what remains comparable and what the new result still cannot establish.

Contents · Previous chapter · Return to Visual Arts