Each guide begins with a short answer and follows one original record through definitions, calculations, evidence checks, changed contexts, misconceptions and bounded reporting.
The worked numbers are stipulated for learning. Authoritative sources define the operational concepts; the examples show how to keep their boundaries visible.
The worked records are original teaching examples. Each keeps the source state, transformation, output state and decision on separate lines so the reader can see exactly where interpretation begins.
A changed-context check alters one consequential condition while leaving the other inputs fixed. This makes transfer visible and prevents one numerical answer from becoming a rule outside its scope.
Keep a compact evidence receipt: object, method, units, reference condition, transformation, result and nearest rival explanation. A future reader should be able to reproduce the narrow claim.
The five reading routes move from definitions to mechanism, evidence, transfer and reporting. Readers may enter by task, but the examples remain connected rather than becoming detached tips.
The practical standard is an answer that can be checked. Preserve raw inputs, intermediate calculations, exclusions and versioned sources, then name one stronger claim the evidence cannot support.
Before choosing a formula, name the reader job in one sentence. A monitoring alert, research comparison, artistic rendering and teaching exercise may use the same observation differently.
Every central case includes a near-match with one familiar surface feature and a different mechanism underneath. A definition becomes useful when a learner can reject the tempting substitute.
The evidence route treats missing information honestly. An unknown method, interval, calibration or boundary stays blank until a source supplies it; an assumption may complete arithmetic while weakening the conclusion.
The transfer route changes scale, timing, composition or sampling without silently changing every other field. Predict the direction first, then calculate or inspect the result.
The reporting route keeps measurement and judgement separate. State what was observed, what rule produced the derived value, what uncertainty remains and which action follows only under declared conditions.
The final reader check is practical: the short answer states the distinction, the worked example can be reproduced, and the sources show where the example ends.
A useful misconception test constructs two cases with the same headline number and different underlying records. If their interpretations differ, the omitted condition belongs in the result.
Where a standard leaves behaviour to a competent analyst, observatory, laboratory or software workflow, the article says so directly. That boundary tells the reader which conclusion can travel.
The examples avoid hidden precision. Rounded display values remain linked to the unrounded calculation, and stipulated inputs are labelled as teaching data.
A complete explanation keeps three layers apart: the thing in the world, the method that turns it into a value, and the decision made from that value. Numerical agreement does not merge those layers.
For each central example, sketch the boundary before calculating. Mark what enters, what is transformed or excluded, what the instrument or algorithm can observe, and what must be inferred.
This specialist guide belongs to the Visual Arts root. It also connects to The Gold Standard of Education | Learning That Transfers into Life and the How X Works hub. The worked cases are original teaching examples; their numbers are stipulated so that the reasoning can be checked.
eduKateSG · Visual Arts
Find your next learning step
Read the short answer first: scene-referred values preserve relationships to the captured or synthetic scene before display rendering; display-referred values represent a reproduction for a named output condition. Trace the output transform, target and operation order before comparing numbers.
ROUTE 1 · CHAPTERS 1–4
Define the quantities
Name the quantities, reference conditions and reporting rules.
ROUTE 3 · CHAPTERS 9–12
Audit the evidence
Check instruments, algorithms, assumptions and rival explanations.
ROUTE 5 · CHAPTERS 17–20
Report and continue
Repair shortcuts, answer questions and follow verified routes.
Chapter contents
Define the quantities · Chapters 1–4
1. Name scene-referred and display-referred states
2. Keep relative exposure separate from emitted luminance
Work the mechanism · Chapters 5–8
5. Work the central highlight case
6. Keep scene-linear and log working encodings distinct
Audit the evidence · Chapters 9–12
9. Audit gamut compression before output
10. Audit target peak and black level
Transfer the reasoning · Chapters 13–16
13. Transfer to SDR and HDR deliverables
Report and continue · Chapters 17–20
17. Repair brighter code means brighter scene
18. Repair tone mapping is just clipping
CHAPTER 1 OF 20 · Define the quantities
1. Name scene-referred and display-referred states
Scene-referred image values preserve relationships to relative scene exposure before a display rendering. Display-referred values represent a rendered reproduction encoded for a specified display and viewing condition. One is not a brighter or more accurate copy of the other.
Write the evidence in the order a reader can audit: the Visual Arts quantity, its reference object, the observation or browser rule, the transformation, and the decision being made. Keeping those fields visible prevents a familiar label from hiding a changed basis.
A fictional linear scene value of 0.18 represents a chosen mid-grey exposure reference. An output transform maps it to 10 cd/m² on a 100-nit target in the stipulated example. Store both values and the transform between them.
First-principles test: ask what physical, observational or algorithmic operation produces each value in this chapter. Then ask whether the second value could change while the first stays fixed. If it can, the two labels are not substitutes. For Visual Arts, this counterfactual is often clearer than a definition learned in isolation.
Keep the scene value fixed and change only the display peak from 100 to 1,000 nits. Predict a different rendered output while the recorded scene exposure stays fixed.
Evidence check: another reader should be able to reproduce the narrow result from the stored definitions, inputs and intermediate steps, then identify the nearest stronger claim that the evidence does not support.
The numeric mapping depends on the named output transform and target; this article does not prescribe a creative grade.
CHAPTER 2 OF 20 · Define the quantities
2. Keep relative exposure separate from emitted luminance
A scene-linear value is proportional to captured relative exposure in its encoding model. Display luminance is light emitted or reflected by the reproduction. A value of one in a scene space does not universally mean one nit on a screen.
Trace the mechanism before interpreting the result. For Visual Arts, mark which value belongs to the source, which belongs to the measurement or layout algorithm, and which is a reported summary. A calculation is useful only when those roles remain attached.
Two pixels at 0.18 and 0.36 have a two-to-one scene-linear relation. After a compressive tone scale their display luminances may have a smaller ratio.
Practical transfer: build two records that share the headline number but differ in the condition discussed here. If the correct interpretation changes, the omitted condition is decision-relevant and belongs in the report. This test turns a vocabulary distinction into a reusable Visual Arts checking habit.
Double both scene values while holding the transform fixed and trace whether the display ratio remains two.
Practical check: keep one counterexample beside the worked case. If the wording fails when that one condition changes, narrow the sentence until it describes only the evidence actually in hand.
Camera calibration, scene adaptation and encoding conventions limit claims about absolute scene luminance.
An output transform commonly combines a rendering stage, including tone and colour mapping, with a display-encoding stage tied to primaries, white point and electro-optical behaviour. Collapsing the chain hides where a change occurred.
Use a compact decision table with columns for condition, input, operation, result and limit. This makes the comparison reproducible and exposes where a missing field would force an assumption instead of an observation.
Record scene ACES values, rendering-transform output, limiting display colourimetry and final encoded RGB as four columns.
Misconception check: a plausible output does not prove the right reference, denominator, ancestor, spectrum or observation rule was used. Reconstruct one intermediate step and one changed-context prediction. If either cannot be stated, keep the result provisional rather than strengthening its wording.
Change only the display encoding while holding the rendered colorimetry fixed. The code values can change without a new artistic rendering.
Transfer check: repeat the reasoning with one consequential condition changed while holding the rest fixed. State what changes, what stays comparable and why the original label alone was insufficient.
Implementation details vary by ACES version and target transform.
CHAPTER 4 OF 20 · Define the quantities
4. Separate tone mapping from display encoding
Tone mapping compresses or reshapes a scene range into a renderable range. Display encoding then represents the rendered colorimetry for a target transfer function and primaries. Applying the display curve alone is not a full tone map.
Preserve the raw inputs beside the derived result. If a later reader changes one condition, the original Visual Arts reasoning should still be recoverable without reverse-engineering a rounded headline or an unlabeled screenshot.
A highlight with scene value 8 is mapped to 80 nits by the teaching tone scale, then encoded to a target code value. Reverse only the encoding to recover 80 nits, not the original scene value 8.
Evidence design: retain both a positive case and a near-miss. The positive case shows when the rule applies; the near-miss changes one consequential condition and shows when the conclusion fails. Together they teach the boundary of the Visual Arts mechanism better than another unlabelled example.
Swap an sRGB-like encoding for a PQ target while keeping the rendered 80-nit value.
Audit check: retain the source version, units, settings, exclusions and rounding rule. A future correction should be able to alter one field without rebuilding the entire explanation from memory.
A display encoding may be invertible over its domain while the creative rendering is not uniquely invertible.
The central record uses scene-linear values 0.18, 1, 4 and 16. A stipulated 100-nit rendering maps them to 10, 46, 78 and 96 nits, preserving order while compressing the upper range.
Ask what would have to stay fixed for the comparison to be fair. Record those controls explicitly, then name the first uncontrolled condition that could reverse the conclusion. This is the practical transfer step, not an optional footnote.
Calculate adjacent scene ratios and display-luminance differences. The highlight values remain distinct even though their output spacing narrows.
Communication test: replace any sentence that says two things are “the same” or “different” with a sentence naming the operation, reference condition and supported consequence. The extra words are useful when they let a future reader repeat the decision without access to the original author.
Change the target peak to 1,000 nits with supplied outputs 14, 107, 300 and 620 nits. Compare which relation changed.
Decision check: end with an action or interpretation that follows from the stated conditions, followed by a boundary sentence explaining when a specialist rule, local procedure or additional measurement is needed.
The table is invented to teach dependency; it is not the ACES 2 formula.
CHAPTER 6 OF 20 · Work the mechanism
6. Keep scene-linear and log working encodings distinct
A logarithmic working encoding can represent scene-referred exposure relationships compactly while its code values are not linear light. Scene-referred describes the reference of the data, not whether the numbers are linear.
Write the evidence in the order a reader can audit: the Visual Arts quantity, its reference object, the observation or browser rule, the transformation, and the decision being made. Keeping those fields visible prevents a familiar label from hiding a changed basis.
Encode two scene-linear values with a stipulated log rule, adjust exposure in the proper working operation and decode before comparing energy ratios.
First-principles test: ask what physical, observational or algorithmic operation produces each value in this chapter. Then ask whether the second value could change while the first stays fixed. If it can, the two labels are not substitutes. For Visual Arts, this counterfactual is often clearer than a definition learned in isolation.
Keep the pictured scene fixed and switch from linear ACES to ACEScct-like working values.
Evidence check: another reader should be able to reproduce the narrow result from the stored definitions, inputs and intermediate steps, then identify the nearest stronger claim that the evidence does not support.
Use the published encoding specification; a log code difference is not automatically a stop count.
CHAPTER 7 OF 20 · Work the mechanism
7. Audit exposure adjustments before rendering
A scene-referred exposure adjustment multiplies or offsets relative exposure before the output transform. A display-referred brightness edit changes an already rendered reproduction and can clip or distort a different range.
Trace the mechanism before interpreting the result. For Visual Arts, mark which value belongs to the source, which belongs to the measurement or layout algorithm, and which is a reported summary. A calculation is useful only when those roles remain attached.
Raise the central scene record by one stop before tone mapping, then compare it with multiplying the final display code values. The highlight behaviour differs.
Practical transfer: build two records that share the headline number but differ in the condition discussed here. If the correct interpretation changes, the omitted condition is decision-relevant and belongs in the report. This test turns a vocabulary distinction into a reusable Visual Arts checking habit.
Lower exposure before rendering while keeping the display target fixed.
Practical check: keep one counterexample beside the worked case. If the wording fails when that one condition changes, narrow the sentence until it describes only the evidence actually in hand.
A creative look can include operations beyond exposure; store operation order.
White balance and chromatic adaptation operate with declared source and destination conditions. Performing them after display rendering is not automatically equivalent to adjusting scene-referred data before the rendering transform.
Use a compact decision table with columns for condition, input, operation, result and limit. This makes the comparison reproducible and exposes where a missing field would force an assumption instead of an observation.
Correct a fictional tungsten scene before the output transform, then compare with neutralising the rendered RGB values on one display.
Misconception check: a plausible output does not prove the right reference, denominator, ancestor, spectrum or observation rule was used. Reconstruct one intermediate step and one changed-context prediction. If either cannot be stated, keep the result provisional rather than strengthening its wording.
Change the assumed source illuminant while preserving the recorded pixels.
Transfer check: repeat the reasoning with one consequential condition changed while holding the rest fixed. State what changes, what stays comparable and why the original label alone was insufficient.
No single adjustment recovers spectral scene information absent from the capture.
Scene-referred wide-gamut values can include negative or very large components that are meaningful for processing but problematic for display mapping. Reference gamut compression and output gamut mapping have different purposes and positions.
Preserve the raw inputs beside the derived result. If a later reader changes one condition, the original Visual Arts reasoning should still be recoverable without reverse-engineering a rounded headline or an unlabeled screenshot.
Track one wide-gamut highlight through reference compression, rendering and target-gamut limiting instead of clipping it at the input.
Evidence design: retain both a positive case and a near-miss. The positive case shows when the rule applies; the near-miss changes one consequential condition and shows when the conclusion fails. Together they teach the boundary of the Visual Arts mechanism better than another unlabelled example.
Hold the scene colour fixed and choose a narrower target gamut.
Audit check: retain the source version, units, settings, exclusions and rounding rule. A future correction should be able to alter one field without rebuilding the entire explanation from memory.
A gamut warning or clipping flag does not reveal the whole transform chain.
A rendering depends on the target peak, black level, surround and viewing assumptions. The same scene-referred image can have different display-referred versions that are all intentional.
Ask what would have to stay fixed for the comparison to be fair. Record those controls explicitly, then name the first uncontrolled condition that could reverse the conclusion. This is the practical transfer step, not an optional footnote.
Render the central case for 100-nit SDR and 1,000-nit HDR targets and store both output-transform identifiers.
Communication test: replace any sentence that says two things are “the same” or “different” with a sentence naming the operation, reference condition and supported consequence. The extra words are useful when they let a future reader repeat the decision without access to the original author.
Raise the target black level while keeping peak white fixed. Predict changed shadow rendering.
Decision check: end with an action or interpretation that follows from the stated conditions, followed by a boundary sentence explaining when a specialist rule, local procedure or additional measurement is needed.
Real displays may not achieve their nominal target uniformly or indefinitely.
A file path and colour-space label do not prove which transforms were applied. Metadata should record input, look and output transforms and whether any step was baked into the pixels.
Write the evidence in the order a reader can audit: the Visual Arts quantity, its reference object, the observation or browser rule, the transformation, and the decision being made. Keeping those fields visible prevents a familiar label from hiding a changed basis.
Two files both named final.exr contain the same scene values, but one has a baked display transform. The evidence receipt distinguishes them before compositing.
First-principles test: ask what physical, observational or algorithmic operation produces each value in this chapter. Then ask whether the second value could change while the first stays fixed. If it can, the two labels are not substitutes. For Visual Arts, this counterfactual is often clearer than a definition learned in isolation.
Remove the transform identifier and ask which reproduction becomes unrecoverable.
Evidence check: another reader should be able to reproduce the narrow result from the stored definitions, inputs and intermediate steps, then identify the nearest stronger claim that the evidence does not support.
Metadata can be wrong; validate against controlled test values and workflow records.
Applying an output transform twice treats display-referred data as though it were still scene-referred. Contrast, highlights and colour can change even when the second transform reports no error.
Trace the mechanism before interpreting the result. For Visual Arts, mark which value belongs to the source, which belongs to the measurement or layout algorithm, and which is a reported summary. A calculation is useful only when those roles remain attached.
Pass the 100-nit rendered teaching values through the same tone scale again and flag the altered output.
Practical transfer: build two records that share the headline number but differ in the condition discussed here. If the correct interpretation changes, the omitted condition is decision-relevant and belongs in the report. This test turns a vocabulary distinction into a reusable Visual Arts checking habit.
Insert an explicit data-state check before the second transform.
Practical check: keep one counterexample beside the worked case. If the wording fails when that one condition changes, narrow the sentence until it describes only the evidence actually in hand.
Software may use ambiguous labels such as display, view or output; inspect the actual transform.
One scene master can support SDR and HDR outputs through separate validated transforms. The outputs need not share identical code values or highlight placement to depict the same intended scene.
Use a compact decision table with columns for condition, input, operation, result and limit. This makes the comparison reproducible and exposes where a missing field would force an assumption instead of an observation.
Create two receipts from the central scene values, one for a 100-nit target and one for 1,000 nits.
Misconception check: a plausible output does not prove the right reference, denominator, ancestor, spectrum or observation rule was used. Reconstruct one intermediate step and one changed-context prediction. If either cannot be stated, keep the result provisional rather than strengthening its wording.
Change only the HDR mastering peak and regenerate that output.
Transfer check: repeat the reasoning with one consequential condition changed while holding the rest fixed. State what changes, what stays comparable and why the original label alone was insufficient.
Creative trims may be needed; automatic transforms do not guarantee identical audience experience.
Light transport and many compositing operations are most coherent in a scene-linear state. Compositing two display-rendered images can combine already compressed values and halos incorrectly.
Preserve the raw inputs beside the derived result. If a later reader changes one condition, the original Visual Arts reasoning should still be recoverable without reverse-engineering a rounded headline or an unlabeled screenshot.
Blend a bright foreground over a background before and after the teaching tone map and compare edge energy.
Evidence design: retain both a positive case and a near-miss. The positive case shows when the rule applies; the near-miss changes one consequential condition and shows when the conclusion fails. Together they teach the boundary of the Visual Arts mechanism better than another unlabelled example.
Keep geometry fixed and change operation order.
Audit check: retain the source version, units, settings, exclusions and rounding rule. A future correction should be able to alter one field without rebuilding the entire explanation from memory.
Some artistic composites intentionally use display-like spaces; label the choice and consequences.
A print transform targets paper, ink, viewing light and limited reflection rather than emitted display luminance. The scene master can remain the source while the output-referred state changes.
Ask what would have to stay fixed for the comparison to be fair. Record those controls explicitly, then name the first uncontrolled condition that could reverse the conclusion. This is the practical transfer step, not an optional footnote.
Map the central scene through a proof transform and keep that output separate from the monitor rendering.
Communication test: replace any sentence that says two things are “the same” or “different” with a sentence naming the operation, reference condition and supported consequence. The extra words are useful when they let a future reader repeat the decision without access to the original author.
Change paper white and black density while preserving the scene file.
Decision check: end with an action or interpretation that follows from the stated conditions, followed by a boundary sentence explaining when a specialist rule, local procedure or additional measurement is needed.
A soft proof depends on calibrated display and trustworthy printer characterization.
Archival scene-referred data can preserve flexibility for future output transforms, but only with encoding, primaries, metadata and transform provenance. An unlabeled floating-point file is not self-explanatory.
Write the evidence in the order a reader can audit: the Visual Arts quantity, its reference object, the observation or browser rule, the transformation, and the decision being made. Keeping those fields visible prevents a familiar label from hiding a changed basis.
Package scene data, transform IDs, viewing target and a rendered reference image.
First-principles test: ask what physical, observational or algorithmic operation produces each value in this chapter. Then ask whether the second value could change while the first stays fixed. If it can, the two labels are not substitutes. For Visual Arts, this counterfactual is often clearer than a definition learned in isolation.
Remove the output identifier and test whether another reader can reproduce the reference.
Evidence check: another reader should be able to reproduce the narrow result from the stored definitions, inputs and intermediate steps, then identify the nearest stronger claim that the evidence does not support.
Long-term readability also depends on open specifications, file integrity and software support.
CHAPTER 17 OF 20 · Report and continue
17. Repair brighter code means brighter scene
Display code values reflect rendering, encoding and target conditions. A higher code in one delivered file does not prove a proportionally brighter source exposure or brighter output on another device.
Trace the mechanism before interpreting the result. For Visual Arts, mark which value belongs to the source, which belongs to the measurement or layout algorithm, and which is a reported summary. A calculation is useful only when those roles remain attached.
Compare the same scene highlight in SDR and HDR files; the numerical codes differ without a new capture.
Practical transfer: build two records that share the headline number but differ in the condition discussed here. If the correct interpretation changes, the omitted condition is decision-relevant and belongs in the report. This test turns a vocabulary distinction into a reusable Visual Arts checking habit.
Decode both to target luminance and trace back only through known transforms.
Practical check: keep one counterexample beside the worked case. If the wording fails when that one condition changes, narrow the sentence until it describes only the evidence actually in hand.
A non-invertible rendering cannot reconstruct a unique scene from the display image.
Clipping discards values above a threshold, while a tone scale can compress a broad range and preserve ordered highlight differences. Both can limit output, but their visual and numerical consequences differ.
Use a compact decision table with columns for condition, input, operation, result and limit. This makes the comparison reproducible and exposes where a missing field would force an assumption instead of an observation.
Map scene values 4, 8 and 16 to 78, 90 and 96 nits; clipping at 80 would merge the last two.
Misconception check: a plausible output does not prove the right reference, denominator, ancestor, spectrum or observation rule was used. Reconstruct one intermediate step and one changed-context prediction. If either cannot be stated, keep the result provisional rather than strengthening its wording.
Move the clipping threshold and compare with a monotonic compressive curve.
Transfer check: repeat the reasoning with one consequential condition changed while holding the rest fixed. State what changes, what stays comparable and why the original label alone was insufficient.
Local tone mapping can depend on neighbourhood and requires additional documentation.
Store source device or synthetic origin, input transform, scene encoding, exposure and look operations, gamut operations, output transform version, target primaries, white, black, peak, transfer function and rendered file.
Preserve the raw inputs beside the derived result. If a later reader changes one condition, the original Visual Arts reasoning should still be recoverable without reverse-engineering a rounded headline or an unlabeled screenshot.
Package the central SDR and HDR examples so another colourist can identify every state transition.
Evidence design: retain both a positive case and a near-miss. The positive case shows when the rule applies; the near-miss changes one consequential condition and shows when the conclusion fails. Together they teach the boundary of the Visual Arts mechanism better than another unlabelled example.
Delete the data-state label and test whether double application becomes likely.
Audit check: retain the source version, units, settings, exclusions and rounding rule. A future correction should be able to alter one field without rebuilding the entire explanation from memory.
A screenshot alone cannot prove the scene values or transform chain.
A defensible report says which values are scene-referred, which output transform produced the display-referred image and which display conditions the encoding targets. It avoids treating either state as universal truth.
Ask what would have to stay fixed for the comparison to be fair. Record those controls explicitly, then name the first uncontrolled condition that could reverse the conclusion. This is the practical transfer step, not an optional footnote.
Write: “The ACES scene master was rendered with the named 100-nit output transform; delivered RGB values are display-referred and must not be reused as scene-linear inputs.”
Communication test: replace any sentence that says two things are “the same” or “different” with a sentence naming the operation, reference condition and supported consequence. The extra words are useful when they let a future reader repeat the decision without access to the original author.
Change the target transform and identify which derived files must be regenerated.
Decision check: end with an action or interpretation that follows from the stated conditions, followed by a boundary sentence explaining when a specialist rule, local procedure or additional measurement is needed.
Judging artistic intent still requires the approved viewing environment and human review.
Continue through the verified reading route
Display calibration and ICC profiling
Display calibration and ICC profiling. Use this when the display state and its measured characterization need separate evidence.
Profile assignment and destination conversion
Profile assignment and destination conversion. Use this for document interpretation and conversion after the image state is known.
Device gamut and rendering intent
Device gamut and rendering intent. Use this when a target gamut and the mapping into it are being merged.
Out-of-gamut clipping and desaturation
Out-of-gamut clipping and desaturation. Use this for clipped channels versus a deliberate reduction in colourfulness.
Frequently asked questions
Can display-referred RGB be used as scene-linear data?
Only after an appropriate known inverse process where one exists; ordinary delivered RGB has already been rendered and encoded.
Does tone mapping simply clip highlights?
No. A monotonic tone scale can compress and preserve highlight ordering; clipping merges values beyond a boundary.
Why keep more than one rendered version?
Different display peaks, gamuts, transfer functions and viewing conditions require different output transforms from the same scene master.
Authoritative sources and further reading
ACES Output Transforms. The official ACES documentation defines the transformation from scene-referred ACES data to rendered output for a specified display.
ACES tone-mapping technical details. The official design notes describe the tone scale, target peak luminance and display encoding stages.
ACES System overview. The official overview identifies scene-linear working data, rendering transforms and display-referred targets.
ACES Metadata File specification. The specification records input, look and output transforms so image states are not silently confused.
Return to the Visual Arts subject guide, The Gold Standard of Education, or the How X Works hub.