Colour difference and luminance contrast can both be described informally as how far apart colours look, yet they answer different reader jobs. A Delta E formula works on coordinates in a declared colour space and reference condition. A contrast ratio works on relative luminance values for a particular visual relationship.
Pair A is stipulated with relative luminances0.20 and0.20, plus Lab coordinates(50,40,20) and(50,-40,20). Its contrast ratio is1:1 while Delta E76 is80. Pair B changes L* from50 to60 with a*=b*=10; its Delta E76 is10, while supplied luminances0.184 and0.281 give contrast about1.414:1. The numbers deliberately show that neither metric can stand in for the other.
The examples are educational. They do not certify accessibility, approve artwork, set a production tolerance, replace calibrated measurement or make a universal claim about perception.
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
Use a colour-difference metric for separation in a declared colour space and a luminance contrast ratio for the specified foreground-background relationship.
ROUTE 1 · CHAPTERS 1–5
Define the quantities
Fix the object, unit, denominator and reference condition.
ROUTE 2 · CHAPTERS 6–10
Work the central example
Carry the calculation through a complete, checkable case.
ROUTE 3 · CHAPTERS 11–15
Audit the evidence
Test sensitivity, measurement choices and rival explanations.
ROUTE 5 · CHAPTERS 21–24
Report and continue
Repair shortcuts, state limits and follow verified routes.
Chapter contents
Define the quantities · Chapters 1–5
2. Calculate luminance contrast
Work the central example · Chapters 6–10
7. Work a lightness-change pair
8. Compare Delta E formula families
Audit the evidence · Chapters 11–15
11. Audit adjacency and geometry
12. Audit display and print conditions
Transfer the reasoning · Chapters 16–20
16. Transfer to data visualisation
17. Transfer to painting reproduction
18. Transfer to brand colour quality
Report and continue · Chapters 21–24
21. Repair large Delta E means high contrast
22. Repair passing contrast means colours match
Begin with the quantity rather than the convenient number. A luminance contrast ratio compares relative light output of adjacent colours; a colour-difference metric estimates separation in a specified colour space and viewing reference.
Work the teaching case in a visible order. Pair A is assigned relative luminances0.20 and0.20 but Lab values(50,40,20) and(50,-40,20).
Make the reasoning independently checkable. Keep colour space, white point and metric beside every difference.
Test transfer by changing one condition while holding the others still. Change only a* while holding L* fixed.
Keep the claim boundary visible. Neither metric alone is an artistic-quality score.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Keep colour space, white point and metric beside every difference. Then run the changed case: Change only a* while holding L* fixed. Keep both outcomes visible.
The evidence receipt for “Name the two questions” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Keep observation, transformation and conclusion on separate lines. For the WCAG-style diagnostic, place the lighter relative luminance first and compute(L1+0.05)/(L2+0.05).
Work the teaching case in a visible order. Pair A gives(0.20+0.05)/(0.20+0.05)=1.00:1.
Make the reasoning independently checkable. Reverse the inputs and confirm ordering restores the same ratio.
Test transfer by changing one condition while holding the others still. Change the second luminance to0.10.
Keep the claim boundary visible. The ratio addresses a defined contrast use, not every visual perception.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: Reverse the inputs and confirm ordering restores the same ratio. For transfer, Change the second luminance to0.10. State why the conclusion moved or stayed.
The evidence receipt for “Calculate luminance contrast” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Use the short answer only after fixing the boundary and units. Delta E76 is Euclidean distance between Lab coordinates under the stated reference conditions.
Work the teaching case in a visible order. Pair A differs by80 units in a*: sqrt(0²+80²+0²)=80.
Make the reasoning independently checkable. Square each coordinate difference before summing.
Test transfer by changing one condition while holding the others still. Change b* by30 as well and recompute.
Keep the claim boundary visible. Delta E76 is a first approximation with known non-uniformity.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Square each coordinate difference before summing. Next, Change b* by30 as well and recompute. Preserve any mismatch as evidence.
The evidence receipt for “Calculate Delta E76” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
CHAPTER 4 OF 24 · Define the quantities
4. Keep L star distinct from relative luminance
Trace the mechanism in the order another reader can reproduce. CIE L* is a nonlinear lightness coordinate derived from Y relative to a reference white; WCAG relative luminance uses a specified sRGB transformation.
Work the teaching case in a visible order. The teaching values are stipulated in their own declared systems.
Make the reasoning independently checkable. Do not substitute the number50 for a relative luminance of0.50.
Test transfer by changing one condition while holding the others still. Convert a supplied Y/Yn through the L* piecewise function.
Keep the claim boundary visible. Similar words such as lightness and luminance do not make scales interchangeable.
Turn the claim into a falsifiable test. First, Do not substitute the number50 for a relative luminance of0.50. Then change one consequential condition: Convert a supplied Y/Yn through the L* piecewise function. Name the observation that would count against the explanation.
The evidence receipt for “Keep L star distinct from relative luminance” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
CHAPTER 5 OF 24 · Define the quantities
5. Interpret a large colour difference with low contrast
Treat a plausible result as a prompt for a check rather than proof. Two equal-lightness colours can be far apart in chromatic coordinates while offering little tonal separation.
Work the teaching case in a visible order. Pair A has Delta E76=80 and contrast1:1.
Make the reasoning independently checkable. Check a grayscale or luminance-only view alongside the colour view.
Test transfer by changing one condition while holding the others still. Reduce a* separation to20.
Keep the claim boundary visible. Display gamut, adaptation and viewers affect appearance.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Check a grayscale or luminance-only view alongside the colour view. Then run the changed case: Reduce a* separation to20. Keep both outcomes visible.
The evidence receipt for “Interpret a large colour difference with low contrast” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Begin with the quantity rather than the convenient number. Record both source colour values, reference white, luminances, Lab coordinates, formulas and results.
Work the teaching case in a visible order. The receipt reports1.00:1 and Delta E76=80 without calling either wrong.
Make the reasoning independently checkable. Recompute both from the saved values.
Test transfer by changing one condition while holding the others still. Swap the colours and confirm both metrics remain unchanged.
Keep the claim boundary visible. The pair is invented to expose the distinction, not recommend a palette.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: Recompute both from the saved values. For transfer, Swap the colours and confirm both metrics remain unchanged. State why the conclusion moved or stayed.
The evidence receipt for “Complete the central pair” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Keep observation, transformation and conclusion on separate lines. Pair B uses Lab(50,10,10) and(60,10,10), so Delta E76=10.
Work the teaching case in a visible order. Using supplied relative luminances0.184 and0.281 gives contrast about1.414:1.
Make the reasoning independently checkable. Check both arithmetic paths independently.
Test transfer by changing one condition while holding the others still. Increase the second L* to70 with chromatic coordinates fixed.
Keep the claim boundary visible. Real conversions require a defined colour profile and white point.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Check both arithmetic paths independently. Next, Increase the second L* to70 with chromatic coordinates fixed. Preserve any mismatch as evidence.
The evidence receipt for “Work a lightness-change pair” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Use the short answer only after fixing the boundary and units. Delta E2000 adjusts weighting and interactions to address non-uniformities that Delta E76 leaves visible.
Work the teaching case in a visible order. Run a validated implementation on a published test pair rather than rewriting the complex formula casually.
Make the reasoning independently checkable. Record the metric name and version.
Test transfer by changing one condition while holding the others still. Compare Delta E76 and Delta E2000 for a small blue-region difference.
Keep the claim boundary visible. Thresholds from one metric cannot be transferred mechanically to another.
Turn the claim into a falsifiable test. First, Record the metric name and version. Then change one consequential condition: Compare Delta E76 and Delta E2000 for a small blue-region difference. Name the observation that would count against the explanation.
The evidence receipt for “Compare Delta E formula families” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Trace the mechanism in the order another reader can reproduce. Lab coordinates depend on the reference white and adaptation assumptions.
Work the teaching case in a visible order. The teaching receipt declares D50 or D65 rather than leaving the white point implicit.
Make the reasoning independently checkable. Convert both samples into one common reference before differencing.
Test transfer by changing one condition while holding the others still. Change the assumed illuminant.
Keep the claim boundary visible. Numerical difference across mismatched references is not a supported perceptual claim.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Convert both samples into one common reference before differencing. Then run the changed case: Change the assumed illuminant. Keep both outcomes visible.
The evidence receipt for “Preserve white point and adaptation” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Treat a plausible result as a prompt for a check rather than proof. Device RGB values acquire colour meaning through a profile; clipping during conversion can change both metrics.
Work the teaching case in a visible order. Convert a supplied display-P3 pair into sRGB and flag any clipped channel.
Make the reasoning independently checkable. Compare pre- and post-conversion coordinates.
Test transfer by changing one condition while holding the others still. Use a wider-gamut destination.
Keep the claim boundary visible. A hex code without colour-space context is incomplete evidence.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: Compare pre- and post-conversion coordinates. For transfer, Use a wider-gamut destination. State why the conclusion moved or stayed.
The evidence receipt for “Preserve colour profile and gamut” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Begin with the quantity rather than the convenient number. A contrast ratio is applied to adjacent foreground-background relationships; area, edge, pattern and spatial frequency influence visibility.
Work the teaching case in a visible order. The same two fills are tested as large fields and one-pixel lines.
Make the reasoning independently checkable. Inspect the actual geometry at output size.
Test transfer by changing one condition while holding the others still. Add a dark outline without changing fills.
Keep the claim boundary visible. The numeric pair cannot represent every spatial condition.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Inspect the actual geometry at output size. Next, Add a dark outline without changing fills. Preserve any mismatch as evidence.
The evidence receipt for “Audit adjacency and geometry” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Keep observation, transformation and conclusion on separate lines. Emission, paper reflectance, ink, ambient light and calibration alter the realised stimulus.
Work the teaching case in a visible order. Compare a profiled display proof with a measured print sample.
Make the reasoning independently checkable. Keep instrument, illumination and substrate metadata.
Test transfer by changing one condition while holding the others still. Dim the viewing light.
Keep the claim boundary visible. A source-file metric is not a guarantee of reproduced appearance.
Turn the claim into a falsifiable test. First, Keep instrument, illumination and substrate metadata. Then change one consequential condition: Dim the viewing light. Name the observation that would count against the explanation.
The evidence receipt for “Audit display and print conditions” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Use the short answer only after fixing the boundary and units. Spectrophotometer repeatability, positioning and surface texture can change measured Lab values.
Work the teaching case in a visible order. Measure each patch five times and report spread.
Make the reasoning independently checkable. Check whether observed Delta E exceeds measurement variation.
Test transfer by changing one condition while holding the others still. Rotate a textured sample.
Keep the claim boundary visible. A small numerical difference near the noise floor needs uncertainty.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Check whether observed Delta E exceeds measurement variation. Then run the changed case: Rotate a textured sample. Keep both outcomes visible.
The evidence receipt for “Audit measurement uncertainty” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Trace the mechanism in the order another reader can reproduce. Colour-difference formulas model typical perception; individual vision, adaptation and task still matter.
Work the teaching case in a visible order. Pair metric results with an appropriate viewing test.
Make the reasoning independently checkable. Separate observer reports from instrument calculations.
Test transfer by changing one condition while holding the others still. Include viewers with varied colour vision.
Keep the claim boundary visible. User evidence does not turn one sample into a universal threshold.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: Separate observer reports from instrument calculations. For transfer, Include viewers with varied colour vision. State why the conclusion moved or stayed.
The evidence receipt for “Audit human evidence” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Treat a plausible result as a prompt for a check rather than proof. Luminance contrast and Delta E both respond to some lightness changes, so treating them as independent points in one score can overweight the same feature.
Work the teaching case in a visible order. Keep the two outputs side by side instead of summing them.
Make the reasoning independently checkable. Explain what decision each metric supports.
Test transfer by changing one condition while holding the others still. Create a weighted score and expose its arbitrary weights.
Keep the claim boundary visible. Composite scores require validation for the task.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Explain what decision each metric supports. Next, Create a weighted score and expose its arbitrary weights. Preserve any mismatch as evidence.
The evidence receipt for “Avoid double counting” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Begin with the quantity rather than the convenient number. Categories may need chromatic separation, while marks and text also need sufficient luminance contrast against their backgrounds.
Work the teaching case in a visible order. Test adjacent chart series and each label-background pair separately.
Make the reasoning independently checkable. Add shape or texture redundancy.
Test transfer by changing one condition while holding the others still. Print the chart in grayscale.
Keep the claim boundary visible. No single Delta E threshold guarantees readable charts.
Turn the claim into a falsifiable test. First, Add shape or texture redundancy. Then change one consequential condition: Print the chart in grayscale. Name the observation that would count against the explanation.
The evidence receipt for “Transfer to data visualisation” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Keep observation, transformation and conclusion on separate lines. A reproduction can be close in Lab at sampled patches yet lose local edge contrast, gloss or texture.
Work the teaching case in a visible order. Compare patch Delta E with tonal relationships across the composition.
Make the reasoning independently checkable. Sample multiple regions and viewing angles.
Test transfer by changing one condition while holding the others still. Change varnish gloss while pigments remain.
Keep the claim boundary visible. Instrument geometry and material appearance must be reported.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Sample multiple regions and viewing angles. Then run the changed case: Change varnish gloss while pigments remain. Keep both outcomes visible.
The evidence receipt for “Transfer to painting reproduction” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Use the short answer only after fixing the boundary and units. Production tolerances often use a declared colour-difference metric, while legibility checks use contrast relationships.
Work the teaching case in a visible order. The logo patch passes a stipulated Delta E tolerance but its text fails the declared contrast requirement.
Make the reasoning independently checkable. Do not substitute one pass for the other.
Test transfer by changing one condition while holding the others still. Place the same brand colour on a darker background.
Keep the claim boundary visible. Tolerance rules belong to the actual production specification.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: Do not substitute one pass for the other. For transfer, Place the same brand colour on a darker background. State why the conclusion moved or stayed.
The evidence receipt for “Transfer to brand colour quality” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Trace the mechanism in the order another reader can reproduce. Contrast criteria address defined text, graphics or interface relationships; colour difference can support but does not replace those checks.
Work the teaching case in a visible order. Evaluate the actual foreground and adjacent background luminances.
Make the reasoning independently checkable. Use the normative criterion relevant to the element.
Test transfer by changing one condition while holding the others still. Add a hue-only state change.
Keep the claim boundary visible. This article is education, not a conformance report.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Use the normative criterion relevant to the element. Next, Add a hue-only state change. Preserve any mismatch as evidence.
The evidence receipt for “Transfer to accessibility review” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Treat a plausible result as a prompt for a check rather than proof. Record source space, profile, white point, conversion, luminances, Lab coordinates, Delta E formula, geometry, viewing condition and uncertainty.
Work the teaching case in a visible order. Another reader reconstructs Pair A’s two metrics.
Make the reasoning independently checkable. Remove the white point and identify the unsupported comparison.
Test transfer by changing one condition while holding the others still. Version the receipt after profile conversion.
Keep the claim boundary visible. Reproducibility does not guarantee identical perception.
Turn the claim into a falsifiable test. First, Remove the white point and identify the unsupported comparison. Then change one consequential condition: Version the receipt after profile conversion. Name the observation that would count against the explanation.
The evidence receipt for “Build a reproducible colour receipt” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
CHAPTER 21 OF 24 · Report and continue
21. Repair large Delta E means high contrast
Begin with the quantity rather than the convenient number. Chromatic separation can be large while relative luminance is equal.
Work the teaching case in a visible order. Pair A provides the counterexample: Delta E76=80 and contrast1:1.
Make the reasoning independently checkable. Check both metrics before using the word contrast.
Test transfer by changing one condition while holding the others still. Hold L* fixed and rotate hue.
Keep the claim boundary visible. The result depends on the declared colour coordinates and rendering.
Before accepting the result, predict what should change and what should remain fixed. Apply this check: Check both metrics before using the word contrast. Then run the changed case: Hold L* fixed and rotate hue. Keep both outcomes visible.
The evidence receipt for “Repair large Delta E means high contrast” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
CHAPTER 22 OF 24 · Report and continue
22. Repair passing contrast means colours match
Keep observation, transformation and conclusion on separate lines. Two colours can have a useful luminance ratio yet differ greatly in hue or fail a reproduction tolerance.
Work the teaching case in a visible order. Compare Pair B’s contrast with its colour-difference purpose.
Make the reasoning independently checkable. State whether the task is legibility, matching or categorical separation.
Test transfer by changing one condition while holding the others still. Keep contrast fixed while changing chromatic coordinates.
Keep the claim boundary visible. One metric cannot answer a different reader job.
Build an audit receipt with input, unit, model, calculation, observation and limit. Use this independent check: State whether the task is legibility, matching or categorical separation. For transfer, Keep contrast fixed while changing chromatic coordinates. State why the conclusion moved or stayed.
The evidence receipt for “Repair passing contrast means colours match” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Use the short answer only after fixing the boundary and units. The method supports choosing evidence for tonal visibility, colour matching or perceptual separation without collapsing them.
Work the teaching case in a visible order. It can reject a Delta E value offered as a contrast ratio.
Make the reasoning independently checkable. Write the visual task before choosing a metric.
Test transfer by changing one condition while holding the others still. Move from a teaching pair to a production artwork.
Keep the claim boundary visible. Use current standards, calibrated workflows and appropriate observer testing for production decisions.
Ask a second reader to reproduce the decision from the stated evidence. They should be able to apply: Write the visual task before choosing a metric. Next, Move from a teaching pair to a production artwork. Preserve any mismatch as evidence.
The evidence receipt for “State the supported decision” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
CHAPTER 24 OF 24 · Report and continue
24. Continue through verified visual-arts routes
Trace the mechanism in the order another reader can reproduce. This distinction sits beside grayscale survival, simultaneous contrast, gamut clipping and spot-process simulation.
Work the teaching case in a visible order. Choose the sibling that owns the next failure mode.
Make the reasoning independently checkable. Keep physical colour, contextual appearance, device gamut and metric purpose separate.
Test transfer by changing one condition while holding the others still. Move to simultaneous contrast without treating the surround effect as a file-value change.
Keep the claim boundary visible. This article owns colour-difference versus luminance-contrast reasoning under declared viewing and colour-management conditions.
Turn the claim into a falsifiable test. First, Keep physical colour, contextual appearance, device gamut and metric purpose separate. Then change one consequential condition: Move to simultaneous contrast without treating the surround effect as a file-value change. Name the observation that would count against the explanation.
The evidence receipt for “Continue through verified visual-arts routes” records the source, input, transformation, changed-context prediction and limit on separate lines. Keep colour space, profile, white point, Delta E formula, relative luminance, geometry and viewing condition attached to the result.
Continue through the verified reading route
Colour contrast in grayscale
Colour contrast in grayscale. Use this when hue separation collapses after luminance-only reproduction.
Simultaneous contrast and physical colour
Simultaneous contrast and physical colour. Use this when surroundings alter appearance without changing the measured sample.
Out-of-gamut clipping and desaturation
Out-of-gamut clipping and desaturation. Use this when device conversion changes colour coordinates.
Frequently asked questions
Can a colour pair have large Delta E and low luminance contrast?
Yes. Equal-lightness colours can be far apart in chromatic coordinates.
Can Delta E replace a contrast-ratio check?
No. The metrics use different inputs and support different decisions.
Which Delta E formula should be reported?
Name the exact formula, colour space, white point and viewing or measurement conditions; thresholds do not transfer automatically.
Authoritative sources and further reading
W3C CSS Color Module Level 4. W3C specification describing Lab, colour conversion and Delta E formula families.
W3C WCAG 2.2 contrast definition. W3C definition of relative-luminance contrast ratio.
W3C Understanding Non-text Contrast. W3C guidance on the scope, geometry and limits of contrast requirements.
Return to the Visual Arts subject guide, The Gold Standard of Education, or the How X Works hub.