1.1_Basic_Color_Science.md

1.1 Basic Color Science

Theory: CIE XYZ, Color Matching Functions, Spectral Power Distributions

CIE Standard Observer System

CIE XYZ [device-independent tristimulus color space based on human visual system physiology established in 1931] forms the mathematical foundation of all modern colorimetry. The system derives from color matching experiments where observers adjusted three primaries to match spectral colors.

CIE 1931 2° Standard Observer [spectral sensitivity curves derived from 2° foveal visual field experiments with 17 observers] defines three color matching functions [CMF: $\bar{x}(\lambda), \bar{y}(\lambda), \bar{z}(\lambda)$]:

• $\bar{x}(\lambda)$: Short and long wavelength sensitivity (red-blue regions)
• $\bar{y}(\lambda)$: Luminous efficiency function $V(\lambda)$ (photopic vision)
• $\bar{z}(\lambda)$: Short wavelength sensitivity (blue region)

The $\bar{y}(\lambda)$ function IS the CIE photopic luminous efficiency function, directly linking colorimetry to photometry.

CIE 1964 10° Standard Observer [supplementary CMF derived from 10° visual field experiments] accounts for macular pigment distribution differences at wider viewing angles. Use 10° observer for cinema applications (large screens, wide viewing angles).

Spectral Power Distributions

SPD [Spectral Power Distribution: intensity of electromagnetic radiation at each wavelength across the visible spectrum 380-780nm, typically measured at 1-5nm intervals] completely characterizes any light source.

Illuminant standards:

• D65 [CIE Standard Illuminant representing average daylight with correlated color temperature 6504K]: $x=0.3127, y=0.3290$
• D50 [CIE Standard Illuminant representing horizon light with CCT 5003K]: $x=0.3457, y=0.3585$
• A [CIE Standard Illuminant representing incandescent tungsten light with CCT 2856K]: $x=0.4476, y=0.4074$

Cinema projectors use xenon illumination with SPD differing significantly from D65. DCI-P3 white point ($x=0.314, y=0.351$) accounts for this difference.

Tristimulus Integration

Tristimulus values [X, Y, Z] result from integrating the product of illuminant SPD, object reflectance, and CMFs:

$$ X = k \int_{380nm}^{780nm} S(\lambda) \cdot \bar{x}(\lambda) \cdot R(\lambda) , d\lambda $$

$$ Y = k \int_{380nm}^{780nm} S(\lambda) \cdot \bar{y}(\lambda) \cdot R(\lambda) , d\lambda $$

$$ Z = k \int_{380nm}^{780nm} S(\lambda) \cdot \bar{z}(\lambda) \cdot R(\lambda) , d\lambda $$

Where:

• $S(\lambda)$ = illuminant SPD (D65, D50, A, or custom measured)
• $R(\lambda)$ = object reflectance factor (0-1) or transmittance
• $k$ = normalization constant: $k = \frac{100}{\int_{380}^{780} S(\lambda) \cdot \bar{y}(\lambda) , d\lambda}$

Y tristimulus equals luminance for reflecting objects. For self-luminous displays, Y is proportional to luminance (cd/m²).

Photometry vs Radiometry

Radiometry [measurement of electromagnetic radiation across all wavelengths] uses physical units:

• Radiant flux: watts (W)
• Irradiance: W/m²
• Radiant intensity: W/sr

Photometry [measurement of light as perceived by human vision] weights radiometric quantities by $V(\lambda)$:

• Luminous flux: lumens (lm)
• Illuminance: lux (lx) or foot-candles (fc)
• Luminance: cd/m² or nits

Critical conversion: 683 lm/W at 555nm (peak photopic sensitivity). All other wavelengths weighted by $V(\lambda)$.

Cinema measurements:

• Screen luminance: 48 cd/m² (DCI standard for projection)
• Monitor white: 100 cd/m² (SDR) or 100-1000 cd/m² (HDR)
• Ambient illumination: <5 lux for critical color grading

Chromaticity Coordinates

Projecting XYZ onto 2D plane yields chromaticity coordinates:

$$ x = \frac{X}{X + Y + Z}, \quad y = \frac{Y}{X + Y + Z}, \quad z = \frac{Z}{X + Y + Z} $$

Since $x + y + z = 1$, only x and y needed to specify hue and saturation [perceived color characteristics independent of luminance].

Dominant wavelength [wavelength of spectrally pure light that matches the hue when mixed with white]: Intersection of line from white point through chromaticity to spectral locus. Purple colors (non-spectral) use complementary wavelength.

Excitation purity [ratio of distance from white point to chromaticity vs white point to spectral locus]: Percentage saturation relative to spectral purity.

$$ \text{Purity} = \frac{d_{white \to chroma}}{d_{white \to spectrum}} $$

Cinema white points:

• Rec.709: D65 ($x=0.3127, y=0.3290$)
• DCI-P3: DCI ($x=0.314, y=0.351$)
• P3-D65: D65 ($x=0.3127, y=0.3290$) with P3 primaries

Confusing DCI and D65 white points causes 6-8 ΔE color shifts, visible to all observers.

Uniform Chromaticity Diagrams

CIE 1976 u'v' [perceptually uniform chromaticity diagram where equal distances represent equal perceived color differences] corrects non-uniformity of xy diagram.

$$ u' = \frac{4X}{X + 15Y + 3Z}, \quad v' = \frac{9Y}{X + 15Y + 3Z} $$

Applications:

• Color difference calculations: $\Delta E'_{uv} = \sqrt{(\Delta u')^2 + (\Delta v')^2}$
• Gamut boundary analysis (MacAdam ellipses become circles)
• Tolerance evaluation (1 ΔE ≈ just-noticeable difference)

JzAzBz [perceptual uniform color space for HDR and WCG applications defined in SMPTE ST 2084:2014] extends u'v' concepts with luminance component for 3D uniformity.

RGB to XYZ Transformations

Linear RGB transforms to XYZ via 3×3 matrix derived from primary chromaticities and white point:

$$ \begin{bmatrix} X \ Y \ Z \end{bmatrix} = [M] \begin{bmatrix} R \ G \ B \end{bmatrix} $$

Rec.709 (ITU-R BT.709-6, 2015):

Primary Chromaticities:
R: (0.64, 0.33)  G: (0.30, 0.60)  B: (0.15, 0.06)  White: D65

[ X ]   [ 0.4124564  0.3575761  0.1804375 ] [ R ]
[ Y ] = [ 0.2126729  0.7151522  0.0721750 ] [ G ]
[ Z ]   [ 0.0193339  0.1191920  0.9503041 ] [ B ]

DCI-P3 (SMPTE RP 431-2, 2011):

Primary Chromaticities:
R: (0.680, 0.320)  G: (0.265, 0.690)  B: (0.150, 0.060)  White: DCI

[ X ]   [ 0.4451699  0.2771344  0.1722827 ] [ R ]
[ Y ] = [ 0.2466551  0.6723116  0.0810333 ] [ G ]
[ Z ]   [ 0.0098017  0.0679384  0.9806962 ] [ B ]

Rec.2020 (ITU-R BT.2020-2, 2015):

Primary Chromaticities:
R: (0.708, 0.292)  G: (0.170, 0.797)  B: (0.131, 0.046)  White: D65

[ X ]   [ 0.6369580  0.1446170  0.1688810 ] [ R ]
[ Y ] = [ 0.2627002  0.6779981  0.0593017 ] [ G ]
[ Z ]   [ 0.0000000  0.0280726  1.0609850 ] [ B ]

CRITICAL: These matrices apply ONLY to linear RGB (transfer function removed first). Apply to encoded values produces incorrect results.

Practice: Color Measurement Workflows

Instrument Selection Hierarchy

Spectroradiometer [measures complete SPD at 1-5nm intervals with single-photon precision]:

• Uses diffraction grating or prism to disperse light
• Provides absolute spectral data: 380-780nm typically
• Measurement uncertainty: ±1-2% (NIST-traceable calibration)
• Primary choice for: display calibration, custom illuminants, metamerism analysis

Best-in-class 2024: Konica Minolta CS-2000a, Photo Research PR-870

Colorimeter [measures tristimulus directly via filtered sensors or array detectors]:

• 3-4 filtered sensors matching CIE CMFs
• Fast measurement (1-10ms vs 1-10s for spectroradiometer)
• Measurement uncertainty: ±2-5% (higher for narrowband sources)
• Primary choice for: production monitoring, daily verification, spot checks

Best-in-class 2024: X-Rite i1Display Pro, Klein K-10A

Critical limitation: Colorimeters assume specific SPD shapes. LED/laser sources produce metamerism errors up to 5-10 ΔE. Spectroradiometer required for accurate measurement.

Measurement Protocol

Display calibration workflow (spectroradiometer required):

1. Warm-up: 30 minutes minimum for display stability
2. Instrument calibration: Use NIST-traceable calibration lamp
3. Measurement geometry: 0/45° or specular component included (SCI) for consistency
4. Spot size: 2-5° measurement field for uniformity
5. Integration time: Auto or appropriate for luminance level
6. Averaging: 3-5 readings per measurement point

Daily validation (colorimeter acceptable):

1. White point: Verify x,y within ±0.002 of target
2. Gamma: Verify at 5 points (R, G, B, neutral, 50% gray) within ±0.02
3. Luminance: Verify target ±5% (48 cd/m² for cinema, 100 cd/m² for SDR)
4. Uniformity: Measure 9-point grid, verify ±10% center-to-corner

Calibration & Validation Hierarchy

Level 1: Primary Standard (annual, NIST-traceable):

• Calibration lamp with known SPD
• Performed by manufacturer or accredited lab
• Cost: $500-1500 USD
• Traceability: NIST (USA), PTB (Germany), NPL (UK)

Level 2: Field Transfer Standard (monthly):

• Calibrated white reference (Spectralon PTFE, barium sulfate paint)
• Reflectance >98% across 380-780nm
• Lambertian scattering properties
• Cost: $200-500 USD

Level 3: Daily Validation (per shift):

• Known color patches (ColorChecker, X-Rite Digital ColorChecker SG)
• Verify repeatability ±0.5 ΔE2000
• Track drift over time
• Establish instrument baseline

Level 4: Drift Monitoring (continuous):

• Log measurements of stable reference
• Control chart analysis
• Predictive calibration scheduling
• Detect instrument degradation early

Color Temperature Measurement

Correlated Color Temperature (CCT) [temperature of blackbody radiator whose chromaticity is closest to light source chromaticity on u'v' diagram]:

McCamy's approximation (valid for CCT 2000-25000K): $$ CCT \approx 449n^3 + 3525n^2 + 6823.3n + 5520.33 $$

Where $n = \frac{x - 0.3320}{y - 0.1858}$

** Robertson's method** (iterative, more accurate):

1. Convert xy to uv
2. Calculate isotemperature lines
3. Find closest point on Planckian locus

Practical cinema application: Tungsten (3200K), Daylight (5600K), Xenon (6000K approx). CCT tolerance: ±150K for acceptable color matching.

White Balance & Tint Measurement

Chromaticity difference vector from target white point:

• Δx, Δy: Chromaticity shift in xy coordinates
• Δu'v': Uniform chromaticity shift
• Convert to tint/green adjustment in camera terms

Measurement procedure:

1. Measure target white (card, diffuse sphere, integrating sphere)
2. Calculate Δxy from desired white point (D65, DCI)
3. Convert to camera white balance parameters (Kelvin + tint offset)
4. Verify with skin tone reproduction

Tolerance: ±0.002 in x,y (≈3 ΔE at skin tones) for professional work.

Failure Modes

Observer Metamerism

Observer metamerism [individual variation in color matching functions causing colors that match for Standard Observer to mismatch for specific observers] arises from:

1. Lens absorption: Yellowing with age (cataracts) reduces short-wavelength transmission
2. Macular pigment: Xanthophyll density varies 0.1-0.8 optical density at 460nm
3. Cone photopigment polymorphisms: Genetic variants shift peak sensitivities 2-5nm
4. Pupil size: Affects effective CMF due to Stiles-Crawford effect

Practical consequences:

• Colors matching for Standard Observer may not match for specific individuals
• Color matching functions vary up to 3 ΔE between observers
• More pronounced in blue region (450-500nm)
• Critical for VFX color matching across viewing environments

Cinema-specific impacts:

• Dailies approval under different lighting conditions
• Final grade verification by multiple viewers
• Director/DP color perception differences
• Cross-facility color matching (different DI theaters)

Mitigation strategies:

1. Use CIE 1964 10° observer for cinema applications (large screens)
2. Establish viewing booth standards (SMPTE RP 431-2:2011)
3. Validate with multiple observers for critical decisions
4. Use reference displays with calibrated white points
5. Document individual observer preferences (if significant deviation)

Instrument Calibration Drift

Spectroradiometer drift causes:

• Wavelength accuracy shift: 0.1-0.5nm/year
• Photometric accuracy shift: 1-3%/year
• Spectral shape changes: detector aging, grating efficiency loss

Failure mode progression:

1. Months 0-6: Within specification (±1-2%)
2. Months 6-12: Drift detectable, may exceed spec (±2-5%)
3. Months 12-24: Significant drift, errors visible (±5-10%)
4. Months 24+: Calibrate immediately (±10%+)

Detection:

• Daily measurements of stable reference
• Control chart tracking (x, y, Y over time)
• Cross-check with secondary instrument
• Sudden changes = instrument failure (not drift)

Impact:

• Uncalibrated spectroradiometer: ±1-3 ΔE (months 0-12)
• Uncalibrated colorimeter: ±3-10 ΔE (worse for narrowband sources)
• Uncalibrated colorimeter on LED: ±5-15 ΔE (metamerism error)

Prevention:

• Annual recalibration by manufacturer
• Monthly verification with transfer standard
• Daily validation with known reference
• Maintain calibration certificate traceability

SPD Measurement Errors

Insufficient wavelength resolution:

• 5nm interval: Adequate for broadband sources, errors <1 ΔE
• 10nm interval: May miss narrow peaks, errors 1-3 ΔE
• 20nm interval: Unacceptable for LED/laser, errors >5 ΔE

Narrowband source errors:

• LED sources: FWHM 20-30nm (blue), 30-50nm (green/red)
• Laser sources: FWHM <5nm
• Colorimeter errors: 5-15 ΔE (wrong CMF assumption)
• Spectroradiometer: 1nm spacing required for accuracy

Stray light [unwanted light at wrong wavelengths reaching detector]:

• Caused by grating scatter, filter leakage
• More pronounced in UV/blue region
• Causes saturation errors, incorrect chromaticity
• High-end instruments: stray light <0.1% (acceptable)
• Low-end instruments: stray light 1-5% (problematic for LED)

Beam geometry errors:

• Specular vs diffuse reflection
• Measurement angle affects chromaticity (especially for glossy surfaces)
• Cinema standard: 0/45° geometry for reflective targets
• Displays: normal incidence (on-axis) measurement

Polarization sensitivity:

• Some instruments vary with polarization angle
• LCD displays highly polarized
• Error: 2-5 ΔE if polarization not accounted for
• Solution: average multiple polarization angles or use polarization-insensitive instrument

Metamerism Failure Modes

Illuminant metamerism [colors matching under one illuminant but not under another due to different SPDs]:

Example: Fabric swatches match under D65 but diverge under tungsten

• Root cause: Different reflectance spectra
• Common in textiles, paints, VFX elements
• CCT change of 1000K can cause 5-10 ΔE shifts

Detection:

• Spectral reflectance measurement (not just XYZ)
• Calculate metameric index (MI)
• Test under multiple illuminants (D65, A, F11)

Prevention:

• Use spectral data for critical elements
• VFX: Capture spectral textures if possible
• Costume design: Verify under all lighting conditions

Instrument metamerism [different instruments produce different XYZ for same sample]:

• Spectral sensitivities don't match CIE CMFs exactly
• Bandwidth differences (1nm vs 5nm vs 10nm)
• Illumination beam geometry variations

Cinema workflow:

1. Use spectroradiometer for primary display calibration
2. Use colorimeter for daily verification (cross-calibrated to spectro)
3. Establish correlation between instruments
4. Use same instrument for time-series measurements
5. Document instrument make/model for reproducibility

Mini Glossary

• SPD [Spectral Power Distribution]: Intensity of electromagnetic radiation at each wavelength across visible spectrum (380-780nm), measured in W/nm
• CMF [Color Matching Functions]: CIE-defined functions $\bar{x}(\lambda), \bar{y}(\lambda), \bar{z}(\lambda)$ quantifying standard observer response to each wavelength
• CIE XYZ: Device-independent tristimulus color space based on human vision physiology; X≈red-green, Y≈luminance, Z≈blue-yellow
• xy chromaticity: 2D projection of XYZ representing hue and saturation independent of luminance; $x = X/(X+Y+Z), y = Y/(X+Y+Z)$
• Dominant wavelength: Wavelength of spectrally pure light that matches a color when mixed with white; intersection of line from white through chromaticity to spectral locus
• Purity [Excitation Purity]: Saturation metric as percentage distance from white point to chromaticity vs white point to spectral locus
• Metamerism: Phenomenon where different SPDs produce identical tristimulus values under specific illuminant; root cause of many color matching failures
• White point: Chromaticity of reference white (D65, D50, DCI); defines neutral axis for color space
• Gamut: Volume of colors reproducible by device within XYZ color space; defined by primary chromaticities and white point
• CCT [Correlated Color Temperature]: Temperature of blackbody radiator whose chromaticity is closest to light source on u'v' diagram
• Luminance: Photometric quantity proportional to perceived brightness; measured in cd/m² (nits)
• Illuminance: Photometric incident light quantity; measured in lux (lx) or foot-candles (fc)
• u'v' chromaticity: Perceptually uniform 2D color space where equal distances represent equal perceived color differences
• CRI [Color Rendering Index]: Metric (0-100) evaluating how faithfully light source renders colors compared to reference illuminant
• TM-30: IES method for evaluating color rendering using 99 color samples and fidelity/gamut indices; supersedes CRI for many applications