CIECAM97s Color Appearance Model Lab | Bradford CAT, Hunt Compression

Primary colour

Comparison colour

sRGB input, adapted via Bradford CAT to D65. X swaps A ↔ B.

Viewing conditions

Degree of adaptation D — 1.00

Adapting luminance L_A — 64 cd/m²

Background luminance Y_b — 20%

CIECAM97s uses the same surround parameters as CIECAM02. The F_LL flag adds √n to z for dim/dark surrounds.

Shortcuts

X — swap A ↔ B · R — reset defaults · C — copy JSON · 1 average · 2 dim · 3 dark · 4 D=1 · 5 D=0

Colours

#D3AF37 Primary

ΔE₉₇

—

—

                    FL = —
                  

#4A90D9 Comparison

CIEDE2000: — ΔE₇₆: —

Opponent Chromaticity — a,b Plane

Filled dot = Primary A. Open circle = Comparison B. Unique-hue rays at h_i: R=20.14°, Y=90°, G=164.25°, B=237.53°. Dashed connector = A→B shift.

Response Compression — CIECAM97s vs CIECAM02

Gold = Hunt power 0.73: 40x^0.73/(x^0.73+2)+1 · Blue dashed = CIECAM02 sigmoid 0.42: 400(FL·|R|/100)^0.42/(…+27.13)+0.1 (scaled ÷10). Vertical bar = current F_L.

Hunt Effect — F_L vs L_A

F_L = 0.2k&sup4;(5L_A) + 0.1(1−k&sup4;)²(5L_A)^⅓ where k=1/(5L_A+1). Red marker = current L_A.

ΔE Multi-Metric Bar Chart

CIE 1931 Chromaticity Diagram

CIECAM97s Appearance Correlates

Colour	J	C₉₇	h°	M₉₇	s₉₇	Q₉₇
Primary	—	—	—	—	—	—
Comparison	—	—	—	—	—	—

J=lightness · C₉₇=chroma · h°=hue angle · M₉₇=colourfulness · s₉₇=saturation · Q₉₇=brightness. Hunt response f(x)=40x^0.73/(x^0.73+2)+1.

Hue Composition H (Quadrature)

Colour	h°	H	Composition
Primary	—	—	—
Comparison	—	—	—

H = hue quadrature (0–400). Unique hues: R=0, Y=100, G=200, B=300. Composition = perceptual mix of nearest unique hues.

CIECAM02 Cross-Reference (CAT02 · Sigmoid 0.42)

Colour	J	C	h°	M	s	Q
Primary	—	—	—	—	—	—
Comparison	—	—	—	—	—	—

CIECAM02: CAT02 adaptation, Michaelis-Menten sigmoid (exponent 0.42), z=1.48+√(50n). Compare vs CIECAM97s above to see how the model revision affects each correlate.

CIELAB D65 Reference

Colour	L*	a*	b*	C*	h°
Primary	—	—	—	—	—
Comparison	—	—	—	—	—

Export & Share

Copy or export current colour data. JSON includes both CIECAM97s and CIECAM02 correlates. Share URL encodes both colours, D, L_A, Y_b, and surround.

Multi-Surround Comparison

Compare ΔE₉₇, CIEDE2000, and ΔE₇₆ across average, dim, and dark surrounds for the current colour pair. Also shows J and M₉₇ per surround to visualise how lightness and colourfulness shift.

Surround	ΔE₉₇	CIEDE2000	ΔE₇₆	J_A	J_B	M_97A	M_97B

CIECAM97s Standards Overview

▶ Historical Context — CIE TC 1-34 (1997)

CIECAM97s (CIE 131:1998) was the first comprehensive colour appearance model adopted by the CIE as a recommendation. Developed by CIE Technical Committee 1-34 under the chairmanship of Mark Fairchild, it unified decades of research by Hunt, Nayatani, and others into a single forward model predicting lightness, chroma, hue, colourfulness, saturation, and brightness under any viewing condition.

The “s” suffix stands for “simple” — a simplified version of the full Hunt model suitable for practical applications. It was superseded in 2002 by CIECAM02, which replaced Bradford with CAT02 and the Hunt power-function with a Michaelis-Menten sigmoid.

▶ Bradford Chromatic Adaptation Transform

Bradford CAT (Lam 1985, refined by Hunt & Pointer 1985) uses a spectrally sharpened cone space. The 3×3 sharpening matrix M_Bradford was derived from corresponding-colour experiments at the University of Bradford. It includes a mild blue-boost row to improve prediction of blues under illuminant change.

In CIECAM97s: XYZ → M_Bradford → von Kries scaling (D) → M_Bradford⁻¹ → XYZ_adapted → M_HPE → LMS for appearance computation.

▶ Hunt Power-Function Response Compression

CIECAM97s compresses adapted cone signals using Hunt’s power function: f(x) = 40·x^0.73 / (x^0.73 + 2) + 1. The exponent 0.73 was derived from psychophysical brightness-matching data. This function saturates more steeply than the CIECAM02 Michaelis-Menten sigmoid (exponent 0.42), producing stronger compression of high-luminance stimuli.

The +1 additive term provides a non-zero baseline response, modelling neural noise at absolute threshold. Compare both curves in the Response Compression canvas.

▶ Hue Quadrature H & Unique Hues

Hue composition H maps the polar hue angle h (0–360°) to a perceptually spaced quadrature scale (0–400). Unique hues are anchored at: Red h_i=20.14° (H=0), Yellow h_i=90° (H=100), Green h_i=164.25° (H=200), Blue h_i=237.53° (H=300).

Between any two unique hues, H interpolates linearly weighted by the eccentricity factor e_s. The composition label (e.g. “65Y 35G”) gives the perceptual percentage of each unique constituent — directly useful for colour naming and specification.

▶ CIECAM97s vs CIECAM02 — Key Differences

	CIECAM97s	CIECAM02
CAT	Bradford	CAT02
Compression	Hunt 0.73	Sigmoid 0.42
z (eccentricity)	1 + F_LL√n	1.48 + √(50n)
Baseline response	+1	+0.1
Status	Superseded (historical)	CIE 159:2004

▶ Known Issues & Why CIECAM02 Replaced It

Chromatic adaptation prediction: Bradford CAT produces larger residuals under tungsten→daylight shifts than CAT02.
Response compression: The Hunt exponent 0.73 over-compresses saturated blues and under-predicts near-neutral lightness differences.
Non-invertibility edge cases: Some combinations of negative adapted cone signals cause undefined behaviour.
F_LL flag: The dim/dark surround z-adjustment is a discrete switch rather than a smooth function.

▶ Applications & Legacy

ICC v4 (2001): Early colour management systems implemented CIECAM97s before CIECAM02 was available.
Automotive paint matching: Bradford CAT adaptation was standard in metamerism indices.
Textile colour specification: Hue composition H provided intuitive colour naming for dye formulation.
Research: Understanding CIECAM97s is essential for reading pre-2002 colour science literature.

Mathematical Models and Formulas

Bradford Chromatic Adaptation Transform

M_Bradford = [[ 0.8951, 0.2664, -0.1614],
[-0.7502, 1.7135, 0.0367],
[ 0.0389, -0.0685, 1.0296]]

Adapted cone responses (von Kries):
R_c = (D·Y_w/R_w + 1−D) · R
G_c = (D·Y_w/G_w + 1−D) · G
B_c = (D·Y_w/B_w + 1−D) · B

XYZ_adapted = M_Bradford⁻¹ · [R_c, G_c, B_c]

Hue Quadrature H

Unique-hue data:

Hue	h_i	e_s	H_i
Red	20.14°	0.8	0
Yellow	90.00°	0.7	100
Green	164.25°	1.0	200
Blue	237.53°	1.2	300

H = H_i + 100·[(h−h_i)/e_i] / [(h−h_i)/e_i + (h_i+1−h)/e_i+1]

Research, Standards and Citations

CIECAM97s & CIE Publications

[1] CIE 131:1998. “CIE TC 1-34 Final Report: CIECAM97s — A Colour Appearance Model for Colour Management Systems.”

[2] Fairchild, M.D. (1998). Color Appearance Models. Addison-Wesley. Chapter 12: CIECAM97s.

[3] Hunt, R.W.G. (1994). “An improved predictor of colourfulness in a model of colour vision.” Color Research & Application, 19(1), 23–26.

Bradford CAT

[4] Lam, K.M. (1985). “Metamerism and colour constancy.” PhD thesis, University of Bradford.

[5] Süsstrunk, S., Buckley, R., & Swen, S. (1999). “Standard RGB color spaces.” IS&T/SID CIC7, 127–134.

CIECAM02 & Successor Models

[6] CIE 159:2004. “A Colour Appearance Model for Colour Management Systems: CIECAM02.”

[7] Li, C. et al. (2017). “Comprehensive color solutions: CAM16, CAT16.” Color Research & Application, 42(6), 703–718.

[8] Sharma, G., Wu, W., Dalal, E.N. (2005). “The CIEDE2000 color-difference formula.” Color Research & Application, 30(1), 21–30.

Hue Quadrature & Colour Naming

[9] CIE 15:2004. “Colorimetry.” 3rd edition.

[10] Luo, M.R., Cui, G. & Li, C. (2006). “Uniform colour spaces based on CIECAM02.” Color Research & Application, 31(4), 320–330.

About this Tool

[11] Auric Artisan Color Science Platform. CIECAM97s Historical Model Laboratory. All maths from modular engine. Zero network calls — all computation runs on-device using IEEE 754 double-precision floating point.

Research Backend

On-Device

Batch Pairwise ΔE Analysis

Enter multiple hex colours (comma or newline separated). The engine computes pairwise ΔE₉₇, CIEDE2000, and ΔE₇₆ for every combination using current viewing conditions and D. Minimum 2 colours.

Enter at least 2 valid hex colours to compute pairwise ΔE analysis.

ΔE₉₇ Distribution Histogram

Bradford vs CAT02 Adaptation Analysis

The CIECAM97s model uses Bradford for chromatic adaptation while CIECAM02 uses CAT02. Both are von Kries diagonal transforms but differ in their sharpening matrix. For D65 sRGB colours these produce nearly identical adapted XYZ values, but under illuminant A (2856 K) at high D, Bradford predicts slightly different blue and red shifts. Use the CIECAM02 cross-reference table in the Lab tab to quantify this difference for any colour pair.

Response Compression Exploration

The Response Compression canvas overlays the Hunt power function (exponent 0.73) with the CIECAM02 Michaelis-Menten sigmoid (exponent 0.42). At low stimulus levels both curves track closely, but above x≈1 the Hunt function saturates more steeply. This produces higher adapted responses for moderate stimuli in CIECAM97s, which cascades through opponent channels into different chroma and colourfulness predictions.

Hue Quadrature Accuracy

Hue composition H is identical between CIECAM97s and CIECAM02 (same unique-hue anchors and eccentricity factors). However, because Bradford and CAT02 produce different adapted cone signals, the polar hue angle h can differ by several degrees for the same input colour. This means H values from the two models are comparable but not identical — the research-grade comparison in the Lab tab quantifies this shift.

Research backend provides: batch pairwise ΔE₉₇ analysis, multi-surround comparison with J and M₉₇ per surround, distribution histograms, CIECAM02 cross-reference side-by-side comparison, response compression visualisation, and export to JSON/CSV. All computation runs on-device — zero network calls, zero telemetry.