RGB, CMYK, Lab, XYZ - a Color Space is your Map in the Land of Color - make sure you're using the correct one!

RGB, Hex, and CMYK are the color spaces most professionals meet first — and, for physical color work, the least useful ones. Here's a practical map of the color spaces that actually run color measurement, quality control, and color matching across industries, and when each one belongs in your workflow.

Ask a room full of professionals — a colorist, a QC manager, a marketing director, a plant technician — to define "color space," and you'll get five different answers, most of them shaped by whichever tool they use every day. That's the problem. A color space is simply a coordinate system for describing color numerically, but different color spaces were built to solve completely different problems: rendering light on a screen, laying down ink on a press, or objectively describing the color of a physical object so it can be measured, controlled, and matched. Mixing these up — say, trying to use a Hex code to set a tolerance for dyed fabric — is one of the most common, and most expensive, mistakes in color-critical industries.

This guide walks through the color spaces you're most likely to run into, grouped by what they're actually good for: the familiar digital and print spaces, the workhorse spaces used for physical color measurement, the legacy spaces still hiding in older instruments and standards, and the raw spectral data that underlies all of them.

RGB, Hex, and CMYK: built for screens and the press, not for physical color

These three are the color spaces almost everyone already knows, because they're the ones baked into every design tool, every monitor, and every printer.

• RGB is an additive color model: red, green, and blue light are combined at different intensities to produce color on a self-luminous display. It describes what a screen emits, not what a physical object reflects.
• Hex is not a separate color space — it's just an RGB value re-written as a six-digit hexadecimal code (#FF6B35, for example) for convenience in web and design software. It inherits every limitation RGB has.
• CMYK is a subtractive model built for four-color (cyan, magenta, yellow, black) offset and digital printing. It expresses color as percentages of ink, and those percentages only mean something in combination with a specific press, ink set, and substrate.

The problem all three share: they are device-dependent. The same RGB value renders differently on two different monitors; the same CMYK recipe prints differently on two different presses, or even on the same press with a different paper stock. None of them describe a physical color in absolute, repeatable terms — which is exactly what you need when you're grading a dyed textile, approving an injection-molded part, or setting a tolerance for a painted panel. For that, the industry has converged on a different family of color spaces entirely.

For a closer look at why this gap causes real production problems, see why some colors are harder to print accurately.

CIE L*a*b* and CIE L*C*h: the color spaces of choice for physical color

When the job is to measure, control, and match the color of a real, physical thing — yarn, paint, plastic, printed packaging, cosmetics, building materials — CIE L*a*b* (and its polar companion, CIE L*C*h) is the standard used across virtually every color-critical industry. It's device-independent, built around a model of human color vision, and precise enough for tight tolerances while still being intuitive enough for a non-specialist to read.

CIE L*a*b* describes any color as a point in three-dimensional space:

• L* — Lightness, from black (0) to white (100)
• a* — position on the green–red axis
• b* — position on the blue–yellow axis

CIE L*C*h uses the exact same underlying space, just expressed in polar rather than Cartesian coordinates — L* stays the same, while a* and b* are replaced by:

• C* — Chroma, the saturation or intensity of the color, moving outward from the neutral axis
• h° — Hue angle, the position around the color wheel, from 0° to 360°

This is also the space that Delta E (ΔE) color difference calculations are built on: ΔE is essentially the distance between two points in L*a*b* space, and the L*C*h breakdown of that same difference tells you whether a failing sample is off on lightness, chroma, or hue — which is exactly what a production team needs to diagnose and fix a color problem.

A number is only meaningful with its measurement conditions

Here's the part that trips up a lot of teams: an L*a*b* or L*C*h reading is not a fixed, universal fact about an object the way its weight or dimensions are. The same physical sample can produce meaningfully different readings depending on how it was measured. Any time a Lab or LCh value is recorded, reported, or written into a spec sheet, the following should be specified alongside it:

Change any one of these and the resulting numbers can shift enough to turn a "pass" into a "fail" on the exact same physical sample. This is also why visual, unaided-eye color decisions are so unreliable in comparison — see visual color inspection vs. instrumental measurement for what that gap actually costs. Reputable references on how these conditions are defined include the CIE's official Colorimetry publication, the background on standard illuminants, and this plain-language explainer on measurement geometry from X-Rite.

Every benchtop and portable spectrophotometer in the Ametra range reports these conditions alongside every reading, which is what makes a measurement taken on one line comparable to one taken at a supplier's facility on the other side of the world.

CIE XYZ and Yxy: the legacy foundation everything else is built on

Before L*a*b* existed, there was CIE XYZ, published in 1931. It's the original tristimulus system — three values (X, Y, Z) representing amounts of three imaginary primaries, derived mathematically from the standard observer's color-matching functions. Every other CIE color space, including L*a*b* and L*C*h, is calculated from XYZ.

Yxy (also written xyY) is a related, derived space: Y is luminance (carried over directly from XYZ), while x and y are chromaticity coordinates plotted on the familiar horseshoe-shaped chromaticity diagram used throughout the display and lighting industries.

Neither is intuitive to work with directly — looking at a set of X, Y, Z or Yxy values doesn't give most people a mental picture of the actual color the way an L*a*b* or L*C*h reading does, since neither space is built to match human perception in a uniform way. In practice, XYZ and Yxy today mostly show up as calculation steps behind the scenes, or in specific legacy contexts: certain color indices — like some formulations of yellowness and whiteness indices used in plastics, paper, and textiles — are still defined directly from tristimulus values, and display and lighting engineers still specify white points and gamuts in Yxy chromaticity terms. Background on the underlying math is available from the CIE 1931 color space reference.

Exotic and niche spaces: Hunter Lab, CIE L*u*v*

A handful of other color spaces persist in specific corners of industry, mostly for historical reasons.

Hunter Lab predates CIELAB by nearly three decades — it was developed in 1948 by Richard Hunter, using L, a, b coordinates (without the asterisks) derived through a different, older mathematical transform than the CIE 1976 formula. It's still embedded in some legacy instrument software and in a number of older US industry and agricultural standards, particularly in food, agriculture, and some plastics specifications written before CIELAB became dominant. Most modern instruments can report both, but for new specifications, CIELAB is the broadly recognized standard.

CIE L*u*v*, published the same year as CIELAB (1976), is a companion uniform color space built on a different chromaticity foundation (u′, v′). It sees more use for self-luminous color — LEDs, displays, automotive lighting — than for the reflective, surface-color work that dominates textiles, coatings, and packaging, where L*a*b*/L*C*h remain the default.

Unless a legacy specification or a specific industry standard explicitly calls for one of these, CIELAB and CIE L*C*h remain the safer, more broadly understood choice for new color programs.

Spectral data: the fingerprint behind every color number

Every color space discussed so far — Lab, LCh, XYZ, Yxy, Hunter Lab, Luv — is ultimately calculated from one underlying source: the spectral reflectance (or transmittance) curve of the sample, typically captured across the visible spectrum at regular wavelength intervals. That curve, combined mathematically with a chosen illuminant and observer, is what produces every tristimulus and Lab value discussed above.

A single spectral curve is illuminant- and observer-independent by nature — it's simply a physical property of the sample. Once you calculate a Lab value from it, you've committed to one specific illuminant and observer combination, and you've discarded the rest of the spectral detail in exchange for a compact, intuitive, three-number result that's more than sufficient for the vast majority of color work: measurement, evaluation, comparison, and pass/fail matching against a tolerance.

Full spectral data earns its keep in a smaller set of specialized tasks where a single Lab snapshot isn't enough:

• Paint, ink, and dye formulation and recipe prediction, where color-matching software needs the full reflectance curve of pigments and colorants to predict how a new mixture will behave — a calculation Lab values alone cannot support.
• Metamerism evaluation, where two samples need to be checked for a color match across multiple illuminants at once, not just the one illuminant a single Lab reading was calculated under. See what metamerism is and how it affects color perception for more on why this matters.
• Recalculating color under a different illuminant without physically remeasuring the sample, since the full spectral curve can be run through any illuminant/observer combination after the fact.

This is also the practical line between a colorimeter and a true spectrophotometer: a colorimeter captures a small number of filtered light readings and outputs tristimulus values directly, while a spectrophotometer captures the full spectral curve, computes Lab and other spaces from it, and retains the underlying data for the tasks above. For a fuller breakdown of what separates these instrument types, see spectrophotometers, colorimeters, and spectrocolorimeters — what's the difference. Browse the mobile colorimeters and full spectral spectrophotometer range at Ametra to see which level of data your process actually needs.

Quick reference

The bottom line

RGB, Hex, and CMYK do their jobs well — for screens and for print — but they were never built to describe the color of a physical object in a way that travels reliably between a factory floor, a supplier, and a customer. For that, CIE L*a*b* and its polar companion CIE L*C*h are the working standard across almost every color-critical industry: objective, device-independent, and intuitive enough for a whole team to use, provided the measurement conditions are always specified alongside the numbers. XYZ, Yxy, Hunter Lab, and CIE L*u*v* still have their place, mostly in legacy specs and specific technical calculations, but they're rarely where a modern color program should start. And behind all of them sits spectral data — the full, unabridged fingerprint of a color, indispensable for formulation and metamerism work, but more detail than most day-to-day measurement, evaluation, and matching actually requires.

Not sure which color space — or which instrument — fits your process? Browse the full Ametra collection or get in touch with our team.