Spectrophotometers, Colorimeters, and Spectroradiometers
A monitor profile, a paint match, a brand color check, a print proof - every one depends on a sensor putting numbers on light. Three families of instrument cover the territory: colorimeters for fast displays, spectrophotometers for reflective samples and printers, and spectroradiometers for self-luminous reference work. Knowing which is which - and how their geometries differ - is the entry ticket to trustworthy color measurement.
Advanced instrument bench
Instrument choice is not just colorimeter versus spectrophotometer. Accuracy is shaped by the sample type, source spectrum, geometry, aperture, wavelength step, integration time, UV content, gloss, repeat reads, and calibration age. This bench lets you stress-test all of those variables before the article unpacks the individual instrument families.
Compare spectral accuracy, geometry fit, repeatability, and traceability
Choose an instrument family and measurement job, then tune the practical variables that decide whether the readings can be trusted. The canvas redraws spectra, detector bands, geometry, repeatability scatter, and a decision map.
What measurement instruments do
A color measurement instrument converts light into numbers. Two families of quantity dominate. Colorimetric quantities - XYZ tristimulus, Lab coordinates, ΔE - describe what a standard observer perceives. Spectral quantities - power per wavelength - describe the physical light itself. From a spectrum you can compute any colorimetric value; from a colorimetric triple you can't recover the spectrum.
Different instruments produce different families. Colorimeters report colorimetric values directly via filter-and-detector arrays designed to approximate the CIE color matching functions. Spectrophotometers and spectroradiometers measure full spectra, then derive colorimetric values as a downstream computation. Spectral measurement is more accurate and more flexible; colorimetric is faster and cheaper.
The three families
Three named instrument classes show up across catalogs. The boundaries between them have softened over the years - high-end colorimeters now ship with internal spectral correction, and many spectrophotometers can measure displays - but the core distinctions still hold.
Colorimeter
Three filtered photodiodes, each tuned to approximate one of the CIE matching functions (x̄, ȳ, z̄). Outputs XYZ directly. Fast, cheap, ideal for repetitive display calibration.
- Display calibration
- Quick QA checks
- Production line monitoring
Spectrophotometer
Measures the actual spectral reflectance or transmittance of a sample across the visible band (typically 31-40 ten-nanometer steps). Built-in light source. Derives any colorimetric value from the spectrum.
- Paint and dye matching
- Printer / press characterization
- Spot color (Pantone) measurement
Spectroradiometer
Spectral measurement of emissive sources - displays, lamps, sky. Higher resolution (1-5 nm bandwidth), tighter wavelength accuracy, NIST traceable. Used for reference work and calibrating other instruments.
- Display reference measurement
- Light source characterization
- Colorimeter correction matrices
Pick the right family for your job
Describe the work. The selector recommends the family and reasoning. There are no hard rules - speed, budget, and accuracy trade off against each other - but the recommendations follow the standard advice in each domain.
- Recommended family
- Colorimeter
- Typical instruments
- Calibrite Display Pro HL, X-Rite i1 Display Pro
- Reasoning
- Colorimeters are fast, cheap, and accurate enough for display work. Spectro is overkill unless you need a correction matrix for narrow-band primaries.
Filter-based vs spectral measurement
Colorimeters integrate light through fixed analog filters: each filter's transmission curve is meant to approximate one of the CIE color matching functions, so the photodiode behind it produces a value proportional to X, Y, or Z. The whole measurement happens in three numbers. Fast and cheap, but the filters are approximations - and the approximation error depends on the spectrum of the source.
Spectral instruments split incoming light into many narrow wavelength bands using a grating or prism, then read each band with a photodiode array. The output is the full SPD. Computing XYZ is a downstream integration step. Slower, more expensive, but accuracy is largely independent of source spectrum.
Measurement geometries
For reflective measurement, the geometric relationship between illuminator and detector is part of the measurement specification. Different geometries produce different readings on the same sample - especially for glossy, textured, or metallic finishes. The CIE publishes standard geometries to keep results comparable.
Five standard measurement geometries
Each geometry specifies the angles of illumination and detection relative to the sample surface. The same paint reads differently between geometries because gloss is treated differently.
SCI vs SCE
In d/8° geometry, a small specular trap inside the sphere can either include or exclude the specularly reflected light. Specular Component Included (SCI) leaves the trap open and measures total reflectance including any mirror-like gloss reflection. Specular Component Excluded (SCE) closes the trap and measures only the diffuse body color.
Calibration and traceability
A measurement is only as good as the instrument's calibration. Two layers of discipline matter: built-in white calibration (the user runs at startup), and traceable certification (the vendor maintains against a national standard).
Repeatability vs accuracy
Two different statistics describe instrument quality. Repeatability measures whether the same instrument produces the same reading on the same sample - the precision of one device. Accuracy measures whether that reading matches the true value - how close to the reference standard.
Same sample, ten readings - how tight is the cluster?
Choose an instrument class. The system simulates ten measurements of a known reference (true ΔE = 0). The plot shows individual readings and the repeatability statistics: standard deviation, max excursion, and a verdict against the instrument's published spec.
- Mean reading
- 0.12 ΔE
- Standard deviation (σ)
- 0.08 ΔE
- Max excursion
- 0.25 ΔE
- Repeatability spec
- < 0.15 ΔE
- Accuracy class
- < 0.5 ΔE inter-instrument
Instrument categories by use case
Most decisions about which instrument to buy come down to combining "what kind of sample" with "what kind of accuracy" with "what kind of budget." The table below shows the typical answer for common scenarios.
| Use case | Family | Typical instrument | Price range |
|---|---|---|---|
| Display calibration (LCD) | Colorimeter | Calibrite Display Pro HL, i1 Display Pro | $200-350 |
| Display calibration (OLED/wide-gamut) | Spectrophotometer | X-Rite i1Pro 3 Plus, Calibrite ColorChecker Studio | $1500-2500 |
| Printer profile creation | Spectrophotometer | i1Pro 3, Barbieri Spectro Swing | $1500-5000 |
| Paint / dye matching | Spectrophotometer (sphere) | X-Rite Ci64, BYK spectro-guide | $5000-15000 |
| Plastic / textile production QA | Spectrophotometer (sphere SCI/SCE) | Datacolor 800, Konica Minolta CM-25cG | $8000-20000 |
| Light source characterization | Spectroradiometer | Konica Minolta CS-2000A, JETI Specbos | $15000-50000 |
| Brand color / Pantone audit | Spectrophotometer | i1Pro 3, X-Rite eXact | $2500-7000 |
| Automotive metallic paint | Multi-angle spectrophotometer | BYK-mac i, X-Rite MA-T6 | $15000-40000 |
UV and fluorescent considerations
Many modern materials contain optical brightening agents (OBAs) that absorb ultraviolet light and re-emit it as visible blue. Printer paper, textile detergent residues, and white plastics frequently include them. They make whites look whiter, but they cause measurement headaches.
A spectrophotometer's measured spectrum depends on how much UV its built-in illuminator emits. Different instruments emit different amounts; same sample reads differently. M-series modes (M0, M1, M2, M3) standardize this.
Spectral workflow advantages
When the instrument outputs spectral reflectance, the workflow gets several things for free that filter-based colorimeters cannot offer.
Common measurement errors
Skipping white calibration
Forgetting to re-zero the instrument at the start of a session. Drift builds up; readings shift by 0.5-2 ΔE.
Wrong geometry for sample
Using d/8 on a glossy paint and d/8 on a textured fabric produces very different results - one is right for each.
SCI / SCE confusion
Comparing SCI from one vendor to SCE from another. Looks like a color difference; actually a geometry difference.
M0 vs M1 mismatch
Measuring papers with OBAs in M0 vs M1 produces 2-5 ΔE differences - not a real color shift but a measurement convention shift.
Edge effects on small samples
Measuring a sample smaller than the instrument's aperture; light from outside the sample contaminates the reading.
Sample heating
Some materials change color when warmed by the instrument's illuminator - thermochromic samples or temperature-sensitive dyes.
Ambient stray light
For displays: ambient room light reflecting off the screen contaminates the colorimeter reading. Especially at black/dark patches.
Out-of-date correction matrix
Using an old colorimeter correction matrix on a new display whose primaries are different. The matrix corrects for the wrong spectrum.
Test your understanding
Six questions on instrument families, geometries, and modes. Wrong answers come with brief explanations.
Quick check
Continue your journey
What Color Is and How Humans See It
The cornerstone connecting light, surface, eye, brain, and standards.
Physics · 04Spectral Power Distributions and Why RGB Is Not Enough
The spectra that spectroradiometers and spectrophotometers actually measure.
Colorimetry · 08CIE XYZ Explained
The tristimulus output every colorimeter approximates.
Colorimetry · 11Metamerism Explained
The phenomenon spectral measurement is uniquely equipped to diagnose.
Digital · 15ICC Profiles and How Color Management Actually Works
The format these measurements feed into.
Measurement · 17How to Calibrate and Profile a Display
The procedure that uses these instruments to build profiles.