Appearance
To identify pigments with imaging, combine three non-invasive techniques: multispectral reflectance to capture each pigment's absorption curve, false-colour infrared to separate pigments that look identical in visible light, and luminescence imaging (UV fluorescence and visible-induced infrared luminescence) to catch diagnostic emissions. Imaging rarely gives a final answer alone, but it maps where each pigment sits and narrows candidates so dramatically that any follow-up XRF or Raman measurement goes straight to the right spot. Here is the step-by-step approach.
Why can't visible colour alone identify a pigment?
Two blues that look identical to the eye — say azurite and ultramarine — can reflect completely differently in the infrared, and a green could be a copper pigment or a mixture. Visible colour collapses the whole spectrum into three channels, hiding exactly the differences that distinguish materials. Imaging across UV to NIR recovers those differences, turning "it's blue" into "it absorbs here, reflects there, and luminesces in IR — therefore Egyptian blue."
What is the step-by-step imaging workflow?
Work from broad screening to specific tests:
- Capture a calibrated multispectral stack (UV–visible–NIR) with a reflectance target in frame.
- Build a false-colour IR image to flag pigments that differ in infrared.
- Run VIL (visible-induced luminescence) to map Egyptian blue and a few luminescent pigments.
- Capture UV fluorescence to catch organic reds, varnishes and retouching.
- Extract reflectance curves from regions of interest and compare to a reference library.
- Confirm ambiguous areas with XRF or Raman at the points imaging flagged.
How do I read false-colour infrared for pigments?
False-colour infrared maps the IR band to red, the red band to green, and the green band to blue. Because pigments reflect IR differently, materials that matched in visible light now diverge in hue:
| Pigment | Visible | False-colour IR appearance |
|---|---|---|
| Azurite | Blue | Stays blue/grey |
| Ultramarine | Blue | Shifts toward red/purple |
| Indigo | Blue | Often dark in IR |
| Vermilion | Red | Bright yellow-orange |
| Red lead (minium) | Orange-red | Distinct from vermilion |
These are screening clues, not proof — but they tell you which blues to test and which to leave.
How do I extract and compare a reflectance curve?
Sample a region of one pigment and plot its mean reflectance across bands, normalised against your target:
python
import numpy as np
from skimage import io
cube = np.stack([io.imread(f).astype(np.float32) for f in band_files], -1)
target = np.stack([io.imread(f).astype(np.float32) for f in target_files], -1)
# normalise each band to the white reflectance standard
white = target[ty0:ty1, tx0:tx1].reshape(-1, cube.shape[-1]).mean(0)
refl = cube / np.clip(white, 1e-3, None)
roi = refl[y0:y1, x0:x1].reshape(-1, cube.shape[-1])
curve = roi.mean(0) # reflectance vs wavelength
for nm, r in zip(wavelengths, curve):
print(f"{nm} nm: {r:.3f}")Compare the printed curve to published reference spectra (e.g. from a pigment checker chart you imaged, or a spectral library) to shortlist candidates.
When is luminescence imaging the deciding test?
VIL is close to definitive for one famous case: Egyptian blue glows bright in the near-infrared when excited by visible light, while every other blue stays dark — so a VIL image effectively maps Egyptian blue directly. UV fluorescence similarly flags organic lake reds, modern retouching and varnish layers that share a reflectance curve with something else. When reflectance is ambiguous, the emission image often breaks the tie.
What are the pitfalls that produce wrong IDs?
- No reference target — curves become uncomparable across sessions; never skip it.
- Mixtures read as a single pigment — a green may be blue + yellow; treat curves as blends.
- Binders and varnishes alter apparent reflectance; account for surface layers.
- Over-reliance on false colour — it is a screen, not a conclusion.
- Skipping confirmation — declaring a pigment from imaging alone where stakes are high.
How do I report a pigment identification responsibly?
State what imaging shows and what it does not. A defensible line reads: "VIL luminescence and the reflectance curve are consistent with Egyptian blue; XRF confirmed copper." Archive the calibrated stack, the reference target reading and the curves you extracted as paradata. Imaging's job is to be a transparent, reproducible screen that points confirmatory analysis to the right pixel.
Key Takeaways
- Imaging narrows pigment candidates non-invasively; pair with XRF/Raman for certainty.
- False-colour IR separates blues and reds that look identical in visible light.
- VIL luminescence effectively maps Egyptian blue on its own.
- Always image a reflectance target so curves are comparable and library-matchable.
- Treat mixtures, binders and varnishes as confounders, not single pigments.
- Report what imaging shows versus what confirmation proves, and archive the paradata.
Frequently Asked Questions
Can imaging alone identify a pigment with certainty?
Imaging narrows candidates and is often conclusive when signatures are distinctive (e.g. Egyptian blue's IR luminescence), but for confirmation it should be paired with a point technique such as XRF or Raman. Imaging guides where to take those measurements.
What is the most diagnostic imaging clue for a blue pigment?
Egyptian blue emits strong near-infrared luminescence under visible/red excitation, glowing bright in a VIL image while indigo, azurite and ultramarine stay dark. VIL is the standard test to map Egyptian blue.
What is false-colour infrared used for in pigment ID?
False-colour infrared (mapping IR, red and green into RGB) separates pigments that look identical in visible light, because they reflect differently in the infrared. It is a classic first-pass discriminator.
Why combine reflectance imaging with luminescence imaging?
Reflectance shows how a pigment absorbs across wavelengths, while luminescence (UV or IR) shows emission. Together they distinguish pigments that share a reflectance curve but differ in fluorescence, such as some reds.
Do I need a calibrated reference target?
Yes. Without a reflectance standard you cannot compare a pigment's spectral curve to reference libraries, and colour-based clues become unreliable across sessions and equipment.
Is imaging non-invasive?
Yes, all the imaging methods here are non-contact and non-destructive, which is why they are used to screen objects before deciding whether any sampling or point analysis is justified.