Appearance
UV imaging and multispectral imaging are not rivals — UV is one band inside a multispectral stack. A single UV capture (usually 365 nm fluorescence) is fast and often reveals retouching or iron-gall undertext, while multispectral adds the visible and near-infrared evidence needed to identify materials and produce reproducible, quantitative results. Most "which is better" arguments are really troubleshooting problems in disguise, so this guide walks through the disagreements you will actually hit and how to resolve them.
Why do the two methods show the text in different places?
This is the most common confusion, and the answer is physics, not error. UV fluorescence depends on materials emitting visible light when excited; NIR reflectance depends on how a pigment absorbs longer wavelengths. Carbon ink stays dark across the spectrum, but iron-gall ink can fluoresce-dark under UV and then fade in NIR. When UV and a NIR band disagree, you are seeing two different materials or two different states of the same ink. Record both — the disagreement is the diagnosis.
My UV photo is a purple blur — what is wrong?
Two faults usually combine here:
- Purple cast: visible light is leaking into a fluorescence shot. Put a UV/IR-cut (e.g. BG39 or equivalent) filter on the camera so it records only the emitted visible fluorescence, and shoot in a darkened room.
- Blur: most lenses focus UV at a slightly different plane than visible light. Refocus while viewing under UV, or stop down to f/8–f/11 to widen depth of field and mask the focus shift.
If you actually want reflected-UV (not fluorescence), you need a quartz/UV-transmitting lens; an ordinary lens blocks most UV before it reaches the sensor.
How do I decide which method a given job needs?
Match the method to the question:
| Need | Use | Why |
|---|---|---|
| Quick survey, spot repairs/retouching | UV fluorescence | One shot, fast, high contrast on additions |
| Read scraped undertext reliably | Full multispectral + PCA/ICA | UV alone rarely separates layers cleanly |
| Identify which pigment/ink | Multispectral reflectance | Material signatures span visible–NIR |
| Reproducible, citable evidence | Multispectral (calibrated) | Single UV shots are hard to normalise |
| Penetrate stains and foxing | NIR band (850–940 nm) | UV cannot see through surface grime |
Why is my UV fluorescence so dim I can barely expose it?
Fluorescence emits far less light than reflectance, so this is expected. Fixes, in order of preference: lengthen exposure (fluorescence is static, so a multi-second exposure is fine on a stable rig); increase UV source power within safe limits; open the aperture only if depth of field allows; and confirm your blocking filter is not over-attenuating the emission band. Raising ISO is the last resort because it amplifies noise that swamps faint signal.
The multispectral stack and the UV shot won't align — how do I fix it?
Misregistration between a separately captured UV frame and the rest of the stack causes ghosting when you composite them. Capture the UV band in the same rig, same geometry, same session as the other bands, then register everything to one reference band with sub-pixel cross-correlation. Never composite a UV image shot on a different day or camera with a multispectral stack — the geometry will never match.
Are there safety differences between the two?
Both involve UV, so both demand care, but a quick UV survey accumulates less exposure than a full multi-minute stack with bracketing. Use filtered 365 nm UV-A LEDs, keep total exposure minimal, log dosage in your paradata, and never bring unfiltered UV-C germicidal lamps near originals. The conservation cost of imaging must always be weighed against the scholarly gain.
Key Takeaways
- UV is one band of multispectral, not a competing technique — multispectral subsumes it.
- Purple, blurry UV shots come from missing UV/IR-cut filters and UV focus shift; fix both.
- When UV and NIR disagree about the text, that disagreement identifies the material.
- Use lone UV for quick surveys and retouching; use the full stack for reliable, citable recovery.
- Capture UV in the same rig and session, then register to the stack to avoid ghosting.
- Filtered 365 nm UV-A only; minimise dose and log it for conservation accountability.
Frequently Asked Questions
What is the difference between UV imaging and multispectral imaging?
UV imaging is a single capture under ultraviolet light (often 365 nm fluorescence). Multispectral imaging captures many bands from UV to NIR, of which UV is just one — so multispectral subsumes UV but adds reflectance and infrared evidence.
When is plain UV enough?
UV fluorescence is enough for quick surveys, spotting retouching, repairs and some iron-gall undertext. If UV alone resolves your text and you don't need quantitative or reproducible results, a full stack may be overkill.
Why does my UV image look purple and blurry?
That purple haze is visible light leaking through the lens because you lack a UV/IR-cut filter on a fluorescence shot. Blur usually means the lens focus shifts in UV; refocus under UV or stop down.
Is UV fluorescence the same as UV reflectance?
No. Reflected-UV imaging records UV bouncing off the surface and needs a quartz-transmitting lens. UV fluorescence records visible light emitted when UV excites the material and uses a normal lens with a UV-blocking filter on the camera.
Why do multispectral and UV disagree about where the text is?
They probe different physics. A stroke can fluoresce under UV yet vanish in NIR, or vice versa. Disagreement is information: it tells you which material you are seeing, not that one method is wrong.
Can UV light damage the manuscript?
Prolonged UV-A and especially UV-B/UV-C can photo-degrade media and parchment. Use filtered 365 nm UV-A LEDs, minimise exposure, and never use unfiltered germicidal lamps near collection material.