Latest or all posts or last 15, 30, 90 or 180 days.
Welcome to
In-depth review coverage is by subscription.
Also by Lloyd: and
First-time visitor
Capacities up to 48TB and speeds up to 1527MB/s

Spectral Transmission of Digital Sensors

The human eye can see from about 390nm out to almost 800nm*.

I recently learned that the Sony A7R II has a 420nm bandpass filter coating on the sensor cover glass, which cuts off a lot of violet light. The abrupt spectral cutoff for violet light means that true violet-colored images cannot be captured at all and that images are inherently biased slightly yellow (because violet is removed). I have never liked Hoya filters for their slight yellowish cast, and yet Hoya HMC Super filters cut off at 400nm, which is much more generous than 420nm.

Presumably the bandpass filter is used to avoid some kind of imaging nastiness that occurs with very blue to ultraviolet light, but why 420nm, why not preserve the 400-420nm visible region? Modern multicoated lenses kill off all but a tiny fraction of ultraviolet, so that is not a valid reason.

It turns out that this 420nm bandpass cutoff is common with digital cameras. Courtesy of Kolari Vision, this spectral transmission chart made with a line scanning spectrophotometer shows the same 420nm cutoff for the Nikon D810. Kolari Vision has graphs for other digital camera sensors.

See also Are your sunglasses protecting your eyes?. Interesting sidenote— I’ve always liked yellow tinted sunglasses and it seems that avoiding violet light is wise:

In ophthalmology, high-energy visible light (HEV light) is high-frequency, high-energy light in the violet/blue band from 400 to 500 nm in the visible spectrum.[1] HEV light has been implicated as a cause of age-related macular degeneration.[2][3]

* That the human eye can see (weakly) out to 850nm can be shown be viewing an 850nm infrared flashlight straight-on. Or I should say that *my* eyes can do so. And I’m assuming that an 850nm IR flashlight is actually emitting 850nm light. But at the least something around 750-800nm seems likely for the human eye. UPDATE: some readers tell me that my 850nm flashlight is simply bleeding other frequencies. I can't rule that out but I will say this: I always get 100% on color discrimination tests (unusually good color vision, genetic from my father) and so who can tell me I cannot see 850nm? I just don’t know for sure.

Nikon D810 sensor stack spectral transmission

Michael B writes:

The blocking of blue/uv light has dramatic implications when taking pictures of blueish-purple or reddish-purple flowers. Many such flowers reflect strongly in this almost-UV-part of the spectrum (because many pollinating insects can perfectly see UV). Taking a picture of these flowers/colors often results in disappointment, because the purple colors turn out plain blue or red, drastically different from their "real" color. Very few cameras get it right (some Nikons do well, but not all). Some fabrics are the same, e.g. some purple satins.

Up to now I thought it was the sensor that somehow could not record these colors correctly. Now I'm learning that the camera manufacturers intentionally block that visible part of the spectrum with the cover glass. Now that just seems stupid. I'm shaking my head. Is there a reason for doing so?

DIGLLOYD: There must be a reason for the 420nm cutoff, but I don’t know the answer. Possible ideas like cutting off violet fringing arise—dumbing down for mediocre lenses? I don’t want to believe that one! In my view, the sensor should record the range of color that the human eye sees. Even if color gamut is a problem for pure violet around 400-420nm, that color contributes to a perceptual response when mixed with other frequencies.

Bruce D writes:

We had a sign on the door of my old laser lab - "Do not stare into the bright light with your remaining good eye".

I read your comment on "seeing" infrared light with some disquiet. Of course, if you can see it, per definition it's not infrared - but that's nit-picking. As you'll doubtless realise, the eye has very little sensitivity to NIR which means that even when exposed to high intensity, the pupil does not constrict and there is no turn-away or blink reflex. This limitation is frequently blamed for people suffering permanent retinal damage while observing the sun through crappy filters. You don't need optics for the damage to be done.

I doubt your torch is capable of delivering enough energy to hurt you but caution really is warranted.


UV is absorbed rapidly in the eye. Besides leading to DNA damage, it will do damage primarily in the cornea. (This is why excimer lasers are used for refractive surgery - all the energy is deposited in a very small volume.)

NIR passes through the cornea and is absorbed in the retina, so there's a second potential damage mechanism. I'd be inclined to call anything longer than about 720 nm NIR, but there's no sharp cut off. Of course, the definition in terms of human vision is totally arbitrary because scotopic vision is blind to anything shorter than about 500 nm.

Microwaves can lead to cataracts. I used to work in radar and we were regularly warned against staring down active waveguides to check their alignment.

As for intensity - the sun puts out about 1400 watts per sq metre, about 40% of which (~560 watts/sq metre) lies in the 0.75 - 2 micron region. This translates to half a milliwatt per square millimeter. This _is_ sufficient to cause thermal damage. You can compare this with your torch's output and the size of your dilated pupil.

DIGLLOYD: I use infrared to mean 680nm - 1100nm (after that is deep IR).

My math (see above) says that 560000 milliwatts (560 watts) over 10000 cm^2 (1 m^2)= 56 milliwats / cm^2. Of course observing the sun is a risky business and a lot of UV-cut filters let through a great deal of infrared.

I was rushed and meant to add that you really need to take the beamwidth of the light and the aperture of the eye into account when making this calculation.

So if you have a beam that's 0.5 m wide and your pupil is 5 mm in diameter, your eye will intercept (5/500)^2 = 1.e-4 of the emitted light .... or perhaps four times that because the beam intensity is not uniform.

In any case, if you get ~ 10 mW in your eye for a second, you can damage it. There's a brief article on laser safety here Laser safety - Wikipedia, the free encyclopedia.

Good advice in general: DO NO STARE AT ANY FLASHLIGHT OR LASER. Though my IR flashlight is not very powerful, I can't know how bright it really is. As a more practical matter there is IMO an issue with sunglasses that do not block IR—consider snow and beaches with dark sunglasses that not bloack IR. As for lasers I would never look at any laser, and I walk out of any idiot’s presentation using a laser pointer—I’m not taking that chance, particularly green lasers.

diglloyd Inc. | FTC Disclosure | PRIVACY POLICY | Trademarks | Terms of Use
Contact | About Lloyd Chambers | Consulting | Photo Tours
RSS Feeds | Twitter
Copyright © 2008-2017 diglloyd Inc, all rights reserved.