9 replies to this topic

[Andrea B.]

I was recently complaining about how UV torchlight and sunlight don't mix well. So I thought I would show you this example. There's nothing earthshaking or hugely important here. Just an illustration of one of the problems that can be encountered in UV photography if using the 'wrong' equipment.

My UV torch for this experiment is fitted with a high-power Nichia 365nm UV-Led. I typically use this torch for illuminating a subject in order to focus Live View through a UV-Pass filter. For this task a UV torch is indispensible. Mine is a well-made beauty - the UV-Haiku by Don MacLeish.

Equipment used: Nikon D600-broadband + 105/4.5 UV-Nikkor

Here is my test subject: Lewisia longipetala 'Little Plum'.
(Or so it was marked. It does not look particularly plum coloured to me.)
Visible: with Baader UVIR-Block filter in sunlight


Ultraviolet: with Baader-U UV-Pass filter in sunlight.
First the original (cropped) version, then the white-balanced version.
You can see that the flower has a lot of UV-iridescence because of surface conical cells.



Ultraviolet: with Baader-U UV-Pass filter in sunlight and torchlight.
I was holding the torch at quite some distance from the flower
in order to fully cover the flower with the beam.
However, the shot appears blotch-ily bi-coloured both before and after white balancing.
FIrst the original (cropped) version, then the white-balanced version.


This last photo shows how I tried to desaturate the yellow false colour in the preceding white-balanced version.
This almost gives a monochrome look, but not quite. Just a reminder -> a monochrome look is expected
when shooting with such a narrowband UV-Led light when little or no sunlight is mixed in.


Here is one more comparison. In this case, the UV torch & sunlight combo turned out better.
Although not particularly monochrome, so go figure.
First the sunlight-only shot, then the mixed light shot. (Both were white-balanced.)



If I desaturate the yellow false colour in the preceding mixed light shot,
then I also lose the greenish false colour in the leaves.
This is mostly an artistic complaint, as I think we need to be careful with how we interpret false colour.
Because there are flowers which do have a yellow false colour after white balancing,
I think it would be difficult to know when to remove it and when not to if using this kind of mixed sunlight/torchlight.

[JCDowdy]

Andrea,

I think you possibly on to something interesting here.

As I understand it, the iridescence normally observed is a function of spectrally differential reflections form a broad band polychromatic source seen as color variation.
With your narrow band LED source it seems to me that you should only see reflections from a fraction of the mcrostructures spatially oriented to reflect that narrow spectral illuminant towards you.

I predict that if series of LED illuminated images were recorded while moving the LED through a series of different illumination angles the pattern of iridescent reflection would change.

This recalls a comment Dr. Rørslett once made regarding flashes of iridescence caused by wind induced movement of blossoms in a wide angle landscape.

Edited by JCDowdy, 28 March 2014 - 19:55.

[Andrea B.]

Oh we do already have seen many times over that indeed the pattern of "UV-shininess", i.e., iridescence, does change depending on how you move around your UV illumination.
That is inherent in the very definition of iridescence.
It's caused by the conical cells on the surface of the petals. Yes, they act as little lenses, in a sense.

Seen also in flowers moving in the breezes. Bjørn has a video of this in UV.

My concern is that the false colour difficulties, sometimes blotchiness, when using a mix of sunlight and UV torchlight, even when the UV torch is covering the flower uniformly, is concealing the actual UV signature.

I suppose this false colour mess could be from different angles of the UV torchlight and sunlight inducing the iridescence in different spectral regions. But it seems broader, or blotchier, than that in many examples. I simply had to quit shooting with UV-Led.

[I'm pretty sure I just repeated much of what you already said JD. So I bolded my main concern after rewriting this about 3 times.]

[JCDowdy]

I guess most of what I said was a statement of obviousness, but I also see potential for accentuating UV-iridescence.
As the bee flies past how much does the iridescent signature appear to flash and swirl inciting the pollinator?

[Andrea B.]

Here's another example. I was trying to shoot this Viola conspersa in a semi-shady area. I evenly illuminated the lower left flower with the UV torch. But after white balancing in the editor, the face of the little flower is blotchy in false colour. I'm not concerned with the surrounding blotchiness where I would expect things to look wrong given that the torch beam does not cover everything.

The third photo shows what the Viola should look like. This third photo did not make use of the UV torch. (..."should look like" given that we are dealing with false colour...)

UV Viola, mix of UV torch and sunlight, out-of-camera.
Aside from the typical red overload, the lower left flower looks OK.


UV Viola (preceding photo) after white balancing.
The lower left flower become blotchy in false colour.


UV Viola, sunlight only, after white balancing.
The lower left flower is not blotchy in false colour.

[Andrea B.]

In Chittka & Kevan (2005) Flower Color as Advertisement, it is noted that bees can see iridescence and do use it for detecting targets.

[colinbm]

Thanks Andrea, this is a nice demonstration.
Col

[JCDowdy]

Cool stuff indeed!

[Shane]

Another consideration is illuminating a subject with monochromatic light and capturing it with a CFA sensor. Throw into the mix, the QE is higher in the 380-400nm region vs the 365nm region.

In the UV region, the G channel tends to fall between R and B with regards to signal quantity. With broadband UV illumination (depending on CFA characteristics e.g. Nikon vs Fuji) R, G and B channels all receive some data, whereas a narrowband 365nm source will leave the blue channel (Nikon) with very little data (more noise).

Interpolation of these two sets of data will no doubt result in some differences and I also suspect the narrowband source to produce a more noisy interpolated image. Is this something you have noticed in your comparison Andrea?

[Andrea B.]

CFA = color filter array.
I had to look up this acronym. So used to seeing 'Bayer array'.

I can't really say anything about noise comparisons from the preceding foto sets because the exposure times vary and the UV-Led lighting was mixed with sunlight and not purely mono. I had no 'controls' of any kind set up. Was only wanting to illustrate some problems with UV torch illumination.

Perhaps I should try an experiment 'in the dark' and shoot a flower in UV-Flash and then in UV-Led ? Although how would you control for the longer exposure time when using the UV-Led. Hmmmm. Maybe hold the UV-Flash further away and the UV-Led closer. Some trial and error needed.