How to use filters for UV photography

[nfoto]

Rørslett, B. 2013. How to use filters for UV photography. http://www.ultraviol...uv-photography/

 

In order to make UV photographs, all spectral bands outside the UV range need to be excluded. Keeping in mind the spectral energy in the 300-400 nm range is much lower than what arrives in the visible, or IR ranges, this makes efficient bandpass filtration quite difficult. Thus, with say IR being many times stronger than UV, even a small leakage of IR can severely degrade the UV capture.

 

When silver-halide technology (also affectionally known as 'film') was dominant, this side effect of IR leakage was not too important because most films had virtually no IR sensitvity. However, a digital camera behaves entirely different and has potentially a very high native IR response.

 

Let us see what influence various filters can have on the final outcome of a UV capture. The subject is the familiar mop of a dandelion (Taraxacum sect. Ruderalia), which is known to have a pronounced UV signature. A Nikon UV-blocking filter, L37C, to the lower left.

Visible light. Nikon D70, UV-Nikkor 105 mm f/4.5 lens, no filter, daylight.

 

Now, add the propriertary Nikon FF filter, an equivalent to the discontinued Kodak Wratten 18B, and according to Nikon's specifications, a UV image should result. However, as this filter leaks significantly in near IR, this component washes out any UV signature of the dandelion flower head. Also note the UV blocking filter, to the lower left, is just a little darker than the unfiltered version. This slightly disappointing outcome simply reflects the vintage of the filter as it was designed for use in conjunction with the UV-Nikkor lens on a film camera. Thus while it worked quite well in the pre-digital era, it is entirely unsuited for digital UV photography.

 

 

Combining the FF filter with a B+W BG-38, to cut down the IR leakage, improves the rendition and now the UV signature of the dandelion starts to appear. However, the contrast between the darker centre and the outer ray flowers of the dandelion is less than desirable. Also note that the UV blocking filter target still is not entirely dark. This results because the filter package does not have a steep cut-off, so is inefficient in preventing leakage from spectral bands outside the desirable UV passband.

 

Only when deploying a dichroic UV bandpass filter, Schott UG11X, in which a steep roll-off is used, we observe a UV scene without noticeable non-UV contamination.

 

Another subject, Colt's foot (Tussilago farfara) also from the Daisy family, serves to illustrate how filter technology has progressed to give much improvement in the resulting UV images.

 

Nikon D200, Novoflex Noflexar 35 mm f/3.5 lens, Baader UV/IR Cut filter, daylight.

 

Nikon D200, Novoflex Noflexar 35 mm f/3.5 lens, Baader U (Venus) filter First generation, daylight.

 

Nikon D200, Novoflex Noflexar 35 mm f/3.5 lens, Baader U 2" (Venus) filter Second generation, daylight.

 

We observe that by using the more recent filter the central disc flowers are now almost black to indicate they do not reflect UV, and their rendition is no longer contamined by residual IR. In particular these "bull's eye" patterns are very susceptible to IR contamination because any leakage of the filter will literally wash away the black signature.

 

Also noteworthy is that due to the dichroic nature of the Venus filter, there is a tendency towards unevenness in its attenuation across the frame when used in conjunction with a wide-angle lens. One can readily observe a slight magenta cast in the peripheral areas of the image, caused by this effect. So this kind of filter is best combined with longer focal lengths having more restricted angular coverage.

 

[Published 5 Feb 2013 Last update 25 Aug 2013]

[OlDoinyo]

Radial tint gradients can be caused by dichroic falloff, as you say here. It is worth mentioning that they can also result from uneven channel balancing and/or inadequate compensation for such, especially with wide-angle optics that have significant cosine falloff. Digital "well effects" near the edge of the sensor can sometimes contribute, where the incoming light is not normal to the sensor surface.

[OlDoinyo]

In case anyone is interested, this author has compared filter stacks:

 

http://myphotojourney.co.uk/uv-pass-filters-a-comparison/

[Andrea B.]

Clark, thanks for the link to Boon's filter test page.

 

Dr. Klaus also has several such tests.

And we have some posts here also.