Testing for camera sensitivity - lens transmission test results

[JMC]

I was thinking about how to test for a lenses ability to pass UV, without using a spectrophotometer. It's a bit of 'Heath Robinson' approach, but I took a set of photos of a PTFE plate in sunlight with a variety of shutter speeds, but keeping ISO and aperture constant (ISO800 and f5.6). PTFE was chosen as I've measured the reflection curve of this plate before and it is pretty flat from 300-400nm. The lenses were defocussed slightly to minimise the effects of imperfections on the PTFE surface. I avoided direct reflection, and did the images in a short space of time in a 'relatively' cloud free sky. After taking, the images were cropped in XnConvert to the central 1000x1000 pixels, and analysed in Fiji using the Histogram function. This gave a greyscale value from 0 (black) to 255 (white) for the cropped image, giving the overall brightness.

 

Camera is a Canon EOS 7D mkI converted to UV by ACS. Image captured as monochrome jpegs. No additional filtration was done on the lenses (only filter was the one in the camera sensor).

 

Experiment 1 - wide angle lenses

Lenses tested; EF zoom 17-40 f4 (enabled me to do 28, 35 and 40mm), 28mm lens from Igoriginals (ebay), 35mm Soligor, 40mm EOS EF pancake lens.

 

 

About 10AM, slightly hazy light.

 

Here's the comparison of the 3 prime lenses.

All 3 primes were fairly similar, with perhaps the Soligor letting through slightly more light at a given shutter speed than the other 2. Very interesting is the Canon 40mm pancake, which is a modern lens, I didn't expect to be ideal for UV. I'm presuming given the size there just isn't that much material to absorb the UV.

 

Ok, does everything look the same, indicating a flawed experiment? Here's the Zoom lens at 35mm vs the Soligor.

Now the Soligor lets through more than the Zoom lens at 35mm, from the look of the values on the graph, about 1.5 to 2 stops. Makes sense, as the zoom should be much less efficient at passing UV (more elements, modern coatings etc). Incidentally, while I've not included the graphs here, similar behaviour was seen for the 28mm and 40mm primes vs the zoom, with 1-2 stops benefit for the primes.

 

Experiment 2 - 80-90mm lenses including the Asahi 85mm Ultra Achromatic

Slightly later in the day, I compared 3 more lenses using the same setup. Sunligh more intense now, with less cloud/haze. Time of experiment 12PM.

 

Lenses tested; Canon 80-200 f2.8 L series, Asahi Ultra Achromatic 85mm, Canon EF 28-135mm IS lens

 

As before all experiments at ISO800 and f5.6. With this comparison the 3 lenses look very different.

 

As expected the Asahi lets through much more light than the the other 2 lenses - about 3.5 to 4 stops more light. I expected the 80-200mm and 28-135mm lenses to be poor and indeed they are.

 

I would not want to compare the results of Experiment 1 and 2 though, as they were done at different times of the day, and different levels of 'haziness'.

 

Overall it looks like a reasonable approach to testing lenses transmission, but I would appreciate any comments. Obviously I am working on the assumption that the sunlight remains constant during the experiment (could be improved by using a constant light source, but heh, sunlight's free), and this is a 'real world' test looking at all the light being collected by the sensor through the ACS filter. I cannot rule out some IR obviously, as I have not seen a filter transmission curve, although other tests I've done suggest that any potential IR contamination would be very small, and much much less than 1%. In a perfect world I'd have a 77mm Baader U, but not sure that's possible. Again I could perhaps make a UV light box using UV LED's to eliminate all IR if I wanted to check that. But the principle here was to develop a cheap, easy way to test lenses, at least comparatively.

[nfoto]

Sunlight is much more variable than one would imagine, even during short time spans. You should at least use a UV light source with a wide spectral range and constant output. Magnification of detail has to be kept similar for all lenses as well.

 

Might work as a "quick'n'dirty" approach.

[Andrea B.]

An interesting test. But you are at the mercy of the geographic constraints* on the amount of UV in sunlight. Although there are obvious relative conclusions which hold up -- the UAT transmits more UV than other lenses. But we already know that. :) And you are always at the mercy of your camera software and conversion software when making histograms. Does your app analyze the raw data, the demosaiced data, the "gamma-ed" data?? And so forth. (BTW, an app like Raw Digger will easily create histograms for you.)

 

*latitude/longitude, altitude, time of day, season of year, sun storms, local reflectivity conditions (by the shore, for example), local atmospheric contaminants......I could go on, but enough. As an example, some of the shortest UV exposures I've ever gotten where while sitting out on a floating dock in a big ocean inlet at noon in mid-summer on a cloudless day. (Well, there were clouds in the distance, but not overhead.) And I was up at lat 44° in the Eastern US. Wonder what the UV would have been like down in Florida under similar conditions?

 

*********

 

I should add: The sunlight never remains constant in its UV content. I've sat outdoors with my Solar Tech Solarmeter (measures UVA/UVB in mW/cm²) pointed at the sun and watched the meter fluctuate during 5 and 10 minute intervals. Sometimes the reason for the changes are not at all obvious - like, something just "happens" in the atmosphere between the meter and the sun and the measurement will go up or down.

[Andrea B.]

Need to add something.

Bjørn's Lemma states that you can force enough UV through almost any camera + lens system given enough UV light and enough patience.

 

Whether that produces a good UV photograph is another story. (And I myself am quite deficient in the patience area.)

 

Many non-dedicated UV lenses will produce a UV photograph readily enough. The only issue is the quality of the UV photo. How noisy due to long exposures? How free of chromatic aberrations due to uncorrected lenses? How sharp? And so forth. There is a reason professional UV photographers use really good, dedicated UV lenses -- they make a better UV photograph more easily and more consistently than an ordinary non-UV lens.

 

Now, please kindly note, that is not to say that you cannot make a good UV foto with a non-dedicated UV lens, you certainly can. But you must learn all the limitations of the lens and figure out how to mitigate any shortcomings when shooting UV with it.

Well, anyway, we support in every way we can the search for good non-dedicated, UV-capable lenses because we understand that not everyone wants to fork over between (US)$4K - (US)$7K for the UV-Nikkor/Rayfact 105/4.5 or CO 60/4 lenses.

[JMC]

Yes, this to me this is part of the story - I could put in a single convex CaF2 lens and get very high transmission but the resultant image would not be good. However this was only ever meant to be a quick way of seeing how much UV a lens would transmit in relation to something else, using a cheap and simple setup. Each set of data for a lens took about 1 minute, with perhaps another minute max to swap the lenses over and run again. So for 3 lenses, maybe 5 mins. The curves themselves are 'relatively' smooth 'S' shapes, which suggests to me that the light has remained relatively constant within a run. Although yes, I fully accept a more controllable light source would be better.

[Andrea B.]

Please know that I do applaud the interesting experiments our members made to determine the UV capability of lenses.

 

We have some very creative members here on that topic, including yourself. I'm going to go look for some of the other posts and make a link list. Have been intending to do that for some time.

ADDED LATER: Informal Lens Transmission Tests: Some Links

 

I walk a fine line between recommending (as best we can) the professional gear -- so that everyone knows the standards -- and supporting the non-professional gear -- so that we can all participate. B) :D

[JMC]

No problem Andrea, I prefer spirited discussion about research. If it's not being discussed, nobody understands it, or it's not being done right (or both). Of course sometimes it's so boring to other people that nobody cares. But if it's interesting I still care :) I'm an empirical scientist, while books are important to me (hence my library on skin measurement at home which my wife loves of course) I prefer to learn by doing. I tinker, I build, I experiment, and I learn.