Jump to content
UltravioletPhotography

The Quest for Ever Wider: Breaking the 20 mm Barrier


OlDoinyo

Recommended Posts

Wide-angle lenses are one of the challenges of UV photography. An ideal UV lens passes as much UV at as wide a range of wavelengths as possible. Dedicated quartz lenses exist which are optimized for this purpose, but aside from their extreme cost, they are marketed for scientific, forensic, and other technical purposes, not for artistic photography. As such, almost all such lenses are made in a narrow range of largely short-telephoto focal lengths. These technical applications have no use for wide-angle optics. UV photographers wanting wide-angle optics must rely on lenses which incidentally happen to transmit enough UV for their purposes. In the long-wide-angle range around 35 mm, there are actually a number of outstanding choices, but below this range the search becomes increasingly fraught.

 

The reason has to do with the nature of the shortest-focal-length optics. Often the nominal focal length is less than the camera's flange focal distance, so especially with older SLR designs, one is faced with having to keep the optics clear of the mirror box in front of the lens, usually meaning that the closest portion of the lens must be kept 40-45 mm in front of the film or sensor. The only way to achieve this is via retrofocus designs, which tend to be complex affairs with many elements and groups. Lenses designed for digital sensors also have extra collimating elements in the rear to minimize the incident angle of the light exiting the rear of the lens (to control well falloff.) All this makes for optics with a lot of total glass thickness and a high chance of containing several types of glass, some of which may not be UV-friendly, and most lenses shorter than 35mm are frankly awful when it comes to UV bandpass. To make matters worse, the front elements of the shortest optics must be quite thick, especially at their edges. Nevertheless, to date three lenses have been identified in the 20-21 mm range which provide more or less usable results: the Tamron 21/4.5, the Asahi 20/4.5, and the Bushnell 21/3.8. But what about even shorter? It would seem nice to break below 20mm into the true short-wide-angle range if something suitable is out there. The rest of this post details such an exploration.

 

Two possibilities came to attention: the Tokina 17/3.5 and the Tamron 17/3.5, and examples were obtained for testing. The first step in the process is a quick bandpass check, carried out with a pinhole test using a brand new aluminum target, a.k.a. a cookie sheet. It would be folly to expect this type of lens to have the bandpass of a Cassar-S or the like; my Asahi 20/4.5 was used as a more reasonable standard of what can be considered adequate.

 

A new test of that lens:

 

post-66-0-27381200-1618282836.jpg

 

Next, the Tokina:

 

post-66-0-20001400-1618282945.jpg

 

The transmitting area of this lens looks dark as well as deeply orange colored--not a good sign. I would rate this result somewhere between poor and dreadful. I will not be discussing this lens further, although a more detailed exploration of what may be a very similar lens has been posted elsewhere on this board.

 

There was reason to hope the Tamron might give a better result. On a Pentax Forum post in 2016, a user had this to say about this lens:

 

Holy crap, does this lens love sky! Blue skies just explode off the frame

 

This kind of experience by visible photographers can be a hint that there might be substantial bandpass depth on the far side of the 400 nm line. A similar remark led me to the Asahi 20mm lens, in fact.

 

And so it proved in fact:

 

post-66-0-05089500-1618283504.jpg

 

Ignoring the specular artifacts from the sun, this result in fact looks not dissimilar from that of the Asahi, implying somewhat similar bandpass--an acceptable result.

 

Before proceeding to further testing, a few words about this lens. It came in two versions: the earlier 51B version, which has an internal filter wheel, and the later 151B version, which has no filter wheel. In addition, a lens hood was sold separately which could be affixed to the front of the lens with pressure thumbscrews. This hood has an 82 mm filter thread inside it for front-mounting filters. By an incredible stroke of luck, I found one of these hoods in the clearance bin of a local camera store--they are not often sold with the lens. The lens in my possession is the 51B version and the filter wheel has four positions: Normal (maybe just clear glass?) 81B (for correcting tungsten film to daylight,) 80B (for correcting daylight film to tungsten lighting,) and Y2 (a 475LP filter of unknown purpose.) The photos in this post use the Normal position. The 80B position gives a very similar result, though perhaps a hair dimmer. The other two filters black the lens out. For the remaining photos, I obtained an 82->77 step-down ring and mounted my U360/S8612 filter on the front of the lens using the hood. This results in some corner clipping, and a better scheme needs to be devised, but it is adequate for testing purposes.

 

The next step is assessment of image quality. A test scene was chosen, and for chromaticity reference, the center of the scene was photographed with the Baader U2 and pinhole:

 

post-66-0-39649200-1618285767.jpg

 

At this bandpass, the two vehicles appear almost the same color, a bright orange. The touch-up paint on the van at right shows clearly--they did not bother to match the UV reflectance.

 

Now the lens at maximum distance, f/3.5:

 

post-66-0-81404400-1618285960.jpg

 

The lens clearly struggles to reach infinity focus at this aperture; there is a substantial focus shift, considerably more than the Asahi.

 

And at f/22:

 

post-66-0-38496100-1618286133.jpg

 

Focus and sharpness are now reasonably good in the center, though not as good as the Asahi's.

 

post-66-0-25260100-1618286275.jpg

 

In the corner, there is quite a bit of chromatic aberration, more comparable to that of the Tamron 21 with rear-mounted filters:

 

post-66-0-12576100-1618286376.jpg

 

Finally, though I am inexperienced at this, I attempted to devise bokeh tests. I used the reflection of the sun off a chrome-plated wrench handle as a point source, which may not have been the best choice. First I tested foreground bokeh at the infinity stop, f/3.5:

 

post-66-0-92165000-1618286642.jpg

 

The presence of trees near the sun contaminated this result somewhat, but as nearly as I can tell the bokeh is near neutral here.

 

Next, background bokeh at minimum focusing distance:

 

post-66-0-59775700-1618286779.jpg

 

This bokeh seems weakly negative overall. The bright spot in the center might be an artifact of the size of the sun's disk in the reflection.

 

In conclusion, the Tamron 17/3.5 is found to be usable for UV photography, although it has a lot more focus shift and chromatic aberration than the aforementioned Asahi 20. This is the widest such lens discovered to date. Certainly, if you need the widest angle coverage possible in the UV, this, at least for now, is the ticket; but it might behoove the user to brush up on techniques for removing chromatic aberration in Photoshop. If you do not need such wide coverage, a lens such as the Asahi 20 will be a bit friendlier to use.

 

Would it have made any difference to use the version without the internal filters? It is possible, but I think perhaps not much--this lens acquitted itself about as well as can be expected in the bandpass department.

 

Where might the search go from here? Even shorter are the Asahi 15 and its Zeiss clone, but both are a bit too expensive to purchase for casual testing, and the Asahi has a potentially more problematic filter wheel built-in. There are also 15 and even 12mm offerings from Voigtländer, but those cannot be mounted on SLR cameras. So this is where it rests--for now.

Link to comment

I hadn't thought to test this lens. But I too own the Tamron 51B with filter wheel. I don't own the front filter hood though. But maybe able to reverse mount a filter.

 

I will say so far in my tests the Sigma 10-20mm f4/5.6 lens is the widest rectangle lens I have tested that is useful in UV. The spectral UV cut off is 378nm.

 

Link to comment

I thought of mentioning the Sunex in my post. But fisheye lenses are a strange niche product, kind of their own thing in a way, and the best ones seem very rare and expensive, though the advent of GoPro video has perhaps made the idea a bit more mainstream as of late. It is mathematically possible to remap a fisheye image as rectilinear, but in practice this results in laughably bad image quality at the edges.

 

The Sigma 10-20 is interesting for being a zoom lens, but it is not full-frame, so the comparison is a bit misleading--I don't think its FOV at 10 mm on an APS-C camera will be much wider than the Tamron's on a FF camera. I guess I have limited my search to rectilinear FF lenses. I believe the widest such ever made was 12mm.

 

Since a number of members now have access to actual spectrophotometers, I offer a humble suggestion: When not posting an actual spectrum, rather than saying the cutoff is 378 nm, which can be puzzling to interpret, one might report the wavelength of maximum transmissivity and then the wavelengths at which this drops by 1, 2, and 3 stops respectively from the maximum value.

Link to comment

I will say so far in my tests the Sigma 10-20mm f4/5.6 lens is the widest rectangle lens I have tested that is useful in UV.

 

Useful only for up to APS-C!

Link to comment

Just quickly tested my Tamron adaptall 17mm f3.5 lens on the first Normal filter setting. Its more than a stop slower than the Sigma 10-20mm. It also only works well with my 390bp25 filter and not the 330WB80 improved filter, which does work on the Sigma.

So at least for me one the first Normal filter setting its poor for UV and most likely only allowing the upper 390s through.

 

In hindsight I should have tested both Normal filter settings. Maybe another time when the sun comes back.

Link to comment

Try the 80B setting.

 

I can't quote actual numbers, but I don't think the images above would be obtainable with a cutoff in the upper 390s, especially the amount of chromaticity and CA. So that is puzzling. Are the two normal settings different?

 

I have occasionally heard of cases wherein two seemingly identical lenses do not act the same. I do not understand the why of it.

Link to comment

Ok quickly scanned the lenses on my Spectrometer. The Tamron 17mm f3.5 has absolute cut off at 350nm. The curve is slightly shallow sloped. The first normal is different than the last normal on my lens cutting 356nm ish.

The Y2 cuts off as expected and 81 setting is similar to my normal 1.

The best UV transmission was with 80B setting, which surprised me.

 

The Sigma 10-20mm f4/5.6 cuts off at 370nm, but has a broader UV peak if that makes sense. The area under the curve for the 10-20mm looks to be greater than the area under the Tamron 17mm.

So now I will have to retest the Tamron with 80B setting outside. Being adaptall leads to more rear filter modifications. I didn't even know a 82mm hood filter adapter existed. I will have to see if I can modify the front. I have seen some people modify 67mm filter rings, I will have to see if that can work.

 

Also scanned my Sigma 8-16mm and its too close to 400nm, but does let some UV through.

These Sigma's can be used an 135 format camera, but not at widest settings. The 8-16mm fills the frame at 14mm, 15mm and 16mm settings. I don't know the coverage of the 10-20mm as I don't have 135 format Canon camera and my lens is in Canon mount.

Link to comment

Hmm.. I suspect the contents of your filter wheel and mine might not be identical. For me, Normal and 80B showed fairly similar transmission. To be fair, I did not test both Normals, as it did not occur to me that they might not be the same physical filter. Presumably, the 80B has its own UV rolloff characteristics, but whether they are dictated by the colorant or by the glass is unclear. I begin to wonder if the 151B version might be a safer bet for prospective buyers, assuming it was not otherwise changed in an unfavorable manner.

 

I am having an adapter 3D-printed which I hope will allow me to fit my 77 mm filter on the front in such a manner that there is no corner clipping. I will save the hood for IR use, as I have some 82mm filters I can use for that (the lens is good for IR, btw.). But I do not know how big your filters are and whether this would be an option for you. I agree that this lens is not the most filter-friendly--filters are very awkward to mount and remove on the hood due to difficulty gripping the edge of the filter.

 

If you can post those transmission spectra on here, that would be useful.

Link to comment
A 67mm to 77mm step up ring fits perfectly on the lens without any vignetteing. I temporarily used double sided tape. I might glue it if I want a permanent solution. That would also allow for better snap cap as mine always falls off.
Link to comment
I have just learned that some examples of the 51B had a red filter in place of the 80B. I do not know if anything else differed.
Link to comment
  • 2 weeks later...

Ok no direct sunlight, but was able to still play with my Tamron adaptall 17mm f3.5 lens. My last Normal setting was 1/3 stop faster than the 80B and first Normal setting, which were the same. I was using a 77mm Bg39 2mm thick, with 77mm ZWB2 filter.

So at least with my copy the last Normal was best outside with only sunlight as a light source.

 

It would be nice to know how the lens comes apart and if a IR filter or ZWB3 could be placed in the 81B spot.

 

I know Cadmium flipped out the in lens filters on one of his Nikon lenses once. I wonder how much space we have in this lens.

A 1.5 mm U360 + 2mm S8612 or Bg39 might work.

Link to comment

I thought of the same thing, but I fear randomly adding thick elements in the middle of the optical train even if physical clearances allow is very likely not going to go well, especially in a lens already struggling with infinity focus. Deleting one of the filters altogether or replacing one with a dummy Spectrasil disk might go better. It is too bad--frankly, the color-compensating filters are useless except for color slide film, while the other Normal, if it is a 395LP, is also without much use in modern times. It would be nice to replace these filters with a Baader, an R72, an S8612 for deconversion, and a #12 (my fantasy.)

 

I also do not know how light-tight the built-in filter system is. It does not matter so much in the visible but it is crucial for UV or IR.

Link to comment

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
×
×
  • Create New...