Saturday, 19 September 2015

Fun With Filters - Now With Dedicated UV

I know this blog is primarily about infra-red and don't worry, it will stay that way, but I want to talk about UV for a bit, since I have a full spectrum camera. Since I managed to get my hands on a proper UV filter recently (the Baader U) I wanted to do another comparison image between the different wavelengths, but this time with UV as well. Unlike the dual-pass Hoya U-360 that I've used in the past this UV filter blocks all infra-red light as well as visible colour.

Now I'm very new to UV photography, so I may mess up some of the technicalities here, but I'll do my best to get them right. If you see anything wrong please do let me know. Anyway, this is my experience with shooting UV after a couple of days of tests.

Ultra-violet (UV) photography not as easily accessible as infra-red, so it's not as common and here are a few reasons why I think this is:

Firstly the filters are rather expensive (this one was £200!), perhaps partly because they're so rare, but I don't think this is a cycle that can be broken for a few reasons. I was lucky enough to be able to borrow one, but otherwise the cost would have put me off trying one. These filters only work on cameras that have been converted to full spectrum.

The second hurdle is that they don't easily come in common filter sizes. Since they're mostly used for telescopes rather than cameras they most commonly (if you can call it that) come in the 2 inch size. If you're an astronomer you probably know what this means. I assumed that 2" was 50.8mm, but after a lot of research I now know it means 48mm. On what planet does that makes sense? Whatever, at least now I know. Lucky (for me this) is close to the filter size of my Sony autofocus prime lenses - the Sony 28mm & Zeiss 55mm. Also fortunate is that the 49mm to 48mm step-down adapters are pretty easy to find, not too expensive and don't cause vignetting on either lens.

Thirdly - Exposure is not a trivial issue either. The amount of UV light that these filters capture is pretty low*. It's nothing like what you get with infra-red, so usable exposures at base ISO are almost out of the question. If this main example image is anything to go by then it's roughly 10 stops worse off than colour - ouch! Don't forget, this is a converted camera we're talking about, not an off the shelf model. Motion and noise are going to be a tricky trade-off and tripods are going to be needed for higher image quality.

* NOTE: I've since read that many lenses don't transmit UV light well. Some old, uncoated lenses can be OK. Some others can be modified. Some lenses have been specifically made to shoot in UV, but these are rare and very expensive. I need to do more research on this. I'm not sure if this has potential to fix the issue, I will keep looking into this. I did try an old Pentax lens, but it was worse than the Sony on exposure times and image quality.

Last, but not least, is the aesthetics. This is largely subjective of course, but at least to me UV photography doesn't seem to be as obviously pleasing as Infra-red. Clear skies come out very bright (if you want any detail on the ground), most other tones look like a pretty 'normal' black & white image and people? Well, they look down right horrific! Perhaps you like how moody people look in UV, you might want to play with sunscreen or try some astro-photography. Then there's the interesting way that flowers come out in UV (bee vision). As difficult as it might be to find, the fact remains that there is still potential here.

I was curious how UV could be used for buildings and landscapes. This was the driving force behind this initial test. Here are the different sample images in full. Hopefully this shows the differences more clearly:

All this really tells me is that UV isn't great for landscapes. It almost looks like a normal black & white photo, with a blown out sky. Most of the material tones shown here are pretty similar to how they appear in colour.

There were some differences in corner sharpness between the samples, but I mostly put that down to the varying quality of the filters rather than an issue with the wavelength, but I could be wrong. This lens doesn't have the best corner sharpness anyway, even at f/8. I've applied all the lens correction to these samples too. This means a significant change in vignetting, distortion and chromic aberration (CA) removal. All this negatively affects corner quality, albeit equally throughout the samples. Obviously the longer exposure on the UV sample has introduced some motion blur in some of the foliage, but hopefully this obvious where it occurs. Anything else that can be gleaned from these samples I leave up to your judgement.

Here is the camera setup that I used for these shots, minus the tripod. The lens hood came in useful for these first group of samples. When I didn't use it I noticed that flaring mostly occurred with the red and IR filters. That's not to say it doesn't occur with colour, it does and especially when using external filters, it's just more forgiving with glance angles.

Time for another sample comparison from a different scene. This was a bit quicker to take due to me not putting the lens hood back on each time. Since the camera was facing away from the sun and in complete shade it seemed OK not to use it. I thought the water in this scene would be helpful to compare the wavelengths too, but this combined images isn't a great example so please check out the individual shots below.

The settings on these four shots were similar to before accept all were taken at f/11 this time. The shutter speed settings were as follows (from left to right):
  • UV: 5 seconds
  • Colour: 1/90th
  • Red: 1/180th
  • IR: 1/15th

Here's one final scene. This isn't as useful as the other two, but since I took the time to get all the shots I might as well show them.

The settings on these four shots were the same as the previous scene (f/11). The shutter speed settings were as follows (from left to right):
  • UV: 20 seconds
  • Colour: 1/60th
  • Red: 1/125th
  • IR: 1/15th

Just as there are lenses good for IR, there are also some that are better for UV photography too. Unlike IR (which is mostly hot spot issues due to the lenses inner barrel coatings), UV is more due to the number of elements and coating on the front. Some lenses were made specifically for UV photography like the UV-Nikkor 105mm for example. Most modern lenses have more elements (for correcting various issues) and are quite heavily coated, so make for poorer transmission of UV light, but they will still work, as these samples show). You can read more about this issue here.

In the second sample (above) I did notice that there seems to be some kind of flaring on the UV photo compared to the others. This Sony lens (like most other new Sony / Zeiss lenses) does produce an infra-red hot spot too and sometimes quite badly. Modern optics have more issues when capturing wavelengths either side of the visible light. By blocking parts of the invisible wavelengths the visible light that remains is improved, so this makes a lot of sense. Using older optics is beneficial when capturing invisible light, so having a camera that can adapt to any lens mount is very handy. This is true of many mirrorless systems, but Sony seems to have great support for this.

I mentioned astral photography, people and flowers as being interesting subjects for UV, but I haven't shown any here (sorry about that). I will try to get more examples for this and will come back to report my findings, but for now I just want to cover the basics.

Here's a quick example of a friend from work in Full Spectrum vs UV. This was supposed to be colour vs UV, but I forgot to bring in the colour filter. Factor 50+ sunscreen was applied to the left half of his face for the test.

Here's me with some factor 100 sunscreen on, after I've washed my face with soap and water, showing that it does indeed do what it says on the bottle.

1/30th second, f/2, iso 2000

Here's an example of my brother in-law (in all spectrums), who's not wearing any sunscreen because he's allergic.

There certainly are reasons to shoot in UV, but they aren't necessarily immediately obvious. Generally the world just doesn't look as ethereal in UV as it does with IR. This makes it a pretty tough project given it's price and limitations, so I can see why it's so niche. I will continue to experiment with it, as long as I can borrow this filter.

To finish off I wanted to show a fully processed colour IR image. All the colour IR photos above are straight out of the camera and I rarely use them like that. Here's an example of a fairly standard process that I put them through. This is the 'False Colour' technique (blue/red channel swap), with some 'Hue/Saturation' adjustments and some extra vignetting.


  1. I would disagree that UV is not for landscapes--one must, however, approach it on its own terms. It does not have the in-your-face surreality of IR, and if you are hoping for an IR-like look, UV will disappoint. But if you get IR out of your head for a moment, some interesting things are indeed possible. I have seen some very good landscape work in UV (and I rather like your portrait, too.)

    1. Apologies if I offended with that comment. I'm curious what you find the advantages are and if you have any examples I'd love to see them.