All You Ever Wanted to Know About Digital UV and IR Photography, But Could Not Afford to Ask

9. Which Camera?

My digital SLR cameras respond quite differently to UV and IR light. This makes some models better suited to specific areas within the UV and IR sectors. Personally, I tend to emphasize their usefulness for capturing UV, because that is my main field of interest, and besides, the most difficult task due to the fairly low UV levels in my country. Other people may have different preferences, of course.

Dandelions are for free and their two-zoned floral patterns, hallmark of true UV recording, are yours to capture with a digital SLR

Nikon D1, UV-Nikkor 105 mm f/4.5, Nikon FF + Tiffen Hot-Mirror filters, SB-140 UV flash

© Bjørn Rørslett/NN

Nikon D1 has a quite useful sensitivity at either end of the visible spectrum to allow it to record nice images by invisible light. For UV work, in particular with the lighter-coloured Nikon FF filter, I found using a hot-mirror filter mandatory, because of the camera's quite high IR sensitivity. The hot mirror unfortunately attenuates some of the UV response so unless you're shooting in direct sunlight, alternative shots without this filter should be attempted as well. If darker filters, such as Wratten 18A or Hoya U-360 are employed, the hot mirror may tentatively be left out.

I have had best processing results with Nikon Capture (v. 1.13 - 2.02) for NEF files from D1 as far as UV/IR imaging is concerned. The typical two-zoned UV reflection pattern of composite flowers can be reproduced by digital means with D1 exactly as with film, at least when IR contamination is kept low or black-and-white records are made. This conclusively documents that digital cameras indeed register pure UV images. D1 records UV mainly in the red and blue channels, and IR mostly in the red channel.

The basic problem with D1, however, is the digital noise which frequently mars such images due to the long exposure times involved, and the issue of IR contamination of the UV images, because D1 (and many other digital cameras) have significantly higher sensitivity for IR. The noise of D1 have two problematic components, those of banding noise at higher ISO setting and "hot" pixels at long exposures. In practice noise reduction for UV or IR could be troublesome and one should always aim for optimal shooting conditions.

Recently, I acquired an old D1 and managed to get it modified by stripping out the entire AA-filter pack in front of the CCD. You should not attempt to do this modification yourself, because the camera needs to be disassembled entirely before the filters can be reached. However, try talking your local Nikon repair facility into doing this for you. After the filter pack is removed, you get a tremendous boost of IR sensitivity, at least 7 stops compared to the non-modified D1 when a Wratten 89B filter is attached to the lens in use. I stress an IR filter onto the lens is mandatory, because once the filter pack is removed, the CCD is exposed to the entire spectral range passed by the lens. And because IR focuses differently from visual light, you will get terribly unsharp images. Also note that the missing filter pack will cause a severe focus shift of the system, but as it happens, since the modified camera gets "near-sighted" and IR focus also is closer than visual, these factors cancel each out to give focus on the chip coincidign with the visual focus in the viewfinder (before the filter is mounted, of course).

I designate the modified D1 as the D1R and have had much fun with it for shooting infrared.

* * *

The D1X and D1H cameras are direct descendants of the D1, and both cameras have their merits with regard to UV and IR photography. Neither of them has the IR sensitivity of D1, however, although D1X comes close.

The flowers of Potentilla anserina prominently exhibit UV-meditated flower marks to aid visiting pollinators to reach their target.

Nikon D1X, UV-Nikkor 105 mm f/4.5 lens, FF + Hot-Mirror filters, SB-140 flash. © Bjørn Rørslett/NN

Nikon D1X has the benefit of much higher resolution compared to D1 and D1H. Using Bibble for decoding the raw NEF files from these cameras, D1X yields 30MB files instead of the 7.5 MB TIFs obtained from D1/D1H. However, while D1X has the resolution edge, its sensitivity to IR is slightly less than that of D1. Improvements make the D1X less noisy than the predecessor, D1, and it handles longer exposures in a more graceful manner too. However, digital noise increases rapidly towards the 800 ISO setting so only the 125-400 ISO range should be contemplated for UV/IR work.

In contrast to D1, UV is recorded by D1X into all three colour channels of the CCD. This makes it especially important to avoid IR contamination (which records mainly in the red channel) of the UV image, unless the aim is to capture UV + IR simultaneously. Thus, a Hot-Mirror filter can often improve the quality of UV images.

Flower of Potentilla anserina, with its clear-cut UV-visible marks on the petals. Compare this to the corresponding shot made by D100, shown below.

Nikon D1H, UV-Nikkor 105 mm f/4.5, FF + CC20C, SB-140 flash
© Bjørn Rørslett/NN

Nikon D1H offers advantages for digital UV photography due to its relatively high UV sensitivity, at least +0.5 EV compared to D1, and the possibility for applying elevated ISO ratings in the 400-1600 range without introducing excessive noise. Concomitantly, the danger of IR contamination is abated because of D1H's reduced IR sensitivity, between -1 EV to -2 EV referred to D1. I have frequently used D1H set to 1600 ISO for UV photography with excellent results and in fact, this is the very reason I purchased my D1H. Strangely the green channel is highly sensitive to UV on this camera, which after all is a benefit because there are more green pixels than either red or blue ones.

On the other hand, D1H has quite moderate sensitivity to IR (after all that's the underlying reason for its success for UV work), and you are advised to shoot IR only under optimal conditions with this camera. Nikon D1H would be my second choice after D1X for IR work. The modified D1 camera ("D1R") is perhaps the best choice for dedicated IR photography, by combining greatly enhanced IR sensitivity with an excellent image quality in the IR band. Don't let its 2.7 MPix resolution fool you into considering it a low-quality imaging machine, because my practical tests show otherwise.

A flower of Potentilla anserina clearly shows the two-zoned floral marks caused by strong UV reflectance. Much of the UV signature is found in the blue channel with this camera. Insert shows the visual appearance of this yellow-flowered species.

Nikon D100, UV-Nikkor 105 mm f/4.5, FF + CC20C, SB-140 flash (UV image). © Bjørn Rørslett/NN

Nikon D100 has low noise even at 1600 ISO and might be a future candidate for UV work. I have just shot a few UV images with it and my verdict isn't finalised yet. I noticed it rendered UV images with a touch of unsharpness and lots of information in the blue channel, which is rather unusual for dSLRs.

Finally, we have Nikon D2H, the newest addition to the Nikon DSLR range. It has excellent potential for shooting UV, and may be slated to become my new UV workhorse (later it turned out this wasn't the case). Although its IR sensitivity all-over is lower than that of the D1-series models, D2H more than makes up for this by having a superb noise reduction so given your IR target is approximately stationary in time, the D2H certainly can capture it in IR with breathtaking detail.

I have recently switched to using the Nikon D70 camera as my main work horse for UV photography. The camera responds quite well even to the very dense U-340 filter, images show plenty of detail, and the long exposure noise reduction performs much better in UV photography than under visible light for some strange reason. The camera is a little handicapped by its small viewfinder, so adding focusing light is frequently needed to ensure adequate focus, and I'd rather refer something else than the IR remote control which masquaredes as a cable release. Nevertheless, the UV images speak for themselves, so D70 it is. Why D70 instead of the professional D2H? Simply because D2H had its UV response confined mainly to the red channel, thus making the post-processing of the image difficult as far as getting nice colours were concerned. With IR the situation is more in favour of the D2H, however.

A second D70 camera has been purchased to be modified the way I did with my D70. By removing the AA filter pack inside IR sensitivity greatly increases, just as I found with D1. However, the output even with dense IR filters such as 87C occurs mostly in the red channel unlike the modified D1, so you get more image pixels but not a better IR image. UV response of the stripped D70 is better than with the original D70, but not spectacular.

Nikon D2X has a CMOS imaging chip inside and behaves accordingly, with low UV sensitivity and response in the blue channel (as to its IR response, I haven't had opportunity to try it out yet, but my hopes are not set high*).

*Testing a production-level camera confirmed my suspicion that D2X is not suited for IR, but it can, with caveats, be used for UV landscape work. All this is covered in my D2X review, which should be consulted for further details.

* * *

All right, you lust for experimenting with UV or IR photography, but the Nikon digital SLRs are out of your financial reach? Several of the lesser cameras, for example, Nikon Coolpix series, can be pressed into service for this kind of photography. My personal experience with these cameras is limited so I cannot go into details here. I briefly tried a friend's Coolpix 5000, and although it did record IR, it obviously was not comfortable in doing so. Thus, IR sensitivity was quite low (excellent news from the designer team's point of view, but bad news for the IR geeks), and there was troublesome hot-spots due to internal reflections within the optical pathway.

Recently, I acquired a Nikon Coolpix 950 just for having some experimental fun. This camera features a 2.1 Mpix Sony CCD with known response in near IR. Eventually I plan to remove the anti-IR filter in front of its CCD, to turn the 950 into a real real-time IR camera. Meanwhile, playing around with it using different filters is instructive. Coolpix 950 records IR easily enough through the 89B filter and the resulting colour balance is pleasing. Substituting a Wratten 87 makes the images quite colourless, but simultaneously the Wood's effect is evident. If any of my UV bandpass filters is put onto the 950, reddish-tinged pictures are obtained. Likely they are anything but UV images, but arise from narrow-band spectral leakages in deep purple and near-IR. Still, lots of fun to tinker with.

Even with a dedicated UV-flash (SB-140) to illuminate these flowers (Leontodon autumnale), the Coolpix 950 doesn't deliver anything similar to the clear two-zoned UV pattern shown with the UV-Nikkor lens on my D1H. Both pictures taken with Hoya U-360 and Tiffen Hot-Mirror filters to reduce IR contamination, and the SB-140 had its black SW-5UV filter attached. Please note the shooting conditions would be suboptimal for D1H, which doesn't need the hot-mirror filter.

* * *

Fuji S2 Pro has been shown to possess a useful IR response, as indicated by the example shown below. Its IR sensitivity using an 89B filter is about 4.5 EV below the corresponding visible exposure, which is quite good. The S2 allows examination of the histogram of each of the three colour channels, which is a beneficial feature for experimental IR work.

Lately, I verified that S2 Pro has the ability to record true UV images. An added surprise is that S2 Pro evidently has most of its UV response in the blue channel. I'll return to this camera when the conditions for UV photography improves, being in the middle of a freezingly cold snow-packed winter isn't the optimal situation for UV work.

Digital IR with Fuji S2 Pro and AFS 24-85 mm f/3.4-4.5 Nikkor, 89 B filter
Ola-E. Hofshagen 2002

Recently, I'll availed myself of a friend's D60 to do some test shooting of UV and IR scenes. D60 might offer a promising potential because of its excellent noise reduction for long exposures. However that feature would not be of value unless the CMOS sensor has sufficient sensitivity to UV or IR radiation.

IR tests with Wratten 89B indicated that Canon D60 gave severe underexposure, and the camera meter setting should be corrected at least +2 EV. Images were grainy and noisy even at 1/6 sec exposure (@200 ISO), and IR triggered a number of hot pixels as well. Probably I did not set up the D60 in an optimal manner so Canon enthusiasts might be able to extract more out of IR shots made with this camera. However, expectations as to its IR performance should be kept at a realistic level.

Canon does not offer any quartz lens, so in order to assess the UV performance against a known reference, I literally grafted my UV-Nikkor 105 mm f/4.5 lens onto a D60 body. What one wouldn't do to get the relevant shot! The owner of said D60, incidentally a close friend of mine, nearly had a heart attack, but managed to calm down and help further test shooting by keeping my SB-140 UV flash into a proper position. With this make-shift setup I quickly shot some close-ups of Potentilla atrosanguinea, a species with a bull's-eye type of floral pattern prominently visible in UV. Exposures were my standard close-up settings for D1H @ 800 ISO and cameras were set to manual mode so SB-140 would be the only light source. Thus, the test images should come out identical if D60 and D1H exhibited a similar response to UV. Ambient light levels were low and could not influence the results.

Comparison of CMOS (Canon D60) and CCD (Nikon D1H) performance in UV. Test subject Potentilla atrosanguinea

(using the same lens and flash for both images)

Canon D60 Nikon D1H
UV-Nikkor 105 mm lens at f/5.6, 1/125 sec, Hoya U-360 filter, 800 ISO, SB-140 flash UV-Nikkor 105 mm lens at f/11, 1/125 sec, Hoya U-360 filter, 800 ISO, SB-140 flash

However, as plainly shown by the test shots (above), Canon D60 did not like UV at all, and I had to open up the aperture several stops from the test setting to get any image. Thus, its UV response with the dense Hoya U-360 filter was more than 4 EV below that of the D1H, giving rise to poorly exposed UV images with plenty of image noise and triggered hot pixels. Substituting a Nikon FF instead of the U-360 bettered the difference, but still D60 produced much darker images (approx. 3 EV below D1H). On a small positive note, the UV response of D60 occurred in the blue channel, similar to film. On another, less positive note perhaps, trying to shoot UV with a Canon EOS 100 mm f/2.8 EF lens @f/5.6, using a Hoya U-360 and the SB-140 UV flash, yielded virtually black images without any detail, confirming that multi-coated lenses are unsuitable for UV imaging.

Well, at least I did give the Canon D60 a fair trial. I can only let the test images speak for themselves, they are directly comparable.

Published, but unsubstantiated, claims have been seen by me that the earlier model, Canon D30, had a useful UV response. I find this highly unlikely because no true UV lens was used, only multi-coated optics which are well-known to block UV, and no stringent measures were taken to ensure only UV light entered the camera. Probably the images (which I have seen) result from filter leakage in deep purpe and/or the border between deep red and near-IR.

A few weeks later I had another opportunity to use a Canon EOS 1D with my UV-Nikkor lens. Again, the UV response was severely attenuated (approx. -2 EV) compared to my D1H using the same UV-Nikkor lens, but colour rendition and detail were quite good. In common with the D60, Canon EOS 1D had predominately a UV response in the blue channel. Excellent as they may be for other purposes, neither 1D nor D60 is really suited for critical UV work. This is not to say their use is impossible, it's just impractical.

All You Ever Wanted to Know about Digital UV and IR Photography, But Could Not Afford to Ask

| To Top | Far Side | Gallery | *UV | *IR | Lenses | Links | Personal | Professional | Reviews | Start |

Last update 15 March, 2005