7. Infrared (IR) photography with D3

One pleasant surprise with many of the digital cameras is their ability to record useful imagery outside the visible spectrum. Typically, the sensitivity to near-IR might be sufficiently high to allow IR photography not only in bright sunlight, but also under less well illuminated conditions. So, a question one is assured of appearing with any new DSLR these days is "can it do IR"? Read on to find the answer.

The infra-red (IR) spectral band, above 700 nm wavelength, makes up a significant fraction of the sun’s rays impinging upon Earth’s surface. We cannot perceive these rays directly as making an impression of light as such, but we do sense the effect of IR heating of the skin. Electronic flash devices, incandescent bulbs, and more esoteric "light" such as that emanating from IR heaters, all contribute IR to our photographic subjects. So when there is some light, IR will be present too. If you doubt that a flash can output IR, just hold your hand in front of the flash tube to feel the heat wave when the flash fires (you are duly warned, the flash will get hot).

The camera maker, however, does not want infra-red rays to enter the imager inside the camera at all. Why is that? The main reasons are that IR comes to focus differently from visible light, and that many subjects including human skin, reflects IR in an unexpected manner. Thus, human skin can take on a blotched appearance, and subcutaneous veins may become visible. Neither response is conducive to a pretty portrait model. Another well-known problem is the high reflectance of IR by certain fabrics, so one can get a suit rendered in ugly magenta hues instead of true blacks. In fact, a darkly-dressed person most likely will appear bright white in an IR shot. The focus difference of IR robs the final image of detail sharpness.

In order to mitigate the side effects from IR, there is an anti-IR filter in front of the imaging chip inside the camera, typically seen as a greenish sheen when the imager chip is exposed to light. The anti-IR filter is combined with an anti-aliasing (AA) filter in a filter pack, which may or may not be removable. Each generation of Nikon D-series cameras has had its IR sensitivity decreased, or put the other way, the efficiency of the filter pack has increased. Nikon D3 is certainly no exception to this general trend.

While I attended a Q&A session at the D3 production facility at Sendai, Japan, in August this year I asked about the IR response of the D3. The senior Nikon engineers heading the session were baffled by the question, but one of the second-tier aides (those smartly suited youngsters sitting with all the papers at the second row behind their superiors) came up to the blackboard and drew the transmission curve for us. It was very obvious that Nikon had implemented a very efficient IR-cut filter in the D3.

Since I'm an inquisitive soul by nature, I availed myself of a temporary clearing in the season's foggy weather to run a quick IR test of the D3. The reference as usual being the modified D200, and I employed the IR workhorse 28 mm f/3.5 non-AI with its CPU modification and the B+W O-93 filter (Wratten 87C equivalent).

The results corroborated what the Nikon engineers - a little unwillingly - had shown me at Sendai. Nikon D3 has a very low response to IR. I had to go all the way up to 25600 ISO and 20 secs. at f/4 to get any recorded image at all. Besides that, the image quality was appalling. Now, this exposure was no less than 15 - fifteen - stops poorer than my reference camera, the D200 (modified) with the same lens and filtration.

Below is the result from D3 (25600 ISO, f/4, 20 sec, 28mm, 87C). Note the lack of contrast and the vignetting in the extreme corners. I have removed a strong reddish cast in post processing, otherwise the image is unaltered. Now, compare this to the D200 IR image acquired 30 seconds earlier. There is a world of difference here (D200: 1600 ISO, f/11, 1/25 sec, 28mm, 87C).

Infrared Landscape - I Nikon D3, Nikkor 28 mm f/3.5 lens (AI-modified), B+W O-93 (Wratten 87C) filter on lens. 30 sec exposure at f/4 @ 25600 ISO equivalent (Hi 2) This exposure is no less than 21.7 stops below the corresponding capture made in visible light

Infrared Landscape - II Nikon D200, Nikkor 28 mm f/3.5 lens (AI-modified), B+W O-93 (Wratten 87C) filter on lens. 1/25 sec exposure at f/11 @ 1600 ISO Compare this to the corresponding shot with the D3, shot 30 seconds later.

In case you are not entirely convinced, here are the "mandatory" 100% crops (actual pixel size in which a pixel in the file is represented by a pixel on your monitor screen for these cameras. I leave it an exercise for the keen reader to explain why the apparent size of the two 100% crops is different).

Infrared Landscape - I (actual pixel size) Nikon D3, Nikkor 28 mm f/3.5 lens (AI-modified), B+W O-93 (Wratten 87C) filter on lens. 30 sec exposure at f/4 @ 25600 ISO equivalent (Hi 2) This exposure is no less than 21.7 stops below the corresponding capture made in visible light. It obviously has a very poor image quality.

Infrared Landscape - II Nikon D200, Nikkor 28 mm f/3.5 lens (AI-modified), B+W O-93 (Wratten 87C) filter on lens. 1/25 sec exposure at f/11 @ 1600 ISO Compare this to the corresponding shot with the D3, shot 30 seconds later.

The outcome will to some extent be set by the ambient weather and to an even greater extent by the IR filtration applied to the lens. Suffice it to say that D3 is nearly immune to ill side effects from IR. As with any digital camera, if the shutter is opened for a long enough time, the imager can record some IR. So also with the D3. But you need to keep the full perspective here. Is the response relevant for the intended applications? Only the end user can answer that question.

A few days later, I set up a new series of test shots of a sunlit landscape. Using the Hoya R72, a very weak IR filter, now gave a response –19 EV ref visible light. These weak filters (89B types) can leak significantly in the deeper red borderline to near IR. The B+W O-92 is an example of this kind and has a prominent deep-red leakage. With the more proper 87C class filter, the response was –21.7 EV ref visible light. I also ran my D200 (modified for UV/IR) concurrently with the same filters and verified my earlier observation of a 15-stop difference against this reference. Note that the difference for D3 from the weak R72 to the much darker 87C was –2.7 EV, but only –1 EV for the modified D200.

If you still insist on using the D3 for IR photography, an obviously unwise project but hey it’s your camera and you can use it at your own will, be warned that LiveView won’t give you much in terms of a live IR black/white view of the subject. In fact, all you see is a black void. This is in sharp contrast to the Fujifilm S3 Pro UVIR Limited Edition, with which LiveView functions perfectly in IR no matter what IR filtration is added to the lens.

You should not do IR with D3. Simple message, simple remedy: Just don't do it. There are so many more satisfying approaches to digital IR than those available with D3. On the other hand, the low sensitiivty to IR implies you get less problems with skin tones, which are unduly susceptible to excessive IR.

Nikon D3 Reviewed