Infrared Photography With A Pentax Digital SLR – Part 1

Richard Gadd tells us about infrared photography with a Pentax DSLR.

29/09/2011 - 14:21

In this article I will attempt to explain how to get started with infrared photography using a Pentax SLR, with some background about what infrared light actually is. A later article will look at using a modified camera, and image processing to make the most of the effect.

Infrared – what is it?
Usually abbreviated to IR, infrared light is simply some wavelengths of light that humans can't see. For photography we are actually using the 'near-infrared' band which would be immediately adjacent to red in the rainbow if we could see it. It can be focussed by your lens and it is seen by the digital sensor, so you can use it to take pictures. This is not the far-infrared visible via specialised thermal imaging systems, that's a much longer wavelength.

The camera manufacturers have gone to a lot of trouble to stop the IR from getting through to the sensor, so the results will vary with different generations of camera body as those measures to stop it have improved.

The individual photo-sites that make up a digital sensor have tiny red, green and blue filters over them so that the sensor can determine the amount of different colours making up the image, the colour of each pixel comes from processing the signal from these photo-sites. Those little filters don't cut the IR, so the manufacturer puts a separate glass filter in front of the sensor to filter out the IR. That filter is the face of the sensor that you actually see when you open the shutter and look inside, or when you attempt to get dust off it.

The earliest generations of digital camera bodies were not particularly good at filtering out the IR, while later generations improved the filtering to ensure that naturally-occurring IR in the scene didn't detract from the accuracy of the colours in the image. This progress is to the detriment of IR photography since it means that trying to use IR involves much longer exposure times with the most recent bodies. The recent bodies have other advantages such as shake reduction, higher resolution and more comprehensive program modes, but in many ways the older bodies such as the *ist series are better suited to IR photography with an unmodified camera.

The Hospital of St Cross, 18-55mm kit lens at 18mm, 720nm internal filter.

The pictures below show the filters. The first one is taken in visible light on a normal camera, and shows an 830nm IR-passing filter at the top and an IR-blocking filter that normally lives in front of the sensor at the bottom. The top one blocks ALL visible light and is used for IR photography.

Taken on a K5, no filter.

The second picture is the exact same view but taken in IR. Now the 830nm filter at the top is clear, while the IR-blocking filter at the bottom is completely black.

Taken on a modified GX1L with 830nm internal filter.

Why do it at all?
Since we can't see IR we don't see the way the world looks at those wavelengths. Many common things in our world do actually look very different in IR. For example, vegetation absorbs visible light of different wavelengths but is predominantly 'green' to our eyes. The chlorophyll in living plants also reflects large amounts of IR, so with a suitable camera the effect is that grass, trees and other green vegetation 'glows' quite brightly. Sunshine contains a lot of IR, but the light from the blue sky has almost none so skies tend to be very dark with strongly-accentuated clouds. Photographically the effect is dramatic, grassy landscapes can take on the simpler visual effect of snow while sunlit trees and strange clouds all contribute to a surreal, dreamlike treatment.

Here are some comparison shots taken in visible light on a K5 and then a few seconds later in IR. These are not intended to be artistic, but to show the differences:

Note how the foliage is bright in the IR shot and the blue sky is dark, giving much more contrast to the clouds.
Again the clouds are accentuated and the foliage is lighter than would be achieved by converting the original colour shot to monochrome. The shadows are stronger because they are lit only by the blue sky which has almost no IR, so the gate in the first one and the railings in the second one take on a different look which can be useful in composition.
These are in the shade, the headstone is dark while the ivy is bright.
The Military Cemetery at Royal Victoria Hospital, Sigma 10-20mm at 10mm, 720nm internal filter.

IR photography with a DSLR – what are the options?
There are really two approaches to using a DSLR for IR photography. One involves using a normal camera with a filter on the lens to block out the visible light, while the other involves modifying a camera to change the IR-blocking filter. This article looks at using an unmodified camera, a later article will examine the more serious approach of using a modified camera.

IR with an unmodified camera
Your camera body and lens can see IR, but there is an IR-blocking filter over the sensor. This doesn't block all IR, but it does stop a lot of it depending on the model of camera. To use IR to take a photograph you need to block the visible light with an IR-passing filter, since otherwise the visible light will completely swamp any IR getting through. With such a filter on the lens the light which is passed through is predominantly IR so can be used to make an image.

Examples of such filters are the Hoya R72 or the Tiffen 87. The 72 indicates that the filter passes light with a wavelength longer than 720nm (nanometres) and likewise the 87 indicates 870nm. Those wavelengths are the cutoff point, the longer the indicated wavelength the more any residual visible light is blocked. The Hoya R72 gives a weaker effect than a filter with a longer wavelength cutoff.

The Military Cemetery at Royal Victoria Hospital, Tamron 17-50mm, 830nm internal filter.

Composing the picture
The first thing you'll notice is that you can't see through the viewfinder to compose the picture. The human eye can't see the IR, and there isn't much visible light getting through. The Hoya R72 lets enough red light through so that sometimes it is possible to get a vague idea of the scene through the viewfinder, but only in very bright conditions and once the eye is adapted to peering into a dark viewfinder. Stronger filters don't give any visible light at all. So that's the first disadvantage of trying to use an unmodified DSLR for IR photography.

Exposure
The second disadvantage is that with most Pentax camera bodies you will probably need to use a tripod to get suitable exposures. In very bright sunshine an older model such as the *ist DS using the Hoya R72 filter will require exposures such as 1/20th of a second as ISO 200. You can hand-hold the older bodies, but the composition of the shot is still largely guesswork due to the black viewfinder. A current model such as the K-5 has a much stronger IR-blocking filter built-in, so in the same lighting it might require exposures of several seconds in spite of the higher ISO that it can use.. Models such as the K10D are somewhere in between. Using a tripod allows the scene to be composed and then the filter attached to the lens before taking the shot. Although cumbersome this allows longer exposures with the lens stopped down more to achieve more accurate focus.

Gate at Danebury Ring, hand-held exposure with kit 18-55 and external Hoya R72 on unmodified GX1L.

The camera's own metering can often give acceptable results with the filter on the lens, especially with the older models. It is often the case that IR pictures involve landscapes and are frequently in bright sun to exploit the 'IR look' of foliage, so a few trial shots with reference to the histogram will usually indicate a suitable exposure and then setting that up in Manual mode is likely to be a good approach. Shooting RAW is best simply because it provides more scope for exposure adjustment afterwards.

Focussing
The focus point of IR is slightly different from visible light. On old lenses you may find a red IR focus mark (or a curved line on a zoom lens). This is provided so that on a manual focus lens you focus in visible light. Let's say for a landscape you are at the infinity mark, and then to achieve correct IR focus for the same scene you turn the focus ring to align the infinity mark with the red IR mark. The lens manufacturer designed the lens to bring all the colours of light to a sharp focus, but IR wavelengths aren't corrected nearly as well. The IR mark is the manufacturer's own measure of how to adjust the focus to compensate. Unfortunately most modern lenses don't have this marking, or even distance markings on the focus ring in some cases.

Autofocus works in IR, with varying degrees of success. The phase AF on older bodies will be seeing mostly IR, and will generally bring the lens to an accurate focus based on the light it sees. In theory the contrast AF via LiveView on recent bodies should be even better, but on the K5 at least there isn't enough light reaching the sensor for LiveView to work well, it doesn't achieve focus at all. The LiveView image is very dark and hard to see with an R72 filter on the lens, but if it is visible enough then manual focussing based on the LiveView image should give good results. This might herald a return to having a black cloth over the photographer and the camera body, although I've not tried that.

Given that many IR images are landscapes the other method of focussing is simply to determine by experiment that perfect focus is when the focus ring is on (say) 2 metres and use that. A piece of masking tape on the ring could be used to indicate infinity focus and some nearer distances, where you can establish the correct positions for those marks by taking a series of shots at different marked distances. This is effectively your own focussing scale for when the IR filter is in place. You only have to do this calibration once for a given lens. Do not be surprised if the lens is apparently focussing quite close on the scale, this is due to the IR focus being different.

Autofocus on many bodies can be adjusted via the debug menu, although that is beyond the scope of this article.

It is worth setting the White Balance manually using sunlit grass. Even if you are using RAW and will make adjustments later it ensures that the thumbnail looks reasonably correct.

Weston Shore, Sigma 10-20mm at 10mm, 720nm internal filter.

Important note – never ever use an IR-passing filter to look at the Sun, since even though the image will be very dim the IR will still be strong and you risk damaging your eyes.

Aperture
Most lenses were never designed to work with IR. As mentioned above the focus point is different, but the corrections to make a sharp image are also optimised for visible light. The designer is unlikely to have corrected for field curvature and other aberrations down as far as IR wavelengths so in practice it turns out that some lenses are better than others when used for IR photography. For this reason it is generally a good idea to stop the lens down two or three stops in order to improve the image quality and cover small discrepancies in focus. Since IR is a longer wavelength it tends to show diffraction effects at earlier apertures than visible light, so stopping down too far is not useful.

Lenses
Having tried different lenses in IR there are some points to note.

The first surprise is that the humble mark one 18-55mm kit lens works quite well in IR, particularly with the 720nm filter. It is easy to establish correct focus by experiment, and even at the wide end it gives good sharpness across most of the frame. Some zoom lenses give a visible central hotspot, but this one gives even exposure across the frame at all apertures. Certainly a very good lens to start with. It works well with AF once that is calibrated, and any false colour tends to be even across the frame

The Hospital of St Cross, kit 18-55mm at 18mm, 720nm internal filter.

The Sigma 10-20mm f/4-5.6 can give good results provided the focus is correct. When used with the Hoya R72 filter the lens tends to give two slightly different planes of sharp focus, one being the from the deep red light and the other being the shifted IR light. This in turn means that off-centre the image can look very poor indeed if the wrong focus is being used in the centre. Assuming that the focus is correct the lens can give sharp results across the frame at 10mm with no hotspot.

The AF on the bodies I tried does not give accurate results for this lens, since it tends to choose the focus point for the residual visible light. In practice leaving it focussed on one particular distance for landscape use (1.1 metres on mine) gave excellent results. Stopping down beyond f/8 gives less sharpness due to diffraction.

The Chapel at Royal Victoria Country Park, Tamron 17-50mm, internal 830mm filter.

The Tamron 17-50mm f/2.8 gives mixed results. Once correct focus is achieved it gives extremely sharp results across the frame even at f/2.8. However, stopping down progressively reveals a large central hotspot due to some internally-reflected IR, possibly a reflection of the sensor itself. I have used it at f/2.8 and f/4 with good results, beyond that the hotspot tends to be too bad. This lens also shows a colour shift from centre to edge so is not particularly suited to IR shots that keep some of the false colour.

The kit 50-200mm gives good clean results, but I have not used it other than out of curiosity.
The Samyang 85mm f/1.4 gives excellent results but is a challenge to focus correctly. On a converted body it can actually be used at f/1.4 with correct focus via the viewfinder, but it is not an ideal lens to try to use with AF or on an unmodified camera.

Personally I find that I favour wide-angle views in IR, since one of the picture elements is frequently the clouds. On an unconverted body with the R72 filter the kit 18-55 gives great results, and the Sigma 10-20 at 10mm is very effective once you have determined the correct (fixed) focus.

False colour effects
One option I haven't explored yet in this article is false colour. This occurs when using the 720nm filters, since their cutoff allows some dark red to get through as well. Since the RGB filters on the photosites on the sensor react differently to different wavelengths the sensor sees some spurious colour information which it turns into colours in the image. Here are two examples where instead of converting to monochrome I've increased the saturation to show the colours:

Winnall Moors country Park, kit 18-55mm at 18mm, 720nm filter, false colour accentuated.

 

Monument on Farley Mount, Sigma 10-20mm at 10mm, 720nm filter, false colour accentuated.

IR with unmodified cameras – a summary
Suitable bodies are readily available, many people have older camera bodies that will be effective in IR with a reasonably-priced Hoya R72 filter or the Ebay equivalent.

Can sometimes be hand-held but normally needs a tripod, so no action shots. Foliage blowing in the wind will be blurred.

No viewfinder image for composition except sometimes in extreme sun, continual swapping of the filter is fiddly.

The camera remains useful for normal pictures.

Hospital of St Cross, kit 18-55mm at 18mm, internal 720nm filter.

---