Saturday, 31 July 2010

Minus blue and back focus

In preparation for the session at IBC in September I have been doing some experimentation.

Firstly, with the assistance of friends at the BBC Natural History Unit in Bristol, I got my hands on a full-frame Canon DSLR which had had its infrared blocking filter removed. Unfortunately the weather denied me the sunshine that really gives a punch to the Wood Effect but the Clifton Suspension Bridge is always impressive.

One thing I was keen to try was using a minus-blue filter (Wratten #12) to emulate Infrared Ektachrome. Despite the cloudy day the result was promising.

Colour infrared Clifton Suspension Bridge

You can see the first stab here. After some channel equalisation/normalisation this image uses the Photoshop channel mixer to map the camera's blue channel to red (only infrared should be getting through the #12 and the camera's blue filter), maps red to green and green to blue. Some blue (an arbitrary 50%) has been subtracted from both the green and blue output channels. This emulates the process used in Kodak's DCS 420, a version of which was a short-lived colour infrared DSLR from the mid-1990s. It looks as if this deserves further study. If you have a modified DSLR and a #12 filter you should give it a try.

My second point concerns back focus. One problem that you hit when you remove the infrared blocking filter from a digital camera is that unless you replace it with some other glass the lens will no longer focus. I was discussing this with engineers at Ikegami (while we explored modifying a broadcast HD camera for infrared work) and they explained why this happens. It was one of those Homer Simpson 'Doh!' moments. (Forgive me if you already know this.)

What I had forgotten is that the filtration in front of the sensor is actually part of the light path that focuses the lens. The light going through the filter is not parallel, it is converging to hit the sensor. So if you remove the filter (glass) you change the refractive index of that part of the light path and optically extend the lens. To shorten it again you either have to physically move the lens (difficult because you probably don't have much room between the back element and the sensor, even if you could move it) or you have to put back some glass with the same refractive properties as the filter you removed.

You have probably removed more than one filter, bonded together. The camera we were looking at had a removable filter unit which contained a high-pass filter in the frequency domain (the infrared blocker) and a low-pass filter in the spatial domain (the anti-alias filter). Fortunately, for landscape work, aliasing is not too big a problem and some DSLR owners actually remove these to get more sharpness. The actual efficacy of this is a subject hotly debated on various web forums and I won't get into it here.