Canon 40D Conversion – Full Spectrum Imaging
I thought I might describe my work with my recently coverted Canon 40D camera, partly for my own records and partly in case others might want to see the pros and cons of working in this extended medium. Many of the images to which I refer and that are reprouced here are to be found in the galleries of the Infrared-Club (Extra-Visible Imaging) , particularly the Reference and Tutorial gallery.
My Canon 40D was converted by Advanced Camera Services (ACS) [link] a company in Norfolk, UK, for about £325 (including VAT). The conversion involved removing the infrared reflecting hot mirror, and replacing it with a quartz filter that allows ultra-violet light, visible light, and infrared light to reach the sensor array, the full spectrum range being about 380 nm at the UV end to about 1000 nm at the IR end, via the visible 400 – 700 nm range.
What is the point of having such a camera with this capability? Do I not get just some weird jumble of colour that falls between the extremes of pure visible and pure infrared? Yes, there is a danger that this is just a new toy that gives differently coloured shots. However, there is more to it: this is a camera system that allows exploration of extra-visible imaging in ways that infrared cameras themselves do not. Versatility is the key here, such being achieved via the use of external filters and software control, particularly white balance settings.
First Shots – Basic Effects
Probably the best way of running this Journal is to work from photographs. If we look at the image “Canon 40D Conversion 1” then we see the result of taking a shot on more or less fully automatic settings: auto exposure (AE), auto focus (AF), and auto white balance (AWB). The aperture was set at f/8.0 (Av mode), the shutter speed of 1/500th second is fast for any immediate movement in the tree branches (and passing ducks) The lens, Canon's EFS 18-55 mm kit lens, is known to be good for IR work with no discernable hot-spotting. The scene was chosen for its good light, its texture and detail, its colour with a blue sky and some green grass, and the reflections in the water. The Sun's position was over the left shoulder, thus there would be little likelihood of flaring. There has been no post-processing, the image being presented as obtained under Canon's “Landscape” settings in Digital Photo Professional (DPP).
My immediate impression of this image was that it is over-exposed, and looking at the histogram (appended via snipping and then layer editing in the Gimp) we can see that the red channel is well clipped, with some small clipping in the blue channel. To overcome this clipping I reduced the exposure by 2 stops, using the exposure compensation on the camera, and as seen from the image and the histogram of “Canon 40D Conversion 2” the clipping has been lost and all channels have significant coverage.
These observations raise two points: first, if the second exposure is taken to be a “good” exposure, then the AE function on the camera needs to be over-ridden by -2 stops. This seems reasonable in that the AE system is working on visible light, but there is IR light here as well, about 2 stops' worth, and it is that which causes the extra exposure. Secondly, we are begging the question of what is a “good exposure” with this system. Should I be aiming for exposures where there is no clipping at all, or does it really matter if the red channel is clipped so long as the others are in range? I am not sure that there is a simple, correct answer to this, as the infrared input is nearly always going to push the red channel to extremes.
Returning to the first image, but this will not be seen clearly on the uploaded images as they are short of pixels, close examination shows that the detail is very soft, indeed unacceptably soft if this were a visible only image. The softness comes from the infrared component being out-of-focus, a consequence of the longer wavelengths of infrared light. Whether this focus issue is seen as a problem, a feature, or an opportunity, I think depends on circumstance.
****APO lenses (such as my Sigma) should not be a problem [link] - need for investigation****
Neutral Density Filter
The image “Canon 40D Conversion 3” shows the result of placing a Cokin 4 stop neutral density filter over the lens. These filters are transparent to infrared radiation, but block the visible – a point clearly seen in the histogram. However, the sky is now darker with a magenta cast. This makes sense because we are removing several stops worth of blue light from the sky to leave behind the general red cast from the infrared influence on the red sensors. Nevertheless, this idea of cutting down on the visible component may be put to good use – another avenue to explore.
Blue Filter 80B/KB12
One of the most effective and startling uses of full spectrum cameras is where a filter that cuts out part of the visible spectrum is put in place. Provided the filter transmits infrared light, the result is an image from the filter colour plus infrared. LifePixel [link] describe the use of such filters including one they refer to as “SuperBlueIR”. (For a review of filters, their transmission curves and designations see [link] ) Although not stated, I think this is a blue filter with designation KB20. I have the junior version KB12 or 80B in my collection, so I tried that with the new system, and the result can be seen in “Canon 40D Conversion 4” .
The immediate result shows a much enhanced blue sky, with some blue channel clipping. As Colinbm1 [link] says in a comment, this is probably equivalent to a “cold mirror” UV+Blue – Green – Red + IR photograph. I suspect that there is a great deal of creative potential here. Some further results, not yet posted, suggest that the 80B filter can cause colour shifts and saturation enhancements – a “colour punch” effect.
Infrared images are obtained by placing a suitable infrared filter over the lens. “Canon Conversion 40D 5” shows the immediate result using a 720 nm filter on AWB and what I thought would be “best” exposure. I think this shows over-exposure, so I cut the exposure by one stop to “Canon 40D Conversion 6” , the histogram of which shows a very good red channel with almost no clipping along with green and blue channels in the shadow zone. If I was working via the AWB, this would be the image from which to derive others.
I had previously set up some aluminium foil (matte side up) in dull sunlight and had shot this with the 720 nm filter in place. I used the resulting image as a custom white balance for subsequent infrared shots. “Canon 40D Conversion 7” shows the result from “best” exposure – no clipping – at this scene.
There are several points to made about this work:
First, because the infrared filter cuts out all visible light, nothing can be seen through the viewfinder, so achieving composition and focus could be problems. However, the Canon 40D has Live View and switching this on allows both problems to be resolved.
Second, regarding focus, using the Manual Focus setting on the lens allows it to be adjusted to give best focus to the eye.
Third, a hood over the head and LCD helps when carrying out adjustments, as does using the magnification facility on Live View. Indeed, working in B/W mode on Live View is good as the contrast is much better on the LCD screen. In this way, I was able to obtain very highly focussed infrared shots – far exceeding autofocus on my Canon 20D, adapted specifically for infrared photography. Another example of achievable focus is seen in “Full Spectrum Canon 40D Sample Images 5”, a hand-held shot focused via Live View using Canon's EF 50 mm f/1.4 lens.
Hot Spots and Lenses
The EF-S 18-55 mm lens used for most of these test shots appears to give no obvious hot-spotting at f/8 or below. In fact, even at f/22 the images look acceptable. With the EF 50 mm f/1.4 lens, this too is acceptable at f/8 and below, but hot spots appear from f/11 and higher f numbers – as seen in previous work with the Canon 20D.
As with infrared photography itself, white balance is a major issue with full spectrum photography. If we simply go for a best exposure shot on AWB, such as #2 above, then we are left with a strong red cast. That may be acceptable under some creative circumstances, but most of us would want to process the image further. This is an area of continuing experiment. I have found, by trial and error, a “green” balance that renders leaves as an appropriate green hue – see “Full Spectrum Canon 40D Sample Images 1”
For other work, I have used the “grey” balance mentioned earlier, from a sheet of aluminum foil, on full spectrum images. The final image for now, “Canon 40D Conversion 8” is derived from #2 by application of this “grey” balance plus an additional stop of exposure introduced using Canon's DPP.
I welcome comments