Transient Eye is five years old today (!), and so to revisit the theme of the very first post we have a selection of infrared images taken recently while hiking in the Italian Alps.
[Note: for a version of this post that is easier to read on small screen devices, click here.]
These photographs were shot handheld using the M10 and a 35mm Summilux lens that was fitted with a Hoya R72 filter to block most of the visible light below 720nm.
Leica cameras have a notorious history for being rather sensitive to infrared, where the need for extremely thin sensor cover glass made it difficult to include effective IR filtration. Unfortunately for infrared fans the latest cameras have largely solved this problem and using the R72 with the M10 gives exposure times not unlike those from shooting with a 10 stop neutral density filter. Great if you like long exposures but not so great if you do not want to hike with the added weight of a tripod!
Fortunately, the Summilux performs admirably with infrared light at all apertures and has no significant visible hot spot (surprisingly, the same was also true for the 21mm Super-Elmar and 90mm Elmarit-M lenses that I tested on this trip). As a result it was possible to shoot these images hand-held at ISO800 – 1600 with the lens wide-open – not exactly a classic landscape photography setup, but certainly able to yield some interesting results. The vignetting seen in these images is, for example, the direct result of the lens+filter and was not added during post processing.
The original intention was to convert the images to black and white, but I really liked the deep red colour of the images straight from camera and so processed them accordingly. The catch is that they are obviously dimmer than full RGB images, and you need a high quality calibrated display to render them with colour and contrast as intended.
Another reason why I liked the red was because of the implied connection to heat and environmental warming. Late summer hikes in the Alps seem to be increasingly prone to heavy rainstorms, and high altitude glaciers are shrinking as the winter snows are now longer sufficient to replenish the ice.
I had hoped to catch some clear blue skies which would render almost as black in these images, but when hiking we experienced mostly extensive cloud cover and rain. While it is impossible to attribute any of the weather on this specific trip directly to climate change, it is clear that the average weather that we encounter now is noticeably different to that even a mere ten years ago. This year in particular has seen an unprecedented number of regional heat problems, even if the mean global temperature is not itself a record.
Focusing with infrared and a rangefinder is at best a guessing game. Infrared light is more strongly refracted by the lens elements and the optimum focus adjustment is much closer to the camera than is indicated by the visible light in the rangefinder window. Leica do not put IR focusing marks on their lenses, so if you shoot film you will need to rely on stopping down sufficiently to obtain enough depth of field to cover any focus error.
Fortunately, the M10 supports live-view and so these these photographs were focussed by using a Visoflex 002 accessory finder with focus peaking. This worked very well, particularly with the peaking highlights rendered in blue to contrast against the red images.
The images were shot in RAW format and processed using a combination of Capture One and Affinity Photo. The workflow was fairly straightforward:
- import the RAW files in to Capture One
- convert to black and white
- edit the image to taste
- pass the image to Affinity Photo (right-click in Capture One and then select “Edit With”)
- in Affinity Photo reduce the green and blue intensity to zero via a channel mixer layer
The reason for using Affinity Photo rather than Capture One’s built in levels or curves tools to render the image red is largely because with black-and-white enabled in Capture One there is no means to completely eliminate the green and blue colour channels. It is possible to process the image in colour this way, but I find the editing process with black-and-white more intuitive.
Although these were deliberately rendered to black-and-red, there are many other possible ways to process these images.
The most popular form of colour processing for IR images is quite straightforward:
- set the white balance using a known neutral grey area in the image (this produces a very extreme white balance adjustment)
- increase saturation, to boost the weak green and blue channels
- swap the red and blue colour channels (it is important to do this only after adjusting the white balance)
- edit to taste
The swapping of the green-blue channels produces a deliberately false colour image, but one where the sky typically remains a more conventional blue colour. The following slide-show illustrates some of the possibilities for a single image, including a comparison with a non-IR image:
The infrared images have poor detail compared to equivalent shots made without the filter, the main reasons being that the optical system is far from optimised for IR photography and also because only 1 in 4 of the pixels on the sensor are working at a sensible signal to noise level. These shots were exposed to give an optimum red-channel signal without clipping, and the R72 filter’s attenuation means that there was very little signal left in the green and blue channels.
The extreme corrections applied for white balance and saturation used in the slideshow above amplify any imperfections, and false-colour images single shots at base ISO on the M10 are difficult to process. The only real solution is to shoot on a tripod and bracket pairs of shots – one exposed correctly for the red channel and the other exposed correctly for the blue-green channels. These can then be combined and processed as required.
The following image shows an example of this, combining the green-blue channels from an unfiltered image with the red channel from an R72 filtered image:
Processing images such in this way is difficult. In the above image there was no correction for vignetting in the IR channel, so the foliage in the lower left should appear much redder than it does. Furthermore, the lens has a slightly different effective focal length in infrared than in visible light, and it was difficult to obtain perfect alignment of the two images.
Hybrid images such as this (when shot and processed more carefully!) can be useful scientific visualisation tools where, for example, the infrared channel provides information about the characteristics of plant cover. But for non-scientific use, it is definitely easier to stick with the plain red-on-black processing of single images.
I strongly suspect that a camera with a monochrome sensor would give significantly better results. It would be fascinating to run a comparison with one of Leica’s monochrome cameras that omit the Bayer CFA. Colour composites can be made by combining images taken with different colour filters fitted to the lens.