Which filter is best for artistic infrared shots? To be honest, I hadn’t though about this for years. Having spent a small fortune on infrared films and filters and concluding the Hoya R72 (now called RM-72) worked well for my purposes, I left it at that.
And then came digital cameras. When I had my Canon Rebel XT modified for infrared use by LifePixel, I got the standard conversion which meant an infrared filter with a cutoff around 720nm. The converted camera worked great and the filter was installed permanently so it was a done deal anyway.
My Canon G1X was is a different story. It got a “dual spectrum” conversion from Kolari Vision which meant that the camera could still be used for regular photography(!) but to work this magic it required a filter in the lens at all times and the two filters had to be swapped, depending on whether I was shooting “normal” or infrared. For infrared photos, I used a Hoya R72, a 720nm cutoff filter, and not surprisingly the results were similar to the Rebel XT with its built-in 720nm filter. So, for practical purposes, the only difference between the two infrared-modified cameras was that on the G1X, the infrared filter was in front of the lens and on the Rebel SLR, the infrared filter was permanently installed in the camera.
If I’ve already confused you, I apologize. At some point I’ll write an article that starts at the beginning and covers the basics. But there is so much of that on the web you needn’t wait for me to get to it.
The way this test came about is that the article I did for Photo Techniques magazine generated a bunch of email saying “I can’t get a blue sky” or “When I adjust the sky to look blue, everything in the photo turns blue”. There were too many variables for a one size fits all solution. But I remembered seeing photos on Flickr that had very blue skies. Very blue. Many of these super blue photos were taken with red filters such as a #25, an ordinary “medium red” filter, something every hobbyist had when we shot black and white film. Even the seemingly exotic B+W 091 “goldie” filter (so called because it admits enough visible light to allow yellowish/reddish foliage in infrared photos) was just a red filter, somewhere between a #25 medium red and a #29 dark red. The reason the goldie filter is an inexpensive infrared filter is because it’s not an infrared filter. But if you want really blue skies it may be a better choice, as we shall see.
All this made me curious and I realized my dual spectrum G1X infrared camera could be used to compare filters simply by putting them in front of the lens. Now, there was flash of inspiration!
Before we go any further, two things need explanation. First, these tests are only relevant to cameras that have been modified for infrared use by having their infrared-blocking “hot mirror” removed. Using an un-modified camera with a red filter–light, dark, whatever–will result in a somewhat darker sky (assuming it was blue to begin with) but you won’t come close to the almost white foliage you see in an infrared photo. Using a true infrared filter on an un-modified camera actually gives pretty good infrared results but exposures range from long to eternal, raising other issues like noise. Still, if you have a tripod and an stationary subject, it’s certainly worth a try.
Second, sharp eyed readers will notice the filters I used are of different sizes. That’s because I accumulated them over many years and for most of that time I used 49mm, 52mm or 55mm. Alas, the G1X takes a 58mm and I was not about to buy these filters again, so I used an assortment of adapters to hold them in front of the lens. Crude, I know.
Here’s what I did:
(1) I took two photos with each of these filters, first with the camera’s white balance calibrated to some green foliage, and second with the white balance calibrated to a a gray card which some people say gives better results.
(2) I opened the images in Photoshop. Depending on the filter used, the contrast of the images varied, but as expected, all were flat (low in contrast). To make things as fair as possible, I used the “auto” button in the levels adjustment in Photoshop so each image was set to “average” contrast. The results at this point were not ideal but at least it was a level playing field.
(3) Using Photoshop’s channel mixer, I swapped the red and blue channels. (If you’re not familiar, the red channel, by default is 100% red and 0% blue so you set it to 0% red and 100% blue. Then you move to the blue channel which is by default 100% blue and 0% red, and you switch these to 100% red and 0% blue).
This is where things got complicated. Almost all the photos had skies that were either to dark or too cyan (blue-green). So:
(4) For the skies, I made them lighter if needed and where necessary, shifted the sky color from cyan to blue. (I did this using the blue hue slider in Lightroom but I could just as easily done it in Photoshop). Since I was only adjusting the blue hue it took just a second and did not involve any selections, paths, masks, etc. Hate that stuff (subject for another post).
(5) Finally, there was the foliage. I wanted it to be as brilliant white as possible but without losing too much detail. If you’re ever taken photos of snow in sunshine, you know there’s a sweet spot between dingy gray snow and brilliant, blank white that has no detail. So, to the extent possible, I lightened the foliage while trying to keep some detail. (Lightening the foliage also had the advantage of reducing color casts in the foliage; almost like bleaching).
So here are the results with my comments. Unscientific? You bet. But given the variables, such as different camera brands and sensors–and my not knowing what you want your infrared photos to look like, I offer this as a starting point. Beyond that, to each his own. We’re not aiming for realism here so whatever looks good to you is “right”.
What can be learned from this comparison? One thing is certain; my yard is in desperate need of landscaping. Beyond that, things are less concrete, but we can make some observations.
If you want to take infrared photos but your camera has not been modified for infrared use, use a true infrared filter in the 650 nm range, such as the B+W 092 or (my recommendation)–or–in the 720 nm range like the Hoya R72 (aka RM72). Exposures will be long with either filter but lighter filters such as red #23, 25 & #29 will not give a true infrared result and at the other extreme, infrared filters that block all visible light, like Hoya’s RM-90 or B+W 093 will act like a costly lens cap on an unmodified camera–you won’t get anything.
If you want to take infrared photos with a camera that has been modified to record infrared, you have 3 options, depending on the look you are trying to achieve.
Option #1: If you want a really blue sky and don’t mind having some blue seep into the rest of the photo, I’d go with a dark red filter such as a Tiffen #29 or a B+W 091. Set your camera’s white balance to a gray card or white sheet of paper to get a very blue sky–or–white balance to a green subject like grass for an ultra, super blue sky. These dark red filters have other advantages as well. They are less expensive than infrared filters and they can also be used to get foliage that looks gold (sort of.)
Option #2: If you want a dark blue sky but with minimal blue color in the foliage or elsewhere in the picture, use a Hoya R72 (RM-72) or a B+W 092. Set the camera’s white balance to a gray card or white sheet of paper.
Option #3: If you want black and white results only, the most dramatic results will be achieved with the 950-1000 nm infrared filters, such as the Hoya RM-90 or B+W 093. No need to set a custom white balance for these–they’re infrared only.
Note that any of the above options can be used to get a black and white final result; it’s just that with option #3, black and white is the only option.
I plan a few more posts on this but there is a wealth of information on the websites of companies that do infrared conversions.