14 replies to this topic

[OlDoinyo]

There has been increasing interest recently in photographic techniques that render foliage in some shade of red rather than the more familiar green. Unquestionably, this has been inspired by the existence (and more recent demise) of certain Kodak infrared films which did this as a consequence of their engineering. A number of techniques have been devised which do this to some extent or another, some involving infrared light and others not. It is beyond the scope of this post to describe all of these, but one of the oldest of them is to construct an image using the same cross-sampling scheme as the old films, termed an IRG image. There are, in turn, two ways of going about this: combining information from separate infrared and visible images of the same scene (the two-exposure method) or taking a single exposure through a sharp-cutoff filter, isolating the infrared signal in the "blue" channel, and performing a combination of channel transposition and signal subtraction to construct the final image (the single-exposure method.) There are drawbacks to either approach--the two-exposure method is precise and conceptually simple, but often unwieldy and somewhat limiting in practice. The single-exposure method is more complex and more approximate, but much easier to deal with at the camera stage.

A rival paradigm gaining some currency of late is the IGB image, which can be constructed by the above two-exposure method (and will be below) but in practice is more often obtained with a converted camera and a special filter or stack of filters and a single exposure. Conceptually, this involves replacing the red channel of an ordinary visible image with an infrared exposure. Advocates of this method point out that the image looks as it does straight out of the camera, without the need of extensive processing, and that it lends itself more easily to video use than the other..

These two image types have both similarities and differences. To illustrate, a test scene was chosen which contained:

-foliage;

-a car taillight;

-red and green LED signage; and

-red painted surfaces.

Here are the constituent RGB and infrared images used in this comparison:



And here is the 2-exposure IRG frame of the scene as constructed, with minor post-processing:



In addition to the reddish-pink foliage color, one can see other signature characteristics of an IRG image. The red painted surfaces and car taillight now show yellow, due to the reflection of both red and infrared. The red LEDs now appear green(ish) and the green LEDs appear blue. The sky is somewhat obscured by chronochromic fringing due to cloud movement, but the top shows the characteristic deeper blue color, less green than in an RGB image.

Here is the two-exposure IGB image of the same scene:



As initially constructed, due to the large amount of green light coming off the foliage, the vegetation appears orange rather than red, and the sky is more cyan than in the other images. Substantial hue adjustments were necessary to make the image more similar to the other. I do not own the relevant filters myself, but it is interesting to note that some users of filter stacks have also reported vegetation coming out more orange than expected. In passing we note that the red painted surfaces and taillight remain red and the green LEDs remain green. The red LEDs are blacked out.

It is not the aim of this post to claim that one image type is inherently superior to the other; artistic taste is a matter of opinion and there is no right or wrong here. Rather, it is hoped that the difference between the image types will have been clearly enough elucidated that informed choices are made possible.

[Cadmium]

OIDoinyo, Nice comparison. :-)

[dabateman]

Wow $2.78 for gas is all I can say
We have it higher than that here. I remember when I first moved to Maryland and it was over $5 a gallon.

I will gave to remember this simpler trick. I have been trying to work out the best in camera single filter, single shot method. The one touch white balance setting seems to be the most critical element.

[nfoto]

Yes, a nicely done and instructive example.

My gripe against the current spate of "Aerochrome" images is mainly they present vegetation in very garish red colour without any of the rendition subtility and delicacy found in the old E-4 processed IE Infrared Ektachrome. The effect is almost like taking a spray-paint can to the entire image.

[Christoph]

OlDoinyo, on 08 April 2021 - 03:46, said:

It is not the aim of this post to claim that one image type is inherently superior to the other; artistic taste is a matter of opinion and there is no right or wrong here. Rather, it is hoped that the difference between the image types will have been clearly enough elucidated that informed choices are made possible.

Interesting comparison. What is the perfect digital Aerochrome imitation is probably in the eye of the beholder. The two-image IRG method seems to be the closest to the original idea, but I don't find the results ideal. The "punch" is missing, and I find the cyan discoloration of the clouds quite ugly. What do you think of the X1 and Orange combo? Have you ever tried that?

[Andrea B.]

OlDoinyo, thank you for this excellent tutorial for two ways of digitally constructing an IR film emulation which "matches". I agree that the 2-exposure method most closely emulates the old IR film look.

I don't find the 2-exposure method too unwieldy after practicing this methodology for years now when making my botanical photographs in Vis, UV and IR. It is only on windy days that matching up the photos can fail. But a dollop of patience while waiting for wind lulls can usually bring home matching exposures.
*****

Christoph, as Clark wrote, it is all a matter of personal taste.

To me, the pink and cyan version which matches the film has a certain funky, artistic "look" which is evocative of decades past. And the way greens and yellows appear is quite distinctive. It seems to have a very 1950s vibe. It would be interesting to photograph some old 1950 cars using this emulation.

[StephanN]

Just wondering how the IRChrome-filter fits into the picture: https://kolarivision...dak-aerochrome/ . If the website is to be trusted, it's pretty close to Aerochrome with one exposure, without any tinkering with channels etc.

[Andrea B.]

The IR chrome seems not to produce the yellow where it should? But I have not seen enough examples to make that a definite statement.

[StephanN]

In a few of the photos on the website, there's a bit of yellow, but I wouldn't know if these are subjects that ought to be yellow.

[Andy Perrin]

IR chrome actually cannot produce yellow from objects that reflect IR+red because it's based on the idea of (R,G,B) = (IR, green, blue) where red light is deleted and replaced with IR. Hence the red light cannot contribute to the image. Since It's the combination of IR+red that makes the yellow in the 2-image method, that cannot occur with no red. You could get yellow from IR+green light in IRChrome, but that's not the same thing!

Edited by Andy Perrin, 08 April 2021 - 19:53.

[nfoto]

The original Infrared Ektachrome had colour dyes that responded IR-> R, R -> G, G -> B. So an object reflecting IR and red would be rendered in yellow. As all three dyes would respond to blue light, one had to use a strong yellow (or orange) filter to keep the incoming blue component down.

Just "deleting" red light will never give a good foundation for an emulation of infrared Ektachrome.

[OlDoinyo]

nfoto, on 08 April 2021 - 07:40, said:

Yes, a nicely done and instructive example.

My gripe against the current spate of "Aerochrome" images is mainly they present vegetation in very garish red colour without any of the rendition subtility and delicacy found in the old E-4 processed IE Infrared Ektachrome. The effect is almost like taking a spray-paint can to the entire image.

This one is for you, Dr. Rørslett!



(some Photoshop tweaking--from a single-exposure image. Original version viewable here.

Christoph, on 08 April 2021 - 10:14, said:

The "punch" is missing, and I find the cyan discoloration of the clouds quite ugly.

The cyan coloration is because the clouds moved between frames--it is not inherent in IRG images as such. I agree that these are not pretty pictures; I chose this scene merely to illustrate technical points. Had I looked around longer, I could perhaps have found something more appealing, but time was short.


Here is a hurriedly-constructed single-exposure IRG of the same scene (not properly worked up, but notice no such coloration in the sky area:)

Quote

Just wondering how the IRChrome-filter fits into the picture: https://kolarivision...dak-aerochrome/ . If the website is to be trusted, it's pretty close to Aerochrome with one exposure, without any tinkering with channels etc

Yann Phillippe, the inventor, has been somewhat coy about the exact spectrum of this filter (part of his secret sauce, I guess,) but in earlier discussions on this board he has conceded that what the filter produces is a form of IGB image. It is possible that he has also included some green suppression to cut down on the orange issue.

[Christoph]

OlDoinyo, on 09 April 2021 - 03:47, said:

The cyan coloration is because the clouds moved between frames--it is not inherent in IRG images as such. I agree that these are not pretty pictures; I chose this scene merely to illustrate technical points. Had I looked around longer, I could perhaps have found something more appealing, but time was short.

I‘ve got the cyan coloration in sky and an overall cast also in my images when I tried the 2-image-method.
I have to insist, since you‘ve been doing this for a long time :-) whats your take on my orange-green combination? You‘ll find examples in the specific thread...

Edited by Christoph, 09 April 2021 - 11:33.

[OlDoinyo]

Christoph, on 09 April 2021 - 11:31, said:

I‘ve got the cyan coloration in sky and an overall cast also in my images when I tried the 2-image-method.

With certain lighting conditions, clouds, etc. that sky color sometimes just happens. I am not convinced it is always an image fault. See here for one of my examples. Color casts (of the entire image) are easy enough to get rid of in post-processing, for the most part, unless they are extreme.

Christoph, on 09 April 2021 - 11:31, said:

whats your take on my orange-green combination?

I'm trying to get my head around what you did. At first I thought you were using your stack for a version of the two-exposure method, then it seemed as though it was a version of the single-exposure method. I threw one of your SOOC images into Pixelbender and adjusted some parameters. Other than some excessive saturation, it seemed to clean up OK, though many variations on this might be possible depending on slider settings and final workup (I desaturated red and blue to arrive at this particular image:)



So if your question is, can this filter stack be used with the single-exposure method, I would say yes, though it is hardly unique in that respect.

If your question is about your workup protocol, and whether that produces a more pleasing result than throwing it in Pixelbender and adjusting sliders, that depends on what you prefer. It sounds like your workup offers less freedom than mine for varying the nature of the final product, but if it satisfies you and seems easier, by all means use it.

[Christoph]

Thank you! Yes, the whole idea is to get an authentic Aerochrome look with just a few steps via PS. Just one photo. With minimal changes in the raw converter. The Hoya x1 makes that the magentas/blues (which become red after the channel swap) are much more dark and intense. The reason for this? It suppresses parts of the NIR in the 700-800nm region. I tested also with other green, even x1 filters like the one from Wratten— they won’t work — too much IR from that region, hence the magentas/blues are too bright. The Orange 16 in addition makes that the blue channel only gets IR. This makes the processing much easier. I don‘t do any additional saturating— you can choose neutral in-camera preset and get „punchy“ images.

Edited by Christoph, 11 April 2021 - 12:56.