One-Image Aerochrome with Green and Orange filter

[Stefano]

Yes, that's very nice. The colors look correct, nice reds, nice sky.

 

In the last video there is a "bleib cool" sign, which appears red but on the magenta side. Is it dark green in visible light? I would expect that as there is a lot of red (actually infrared), no green (actually red) and some blue (actually green).

 

Can you test a rainbow/diffraction grating? I think this kind of test is necessary to verify whether you are doing things right or not if you do multispectral photography such as IRG or GBUV (what we usually call BUG here).

[Christoph]

Thank you for the motivating words. I like it too, especially since the steps are simple and repeatable. Post-processing involves only the channel mixer for this result. As for the "correctness" of the colors. To be honest, I've strayed a bit from the path to "one hundred percent" authenticity. The reason is simply, first, that I like it so much and better than previous Aerochrome emulation attempts (don't want to badmouth those of course, it's a matter of taste) and, second, Aerochrome didn't have ONE look, but looked quite different depending on filtration, film condition and processing. The images are likely to get IR contamination. Still, to answer your question, yes the sign is green-you can see the VIS image of the scene on my blog (linked that somewhere above). Red signs and taillights etc turn yellow and human skin pale-yellowish...

[Stefano]

Thanks. The green sign indicates you are quite close to scientific "correctness", even if this isn't your aim. One day I want to do what you did.

[Christoph]

Thanks. The green sign indicates you are quite close to scientific "correctness", even if this isn't your aim. One day I want to do what you did.

 

You only need a X1 and a yellow orange filter :-) ... and some tool to do the channel mixer... I'd guess that all types of strong yellow and orange filters (15, 16 range) would work.

[Christoph]

Some more footage with different "subjects of interest" like red signs, flowers, taillights etc... again, only channel mixing, no additional processing...

 

https://vimeo.com/513426398

[colinbm]

Very nice Christoph.

One good thing with the video, is the wind doesn't spoil the shot.

[Andy Perrin]

Some more footage with different "subjects of interest" like red signs, flowers, taillights etc... again, only channel mixing, no additional processing...

 

https://vimeo.com/513426398

I don’t know what “additional processing” you think would be involved! Literally all methods are just that.

[Christoph]

I don’t know what “additional processing” you think would be involved! Literally all methods are just that.

 

Well, if you take one of the common filters (550 or IR Chrome) to get Aerochrome-like images, then the Channel mixer won't do... And the colors don't match anyway... The point is that orange-green combo works quite well to recreate Aerochrome, even with video....

[Andy Perrin]

I see, yeah, I guess I don't consider the IR Chrome to even be the same thing since you don't get the right colors except for the red trees.

[Christoph]

I had to share these (orange-green combi plus polarizer)... finally, the pink tones came through...

 

[nfoto]

If the Infrared Ektachrome is the yardstick, I'd say the images come out with far too much red, and the reds in addition lack definition and variation. Somewhat similar to the late EIR film processed in E-6, which was not much to my liking at all.

 

The conifer-dominated forests in the background do, however, have a good rendition in terms of IE expectation, and of course the sky is blue as it should be. Thus these objects are well covered.

[Christoph]

Thanks for the feedback!

[colinbm]

These are looking really good Christoph

[Cadmium]

I like the second and third ones.

[Christoph]

Ever wondered how toads look like in IR/"digital Aerochrome"?

 

[colinbm]

Are they blushing because you have found them in a compromising position ?

[Christoph]

Yeah

[Christoph]

The issue of IR contamination of the red and green channels has been discussed here repeatedly. The fact is that it significantly affects the colors of a photo and interferes with achieving an Aerochrome-like look. For example, it causes vegetation to appear paler and the color changes (red -> yellow, car taillights for example) to be off. I now made an experiment with the green-orange combo to see how much there R and G are "bothered" with IR light and how this can be turned off. To start with: The way to eliminate the contamination coincides well with my previous post-processing. The following picture should illustrate this.

 

Used were: Iphone flashlight, two IR filters (Hoya 720nm and B+W 830nm) stacked, the full spectrum Nikon D7100, and an older Nikon 50mm f/2, green filter (Hoya X1) and orange filter (Tiffen 16), Kolari Hot Mirror.

 

Standard setting in-camera, white balance from white card (in post).

 

 

Top row.

Left: Visible light

Right: Orange-Green (without any changes in PS Raw converter except white balance), all RGB channels together.

 

Middle row: Orange-Green (without any changes in PS Raw converter except white balance), RGB channels split. You can see that IR is in all channels, strongest in B.

 

Bottom row: Orange-Green, RGB channels split. Now we tried to find out where the threshold is, i.e. at which adjustments the IR from R and G disappears. The following values were determined: In the Raw Converter -30 at the "blues" in the color mixer plus subsequently in the channel mixer at G (before the channel swap): -30 at blue, compensated in red, so there +30. Now IR can only be found in the blue channel.

 

Conclusion: I could not remove the IR contamination in G in the raw converter, so the channel mixer was used here. The IR contamination in R, on the other hand, can be eliminated with a value of -30 in the blue tones. The result partly coincides well with my earlier (by naked eye) calculated, suitable values for a good Aerochrome look, above all the -30 for blue. Maybe there are ways to get the values right in the raw converter. Further tests would be necessary.

 

But be careful: other cameras, lenses, filters can of course influence the threshold. But this method can be used for appropriate calibration.

 

Processed with the values listed above:

 

 

 

The same method can be done with the TB550/660/850 (which is discussed here: Ultraviolet Photography). There the threshold is higher: -50 instead of -30 in each case. And the result is less good for my taste, because the red tones are paler:

 

[Christoph]

I love moss-covered trees. There is a place near where I live, under a gigantic rock wall, where the sun shines only a few hours. The ground and trees there are covered with moss. I always had trouble teasing IR light out of these plants. I tried everything: 720nm, 590nm and so on. Never did I get satisfactory colors. With the orange-green combo, it's different...

 

 

[OlDoinyo]

If you have any of the TB files SOOC, and care to post them, I could see what the Pixelbender workup might make of them.

[Christoph]

I can post or send you raw files from the Triple Bandpass filter... I don‘t have access to them right now but can do it tomorrow. There‘s also a Raw attachement in the specific TB thread. To be honest: I don‘t like the Pixelbender results. Colors kind of of look monotone to me... not Aerochrome-y... For me: Orange-Green with minimal PS work does the job perfectly, even better than TB...

[Christoph]

If you have any of the TB files SOOC, and care to post them, I could see what the Pixelbender workup might make of them.

 

Here you go: _DSC0021.NEF

If you want more of the TB raw files for experimenting, send me a PM...

[Christoph]

Another image from yesterday... the color variation is pretty staggering... the only difference to the previous images: "landscape" instead of "neutral" preset and a little boost in contrast...

 

[colinbm]

I really like this one Christoph

[nfoto]

When your "one-trick-pony" solution can deliver something like this, I'll buy into it ...

 

 

Fuji S5Pro with 85mm f/1.8 Nikkor.