The variability of IR colors
Posted 16 January 2013 - 11:09
The recent thread from Bjørn showing his selection of 52 mostly-IR shots from last year has whet my appetite to understand more.
I usually get very flat NEF files (custom white balance set on grass once for all), that need a strong massage to reach the a satisfactory (to me) color level. I usually play with Lab curves and especially narrox dow the "b" curve in order to boost colors; Sometimes the "b" histogram is fairly wide from the beginning which of course is favorable to high colors, sometimes it is so narrow that the situation is hopeless. The "a" curve is usually poorer in information. This obviously is related to the light that was available when I shot, but I have a hard time to determine the optimal conditions.
I had the impression that overcast skies were helping, but I have many shots taken under gray skies that ended up being almost monochromatic from the beginning.
It is of course always possible to shift colors to whatever dominent tone I want, but it is still monochromatic.
Without wishing that anyone reveals his post-processing know-how, I would be interesting to get some insight about the possible link between the "quality of light" when the shot was taken, and the ease of extracting rich colors from an IR raw file.
Of course some subjects lend themselves much better to color IR, but I am also convinced that conditions play a major role.
Posted 16 January 2013 - 11:33
Of course IR by definition has no colours, so "unreal" or "false" colours are what in reality we're chasing after. Whilst the definition is true in a philosophical context, it is far from true in a more practical down-to-earth setting. The camera does not know anything about IR light being without colours, the Bayer dyes respond in peculiar ways to IR bandwidths, some subjects have strong spectral peaks or dips in the IR band, and finally, software has not the slightest inkling of what IR ought to be. All of this leads us to a potentially rich colour palette in IR yet the existence of it is frustratingly elusive at time.
In my experience some of the lack of colours boils down to interplay between illumination source and subject, camera-specific response, and software quirks. Cameras do *not* respond equally to IR. In fact, there can be a difference between having the filtration done internally, over the sensor, or in front of the lens by an attached filter there. Thus, also the lens influences the IR output.
I sometimes shoot "multi-band IR" using narrow (approx 10 nm) bandpass filters within the 700 to 1000 nm range which my cameras are capable of recording. When I space the filter passbands approx. 100-125 nm apart, three out of four composites appear with dull and muted colours. A few can show an intense colour palette. So in this case it is obvious the spectral reflectivity of the subject plays a major role (I have compensated for exposure differences so basically the camera "sees" IR-white light). When I space the passbands closer, the tendency towards even response in all channels increases but once in a while one or two of the channels register very differently.
Due to subject movement, few motifs are suitable for this technique and it is also time-consuming to conduct without generating much in terms of visual values. Thus I mainly do it once in a while to satisfy my own curiosity.
Posted 16 January 2013 - 14:24
It is clear for me that we talk about false colors, I just don't seem to be able to control their apparition. Maybe it is beyond my reach, I am always trying to find simple answers to complex problems...
For me, "useless" files for IR colors ar the ones that either show the same information in the three channels, or at least in two. In the latter case, I get a color cast but it is just one color, not what I am after.
I have had little success by playing with the white balance as when it is really off (but what does that mean for IR?) it simply get a picture strongly dominated by one color.
So things start to be interesting when a little bit of several colors exist, and that I manage to have them grow...
So right now I am like an uneducated prehistoric man: I know how to nourish the fire, what I really want is to start it anytime!
I will continue to experiment and try to pay more attention to the conditions. I suppose that using a camera with liveview (or even better with a EVF) would be helpful.
Edited by olivier, 16 January 2013 - 14:27 .
Posted 16 January 2013 - 14:43
I have all my cameras for UV or IR set to the lowest colour temperature possible, meaning incandescent light. That gives me the very least an idea whether the exposure is ball-park or way off. Of course the w/b doesn't mean much when you shoot RAW and use software that ignores the in-camera setting(s).
My Panasonic GH-2 allows me to set an "IR-white" balance in-camera, which can be useful to understand the tonality of the subject. However, it is not a good starting point for coaxing IR colours to appear later. The Nikons cannot be set to IR white so I don't even bother trying with them. Same with the Fuji S3.
0 user(s) are reading this topic
0 members, 0 guests, 0 anonymous users