How can I avoid or correct the "inverse Ageratum effect" in digital photography?

Question

I was used to the ageratum effect with film cameras: as the red pigment on film is much more sensitive to IR than the humman eye, some flowers (such as the ageratum) appear pink in a photograph while we perceive them as purple in nature.

Gone digital, I though that this kind of problem was over... however, I have shot some violas and vincas (in outdoor natural light with a Contax T* 60mm makroplanar on my Olympus E630) and the flowers are much more blue than I see them (they looked purple).

This is not a problem of white balance since the surrounding grass has its natural color.

Has anyone also observed this effect?

Is this a problem of the Olympus sensor or a more general effect in digital life?

What workarounds might I use — filters, post-processing, ...?

Here is an "as shot" image:

and a corrected one (as suggested in the comments I have corrected the flower color disregarding the effect on the grass which really looks as shown in the first picture):

Answer 1

White balance is only a broad-stroke correction. Depending on the actual color temperature and the spectral composition of the light, you may find a boost or a "suck-out" in some part of the spectrum even if the neutrals are rendering correctly. To get absolutely (or as close to absolutely as possible) correct color renditions, you need to profile your camera/lens combination under the light your are using. Both of the major names in color calibration tools, X-Rite and Datacolor, make profiling targets that you can use to correct your camera's color rendition across the spectrum.

Datacolor's SpyderCheckr is a new product. Much better known is the similar but older X-Rite ColorChecker Passport with its profiling software. You introduce the calibration target (a miniature Gretag MacBeth color target) into a frame in your shooting series, and then export the image to the profile creation software. Since it samples calibrated color targets across the spectrum, it can create a neutral rendition profile for that camera, lens and lighting combination that you can then apply to the rest of the images in the series.

Answer 2

It's important to understand that film and a camera sensor have many different properties. White balance can make an obvious and substantial difference, yes, but there are additional things to consider. Just as different films had different looks, different digital cameras (or, perhaps more specifically, different digital camera manufacturers) have different looks. Some have more saturated reds, others have more true skin tones, etc. Digital cameras are more flexible (complex?) in that you can easily adjust these things in-camera as well as in post. Throw in color crossover, noise patterns, IR/UV sensitivity, age, and it's arguable that film characteristics can not be recreated digitally.

It's also worth noting that, while minimal in daylight, the IR and UV transmittance of a lens can effect what the camera can record.

Doing things like getting white balance correct, as Stan writes, is important to making an accurate photo but perhaps more importantly you probably need to create a custom profile for your camera to ensure the color it's recording is correct. And then of course, you also need a calibrated and profiles monitor and printer to see your results.

All that calibration and profiling creates a new problem, IMO: everything is perfectly correct and neutral. Personally, the imperfections (like color shifts) that you see in film and digital are creative aspects that I enjoy and sometimes overemphasize (including things like increasing saturation in post). Certainly this can be introduced in post.

I would consider saving a preset with a correction like this for use in post whenever necessary, and otherwise just accept what the camera gives me.

Answer 3

It is probably due to UV sensibility being different from the sensor and the eye, as well as the fact that the shape of the color sensibility of the three pigment in the filters are probably different from the ones for the human eye.

To correct such a thing, I normally use an option in DarkTable called "color zones": in the "hue" tab, you can subtly shift hues of only one zone of the spectrum. Example --- it's approximate, I just did a five-second adjustment from you jpeg:

I imagine that other programs have something equivalent.