Color Checking - the Hard Way (warning: CFL content!)

[xpatUSA]

After venting mild disappointment with Foveon color rendition here it became clear that CFL's are not highly regarded by many photographers. So, after a bit of Googling I found an example of that discontinued and fast disappearing species, namely the GretaMacbeth mini-card, just right for a table-top photographer to investigate other means of lighting. While waiting, I wondered if there was another way since, as a watch photographer, I'm not too concerned about skin color, nor greenery for that matter.

I had taken a shot of a tri-phosphor CFL lamp's spectrum sometime back using a Sigma camera. I also found a spectral power distribution graph (see item [6] on the page), which was conveniently annotated with the phosphor emission peaks. Here they are:



The chemical symbols are for the three phosphors used: mercury, europium and terbium - the latter two being doped. Knowing the phosphors used allows one to look up their emission wavelengths, for example, here.

Thus we find that the highest two peaks above (Hg and Tb3+) are for wavelengths of 547 and 542nm, almost as green as it gets in the world of color ;-)

By this means, one is able to guess that the Foveon was a little bit off in this shot, providing something more like yellow.

Almost made me want to go out and buy some Halogens and burn my fingers :-) but, instead, I sold the camera and bought a D90 in it's place . . .

Ted

[Manfred M]

Okay; I'm impressed. Not everyone would go out and build a spectroscope to figure out what the spectrum of a CF was. Is the D90 going tp perform any better? You are still going to have to deal with that ugly green spike.

[xpatUSA]

Okay; I'm impressed. Not everyone would go out and build a spectroscope to figure out what the spectrum of a CF was. Is the D90 going tp perform any better? You are still going to have to deal with that ugly green spike.

Well, remember I also had a D50 which handled reds quite well under the Sylvania CFL's, for example this old watch looks OK:



The red border around the lume at 12 looks fine to me. So I don't expect the D90 to be any worse than the D50 which, in it's turn, rendered color much better than the SD9. But I am looking at other means of lighting. I might put a Philips TL950 tube in the overhead position (claimed CRI of 98%, not that I trust CRI that much). And I'll give halogens a shot. All on-going after I receive the Macbeth color card. Must have continuous lighting, flash is still out for me (just my personal preference for no particularly good reason). LED's are still on my list to try too.

It's all such fun . . .

Ted

[Manfred M]

Good luck with the halogens. I hope you don't cause a meltdown or worse. Flash with diffusers work for me, but then I'm not into watches.

As for the D90, other than getting a little long in the tooth from a technology standpoint, it's a nice camera to work with. My wife and I both have one and have dragged them all over the world.

[xpatUSA]

As for the D90, other than getting a little long in the tooth from a technology standpoint, it's a nice camera to work with. My wife and I both have one and have dragged them all over the world.

Thanks Manfred, the D90 has yet to do any serious work around here, although it did shoot the eBay pics for the dear old D50. Imagine, using a 12MP camera to shoot pics that get re-sampled down to 640px wide . . ho hum.

And now for the deliberate mistake . . checking camera color rendition per se using real spectral sources is not possible. Nope. According to this gentleman , spectral colors lie way outside of the sRGB zone that most cameras are set to use and, apparently, it's not even easy to approximate spectral colors within an RGB color space. An example from two different websites tells the tale:



Although I was careful to line up the wavelength scales, the RGB colors don't match that well! Whereas the Macbeth cards have those subdued colors with well-publicized RGB values as as L*a*b codes (whatever they are). Can't wait for mine . .

[Edit] The implication is that, when we look at the emission spectrum of various lamps and worry about peaks, it's not possible to calculate exactly what RGB code is represented by such a peak - because that depends on whose calculation is used by the camera manufacture and/or by the post-processing RAW conversion software. For me, that is a sobering thought which is telling me that I was perhaps a little hasty in dumping the SD9?

Ted