This is all nice and well, and i really appreciate the high-CRI emitters that have started to spawn out lately, so please don't take this as bashing. Please read the whole post carefully before flaming...
The photos posted in this thread tell you absolutely NOTHING about the color reproduction. Nothing at all. You could've just as easily set the white balance to favor the colder light and get a Nichia high-CRI shot that lacks as much in the blue part of the spectrum as the "normal" shot in the first post lacks in the red part of the spectrum.
Here - look what happens when you neutralize the white balance for both shots (also fixed the contrast and brightness to show it better):
Can you tell which one was the high-CRI shot and which one wasn't?
Take a closer look at the color chart. Any differences in the saturation of any particular patch? Or its tint?
What does that tell you? It tells me the
exact same amount of color information was "picked" off the scene by BOTH lights. The only difference was the color balance. And that's what's wrong with comparisons like this and the way CRI is defined.
It favours certain color temperatures and CAN'T really tell you much about the color reproduction of two lights with DIFFERENT color temperatures. It's ill concieved for that particular task. What it CAN tell you is how two lights of the SAME temperature compare to eachother, though.
orcinus,
The fact that you have been able to digitally adjust both of these pictures, by changing the white balance and brightness and contrast, so that they are nearly indistinguishable, does NOT say that the
exact same amount of color information was picked off by both light sources (that's a fairly absolute statement, that is!). In my opinion, it says that both sources are full enough spectrum, with no holes and spikes present in the important areas,
for this set of images that
nearly the same amount of color information is present in both of the pictures I posted in the first post of this thread.
But, what if the images were of a painting with many different shades of red? And what if the observer were an artist or art critic? What then? Then, I would be willing to bet, this fancy digital footwork would not be enough. Or if there were colors near some of the spikes and rough parts of the Cree's spectrum (or 083's, for that matter)? Read back in this thread to BabyDoc's first post where he was able to see something with the 083 that he wasn't able to with a Cree. For most of us, the Cree may be more than good enough. Hell, I use a Golden Dragon or Nichia 3mm LED's for most of my daily flashlight tasks, and they are good enough for my purposes. Few people
need more, as today's LED's are full enough spectrum that the information is there. I essentially agree with you, here, I think.
So, let me go further: let's say you're right. Let's say I concede the point that the full information is there in both images, and just needs to be processed.
What does that matter? I'm not an instant computer running Photoshop! My eye sees what my eye sees. And if that is an image that is yellowy due to a low CCT light source, or of an image that is cool bluish due to a very high CCT light source, then
it matters!!! (And the fact that people prefer some form of sunlight isn't just arbitrary. We evolved seeing under that light source.)
This is EXACTLY the reason that so many people have been trashing incans all these years, wondering when they would die, and what was the point, and who would be stupid enough to want such **** yellow light! Despite the fact that the spectral curve was an almost perfect plankian black-body curve, people didn't like it. It was too low a CCT, too yellowy. And remember the binning of the Luxeon LED's, with the spot in the four letter code that was expressly for deviations from the PBBL (Plankian Black Body Locus)?
The fact that "the information is there" doesn't mean that it's just as good as any other light source where the "information is there" also.
Not good enough. CCT is important. And once you have that, deviations too far away from a black body spectral curve--holes or spikes--are also important. Our brains adjust to things, it is true, but it still matters.
To say that the white balance of the light is unimportant for color reproduction is a bit crazy. Of course it's important, which is why we have all these tools to keep track of it and adjust for it,
so that all images come out looking as if they had been photographed with natural light.
So you made the Cree image look almost indistinguishable from the 083? So? I didn't have to do ANY processing to make the 083 image look like a daylight image.
And THAT is the point.