I don't know whether anybody would be interested at all, but yesterday I tried to take some photographs of various colored objects using various light sources in order to show what I'm seeing and have been discussing with regard to spectral curves. Well, it was quite frustrating, to be honest, and the limitations of my sad little point & shoot camera as well as my relative inexperience at taking photos didn't help. Rarely have I seen such photos that were useful in drawing any conclusions, and now I understand why. :hairpull: Usually the photos don't quite match what people actually see, and comparatively many colors that look different to the human eye can look the same in photos.
Infantile whinging aside :whistle:, I finally did manage to come up with a set of photos of one particular object that fairly accurately shows what I've been talking about--Johnson & Johnson's Baby Shampoo, which apparently is purple these days (it was always orange when I was a child). Most of the following photos have a couple of other colorful fluids to either side, but one doesn't look as different between various light sources in the photos as it does to my eyes, and the other just doesn't look accurate in general, so ignore them. The photos were not retouched with regard to color, although I did make in-camera adjustments to the white balance when necessary in order to make them come out as accurate (to my eyes) as possible--while using a fixed white balance would seem to be ideal in theory, it seems to accomplish very little in practice. Oh, and one more thing--the photos should look alright on computer monitors set to a 6500K white reference, so use that if you can (otherwise they will look too bluish, although that's probably OK for doing comparisons).
First, here is the photo of the baby shampoo under noonish sunlight and a clear blue sky (an ideal reference). It's maybe a tad more reddish in the photo than it appeared to my eyes, but it's fairly true-to-life.
Here is the same bottle (in the middle, obviously) under incandescent fixed lighting (65W BR30 bulbs), just to provide another common baseline for comparison. The color of the shampoo seemed a tad more saturated than the photo shows, but it was rather pinkish, as shown. What happened to the purple? This light has too much red and not enough blue, I guess.
Here is the photo I took under a warm white (2700K CCT) CFL bulb. :sick2: Nearly everything has an orangish and also a greenish tint under this thing, probably due to the
gigantic spikes in its spectral curve (tons of orange, but not much actual red). Instead of purple, the shampoo looked sort of grayish-violet under this light source, and the photo did a decent job of capturing it.
Now let's try the "cool white" fluorescent tubes (Philips Alto 4100K CCT, 62 CRI) under my kitchen cabinets. The photo came out a bit yellower than the real light looks, but the color rendering of the shampoo is close enough. It's only a slight improvement over the warm white CFL, despite the neutral white (in LED terminology) CCT.
For kicks, here is the photo I took under a not-so-bright old "daylight" (not even close, and not one of the specialized type) fluorescent task lamp (CCT looks to be over 9000K). The purple shampoo actually looked quite rich in color, surprisingly, but there was far too much blue and not enough red in comparison to sunlight.
Here is the photo I took under an incandescent flashlight--this is what high-CRI warm white LEDs strive for. The colors look nice, as expected, but obviously there is far too much red, so the color rendering is nonetheless inaccurate in spite of the light's 100 CRI.
The photo I took under a typical cool white LED shows, unsurprisingly, that it lacks sufficient red, although it should be noted that these LEDs perform better than any of the fluorescent lamps I used.
And finally here is the photo I took under the ZebraLight H51c with its 85 CRI neutral white (4000K) Rebel LED. The purple looks pretty close to how it does in the photo taken under sunlight, which demonstrates the excellent blue-red balance that I had mentioned in previous posts. The tint of the beam is just a touch more yellowish and just a smidge more pinkish than it appears here (what I believe people often describe as "creamy"), but the purple looks fairly true-to-life (actually slightly more reddish than it appears in person, but the same is true for the photo taken under sunlight, so the comparison is valid).
Unfortunately, I do not currently have any other neutral white (with a normal CRI) LED flashlights to use in this comparison. I used to have one, though (I was more concerned with beam whiteness than color accuracy back then :o), and from what I remember its beam was a bit more greenish-yellow in tint and it was somewhat lackluster on red hues. I wasn't nearly as impressed with it as I am with my H51c. Since I can't provide a photo myself for comparison at the moment, I went searching for such a comparison and found this:
http://www.candlepowerforums.com/vb...-CRI-Release&p=3651029&viewfull=1#post3651029
Save the second (ZebraLight SC51c H1) and the fifth (ZebraLight SC60w M1) on your local drive, and flip back & forth between them (their overall brightness is similar, which makes the comparison fair). See how the colors "pop" more in the SC51c photo, especially the red tones? It just looks more accurate and "natural" to my eyes in person, although these photos are able to capture some aspects of the difference. The color of the beam itself is also rendered fairly well in this photo, by the way.
Now look at the photo taken with the McGizmo Sundrop. I've never used one personally, so any observations I make based on photos alone therefore carry limited weight, but doesn't this photo look rather too reddish? :thinking: The spectral curve of the high-CRI Nichia 083 LED that it uses is even flatter than that of the SSC S42180 that I critiqued in a previous post, but it seems to have a similar problem with too much red. Actually, the S42180 has an overabundance of orange that makes its light somewhat resemble that of an incandescent bulb, while the 083 appears (in this photo as well as others I've seen) to be plain reddish, just as its spectral curve would appear to indicate. Another example (more subtle) can be found in this post (by McGizmo himself):
http://www.candlepowerforums.com/vb...significance&p=2501298&viewfull=1#post2501298
The color rendering looks nice, but in terms of accuracy the purple is too reddish, at least in the photo. Based on my experience thus far (I've tested many purple hues), I bet that the H51c/SC51c would render it more accurately (if not quite as colorfully, due to its slightly yellowish tint), despite the lower CRI of its emitter. Perhaps the Nichia 083 has its own advantages in other areas (it has more cyan, for one thing), although at present I couldn't say what they are in practical terms.
