Originally posted by simico I guess these were shot in jpeg, right?
I had similar issue with a series of flower macro shots before, violet/purple flowers became blue in camera jpeg and with default raw conversion (in LR, Bibble and RT). The cause of it turned out to be a combination of slight hue shift of "bright" image mode (used by cam jpg and PPL using camera settings), not wide enough working color space (default sRGB instead of ProPhoto in Bibble and RT), curves applied by cam jpg and raw developer apps and slight oversaturation. Using "Neutral" in PPL, setting ProPhoto as working space and a preset with the most neutral/conservative curve in other raw apps resulted in much better colors. I guess all the "wrong" settings combined together somehow shifted the original colors.
That sounds about right - however there is another wrinkle as I found out - I had posted the same question over at CPF - where there are some people who are very experienced on lights and lighting -
and I got this reply in post #
26 (link):
Originally posted by blasterman: Shooting in sRGB -vs- Adobe doesn't help for one thing.
Also, the blue cutoff for most dSLRs is around 455nm, which is deep blue, but not violet.
If these are high end stage lights using HID with filters, then the filter used might be a bit less than 455nm. I looked this data up and was shocked at how much energy deep blue / violet theater filters have below 450nm. This is why it looks deep violet to your eyes, but looks blue on camera.
There's a difference between purple, which is just standard blue+red and violet, which is < 450nm.
The problem of course is that for any photo that is displayable on the web has to be sRGB - so photos kind of becomes self-limiting -
Of course one might have 42- or even 48-bit RAW and process only with Adobe RBG or other superior color space -
BUT when one has to display the thing on the web it has to be one of the common graphics files like JPG, and has to be sRGB......
so what becomes of one's 48-bit color wide gamut?
I think the explanation that there is a cut off on dSLRs (any Bayer matrix digicam) of 455nm makes the most sense in this case.
RAW and any wider gamut processing is not going to help if that part of the spectrum is not captured in the first place.
Let me also clarify my colloquial mistake - violet and purple are not the same (as pointed out) - the problem is in trying to be clear I described violet as "purple" which is kind of the dictionary description of violet "purple-blue" -
but violet is a wavelength 380-450nm - whereas purple is a mix of red and blue -
even the Wikipedia uses that colloquialism :
"
Violet (color), which is primarily used to describe a color also called
purple "
In that link for
purple there is a section
2 Purple versus violet :
"
Violet is a spectral color (approximately 380-420 nm), of a shorter wavelength than blue, while purple is a combination of red and blue or violet light.[9] The purples are colors that are not spectral colors – purples are extra-spectral colors. In fact, purple was not present on Newton's color wheel (which went directly from violet to red), though it is on modern ones, between red and violet. There is no such thing as the "wavelength of purple light"; it only exists as a combination.[4]
On the CIE xy chromaticity diagram, violet is on the curved edge in the lower left, while purples are the straight line connecting the extreme colors red and violet; this line is known as the line of purples, or the purple line.[10][11] One interesting psychophysical feature of the two colors that can be used to separate them is their appearance with increase of light intensity. Violet, as light intensity increases, appears to take on a far more blue hue as a result of what is known as the Bezold-Brücke shift. The same increase in blueness is not noted in purples. Pure violet cannot be reproduced by a Red-Green-Blue (RGB) color system, but it can be approximated by mixing blue and red. The resulting color has the same hue but a lower saturation than pure violet."
It seems that the Wikipedia is suggesting that RGB Bayer matrix sensors do not capture violet wavelengths but approximate it with red and blue - which is really "purple"?
So I think the reason why the true violet light photos are shifted toward deep blue is simply because my digicams have a cut off around 455nm and violet is below that, and the RGB Bayer matrix is unable to capture true violet - instead approximates with a combination of red and blue to give a hue of "purple" -
but because of the cutoff the colors are rendered the shortest wavelength (adjacent) to violet - which is deep blue.
My other photos showing "purple" were not violet - but real purple - which was probably made up of mixing red and blue.
Thanks for the discussion I have learnt a lot here.