In message
,
Pooua writes
On Mar 21, 2:47 pm, Bob Williams wrote:
Pooua wrote:
http://web.mit.edu/neltnerb/www/artw...x.htmlfeatures artwork
illuminated by super-bright LEDs, but the photos do not accurately
reflect the colors of the lighting. The artist says that his camera
has trouble picking up the purple lighting, instead showing it washed
out, apparently because it is outside the normal color space of the
imaging sensor. Does that sound likely? What might a photographer do
to take better photos of these tricky lighting situations?

I think the reason is, that NO combination of RGB used in sensors can
produce violet (purple?) light. The visible color spectrum is ROYGBV.
All colors between R and B can be generated by mixing appropriate
amounts of R, G, and B. But Violet is a shorter wavelength than any of
the frequencies captured by an RGB sensor. So no combination of longer
wave lengths can produce a shorter wavelength.

But the human eye perceives purple as blue with a red leakage. So you
can get a very convincing purple and magenta out of the RGB colour
space. The problem is more in the image capture device itself that to
optimise its performance on flesh tones where our eyes are very
sensitive to slightly off colour casts they have to lose the residual
errors somewhere and usually it is along the line of purples.

You can't fix it with Photoshop either because PS also uses an RGB
palette (e.g. Adobe RGB).
Bob Williams

The wavelength doesn't really matter only how it stimulates the eye
sensors. This could make for trouble if you do have something
illuminated with specific narrow wavelengths (monochromatic LEDs are
typically 50nm bandwidth on a nominal wavelength of somewhere between
650nm (red) and 350nm (violet) which makes them a much purer colour than
a typical gel filtered light source. At least with digital you can
experiment cheaply.

I don't know if that explains the problem.

1) Purple is actually a combination of red and blue.
2) The cones of our eyes only detect red, green and blue (some
extremely rare women can see a 4th color)

They actually detect green, yellow and blue. Red is a creation of the
brain as the difference signal of yellow-green raw signal.

I just found a website that states that humans detect violet by
comparing the ratio of blue light to red light. According to the
article, humans could just as well use the ratio of blue light to
green light, but the human spectral response to green and red light in
the blue portion of the spectrum is about the same, so it does not
make a difference which system humans use.

Digital cameras, OTOH, are more likely to sense violet by the ratio of
blue to green light. So, the response is not the same. That's my take
on what the article is saying, anyway.

http://gene.bio.jhu.edu/violet/violet.html

It would seem from this that it would be possible to switch the green
and red channel to find the violet. Maybe. Possibly. But, what happens
to everything else in the photo?

Anyway, I am thinking that by beating the scene into submission
through the clever use of RAW mode, bracketing and maybe HDR, I ought
to be able to help this artist. Too bad that he is in Boston, while I
am in Dallas.

Your best bet is to take a series of photos that span a range from
wildly underexposed to capture the true colours of each light source to
over exposed to capture the overall lighting effect. Finding a nice way
to combine them is a bit of a challenge, but provided nothing moves
between the shots it should be possible using some of the high dynamic
range tricks from combining multiple exposures. Good luck!

Regards,
--
Martin Brown

--
Posted via a free Usenet account from http://www.teranews.com