View Full Version : is there an optimal color temperature for the digital imager ?
psychonaut
7th of April 2005 (Thu), 00:43
the imaging chip ( CCD or CMOS or whatever it is ) in a camera ... is there an optimal color temperature ( color temperature of lighting lets say ) that will produce best performance, as in signal-to-noise ratio so to speak ?
im asking because i am trying to figure out the purpose of warming filters or otherwise filters that impact color temperature. the only reason i can think of why this should be done in a filter instead of photoshop is perhaps if we illuminate the CCD with optimal color temperature image then signal to noise ratio will be better ? that is entering photoshop we will be starting with a higher information content so to speak ...
its a wild ass guess of course, thats why im asking.
and if so ... to what color temperature should we aim to bring down whatever the ambient light temperature is ?
:shock:
Jon
7th of April 2005 (Thu), 09:02
There's no optimal colour temperature; rather to reduce noise you need to make sure the sensor gets enough light (of whatever's available). Colour temperature adjustment is so the picture looks right to your eyes (which have the finest AWB around) even though the camera saw the orange/blue that was there. Doing white balance in-camera, however, is much easier than doing it in post-processing.
psychonaut
7th of April 2005 (Thu), 20:51
i guess to answer this question i would really need some understanding of how sensors work and how camera adjusts white balance.
my reasoning is this ... if the color temperature is say 15,000K like some kind of a shadow ... if your camera gets enough reds in that situation then its gona get too much blues, no ? or by adjusting white balance can it selectively lower the sensitivity of blue pixels relative to red, like a different ISO or something ?
so the question is - does the camera do white balance right on the imager in analog domain or is it done digitally ?
Jon
11th of April 2005 (Mon), 13:48
It's done digitally in post-processing.
J Rabin
13th of April 2005 (Wed), 00:17
Not completely sure, but digital sensors see a lot of green. Even the .jpg histogram is a green channel histogram as a proxy for what's in the image, and sensors are marginally a little "weaker" in the blues, if you believe this guy: http://www.plumeltd.com/index2.html and thus do not perform their best under the wasted heat of tungsten 3000K +/-. I believe digital does it's best in the 5000-6500 range, which fortunately is typical outdoor or color balanced fluourescent. But heck, as long as lighting is not mixed, or affected by IR heat, then fixing color casts is soooo much easier in digital than film has been.
Jon
13th of April 2005 (Wed), 12:20
Actually, digital sensors don't "see a lot of green". Human vision's more responsive to green, so typically a digicam sensor will gel 2 pixels green for each red and blue. There are variations on this, like Sony's "emerald" instead of one of the greens, but 1R 1B 2G is a typical cell ratio. But any kind of colour correction assumes that the light source is a continuous spectrum, rather than the discontinuous spectrum typical of gas emission light sources like sodium or mercury vapour lights or any fluorescent. Tapping that with ony 3 narrowly-tuned colour samplings isn't terribly easy under any conditions.
J Rabin
13th of April 2005 (Wed), 19:18
Jon:
Is this why I can easily white balance just about any lighting color cast in RAW, EXCEPT when shooting in gyms or commercial settings with vapor lamps. Even with my WhiBal, I can't seem to get it just right under vapor discharge lighting.
Yes, what you wrote about green is what I meant to say to the poster, but I did not articulate it correctly. Thanks. J
Jon
14th of April 2005 (Thu), 08:21
Yep. I don't think it's possible to properly colour-balance under gas discharge, especially sodium, lamps. You almost have to treat every colour separately, since proper colour rendering assumes a continuous spectrum source.
PhotosGuy
14th of April 2005 (Thu), 08:34
I don't think it's possible to properly colour-balance under gas discharge, especially sodium, lamps. Right. Pure sodium vapor lamps have only one color, made up of 2 spectrum lines in the yellow region. If you're lucky, there may be some other lighting present that will help out.
vBulletin® v3.6.12, Copyright ©2000-2012, Jelsoft Enterprises Ltd.