How much can DSLR's correct colour casts?

[ron c sunshine coast,qld,a]

I'm not sure exactly how to word this so hopefully what i'm asking

will make sense.

<p>My question is specifically to do with the ability of digital

cameras to correct various colour casts (allthough film cameras had

similar problems i remember).

<P>As far as i understand it,something that is too cool or too warm

is obviously easy to fix -so much so that one can get a perfectly

ballanced true colour image.In a sense the colours are all there -

they just need shifting abit.<P>

<BR>But what about discontinuous spectra such as flouro's ,etc.Is it

*REALLY* possible to completely correct for such colour tints?

<BR>I ask this because i assume that the only way to fully fix such

a tint is to reduce the offending colour and increase the subdued

colours (?).If it does happen this way then there would be definite

limits.Increasing the subdued colours would inevitabley lead to an

increase in noise.

 

<P>Am i on the right track with the way i understand it so far?

[ajweiss]

What an interesting question! As a photographer and a physicist, I am very intrigued.

 

<p>I don't know what algorithms are used to process the color information in a raw

converter, but my guess is that they assume a blackbody radiator of a certain temperature.

Blackbodies are theoretical objects that emit radiation at all wavelengths with a smooth

intensity curve peaking at different points depending on temperature.

 

<p>What this means is that they change what they think white is and adjust the other

colors to match that white point. I highly doubt that they take the spectra of different

kinds of lighting into account. When I process RAW images, I only have control over

temperature and tint, not effective gain at specific wavelengths. I assume the camera does

the same.

 

<p>It would be possible to do so (though I know of no software that does it), but you

would definitely increase noise at certain wavelengths. Some common kinds of lights emit

almost nothing at some visible wavelengths. Sunlught isn't a smooth curve either, so what

would we be correcting to? Sunlight? Tungsten? A Blackbody?

 

<p><IMG SRC="http://mis15.ncarts.edu/film/282/images/spectrum.gif">

<br>Source: <a href="http://mis15.ncarts.edu/film/282/fluorescent_lighting.html"

target="_blank">NCSA School of Film</a>

[ron c sunshine coast,qld,a]

Wow,excellent info and comments adam!

<BR>While it doesn't answer my question it does word it alot better :)

<P>The thing that really got me thinking about this was the fact that the latest DSLR's have flouro light correction adjustable on two axies-blue/amber and green/magenta .Does this feature completely correct such lighting (or does it come very close to doing so)?

<P>Coincidentally i was just reading the new 5D white paper and saw the gamut graph where they describe this feature.It has the black body line that you mention running through the graph

[byronlawrence]

I would have guessed something like that,

 

all I can say is that while the temperature is set for the white point the rest of the lighting (in different ligthing situations) looks invariably different. which makes sense after looking at those charts chart.

 

correcting for the differences in different light in my experience (which isn't vast) is next to impossible in photoshop. I don't want to say that it is not possible,,, just beyond my ability. although I feel I can get pretty close sometimes. when I do , I get ugly results, so I don't.

 

so I bet you are on the right track.

[ron c sunshine coast,qld,a]

Sorry-meant to say "while it doesn't answer my question *completely*,etc "

 

<br>I shouldn't post/respond when i'm tired...:)

 

<P>By the way-how does HUE figure into this question (if at all) ?

[robin_sibson1]

By selective enhancement or reduction at particular frequencies, you can adjust one continuous frequency spectrum to match another; it's called a filter (in the more general signal processing sense than the piece of glass we are all familiar with, although the effect is the same!). However, there is no way of turning a line spectrum into a continuous spectrum - you can't make owt out of nowt, as they are alleged to say in Yorkshire. The saving grace is that what the eye sees is, to a good approximation, sampled from the spectrum in three channels, red, green, and blue, so provided that a line spectrum has a reasonable number of different frequencies (at least three) in it, spread across the visible spectrum, you can re-balance the three components and do not need to worry about the spectrum as a whole. There are, as usual, some subtleties in this: the eye is really a bit more complex than that; one frequency of light falling on something can stimulate re-emission at a different frequency; and so on. But to a first approximation it provides the answer to your questions.

[digitmstr]

I suppose a software could be developed which captures a "print" of the light source and make appropriate adjustments. In the audio world we have had such a thing for quite sometime for noise reduction, etc...

 

As it is, we are "limited" to color temp corrections using the Kelvin scale.

[matthias_meixner2]

Cameras sample 3 color channels: red, green and blue using filters in front of the sensors. Color correction is performed by amplifying or damping one or more of these 3 channels, which may increase or reduce the noise regarding this channel.

 

Then there is another problem: The characteristics of the filters used does not fit the characteristics of the sensitivity of the receptors in the human eye. Therefore, two colors may look different to a camera although they look the same to a human (e.g. monochromatic yellow light vs. yellow light composed from red and green, depending on the filters used, the monochromatic light may end up looking greenish or orange). Obviously you cannot correct this effect in software.

[edgreene]

If I understand the original question it might be answered at photosig.com.<br> Portfolios by the same photographer, same day-different models, the color cast can vary wildly.<p>I've wondered why a(ny) DSLR cannot perform the identical way shot to shot. <p>

[Spearhead]

The problem in that case is the photographer, not the camera. In the same lighting, if properly set up, a digital camera will give the same color appearance on every shot.

 

The issue is really with mixed lighting, which requires RAW files that are converted for multiple white balance settings and then blended.

[ron c sunshine coast,qld,a]

Anyone want to comment on HUE? (how it relates to all this?)

[mark u]

There's a neat illustration of the eye's colour response right here on photonet:

 

http://www.photo.net/photo/edscott/vis00010.htm

 

As you see, it's not monochoromatic, but rather broad specturm for each colour channel (leaving aside the very broad spectrum nightvision sensing). Moreover, the red channel actually has an important secondary peak that's actually at a shorter wavelength than the peak of the blue channel (contribution to completing the colour wheel, and sensations of purple as red/blue combinations). These features mean that the eye's native sensitivity differs somewhat from the Bayer filters. Things get even more complicated if you add in the eye's own white balancing tricks - an area that was much reasearched by Edwin Land, the inventor of the Polaroid cameras - see for example:

 

http://www.rowland.harvard.edu/organization/land/theory.php

[ron c sunshine coast,qld,a]

For those that are interested i've been looking into 'daylight spectrum' lighting and have discovered some very interesting info.

<P>There are two fundamental measurments that describe light (in terms of photography anyway).

<BR>They are colour temperature and Color Reproduction Index.

<P>From what i've read so far, colour temperature can be easily compensated for (warming/cooling filters) while the CRI has a much stronger influence on how lifelike an object apears on film.

<P>Just as an example- normal fluoro tubes have a CRI of about 70 which can not be properly compensated for.There are high CRI (90+)fluorescent lights that give excellent colour rendering.

If anyone is interested in this subject,just do a search for "cri" on this site