Is camera light metering sensitive to colour temperature?

(Talking 350D with kit lens)

The recent discussion on RAW files prompted me to give them a go (and by the
way - having done so I'll NEVER be going back to just High-quality JPEGs if
it's something that has to look top-notch - but I digress).

I made an interesting observation ...

Whenever I setup my "studio quality" halogen work lights (diffused with
"studio quality" oven baking paper) and take a photo or two I used to find
two things of interest ...

First up, as expected, the colour temperature was always way off (very
cold) - so I did a custom white balance correction and got that bit pretty
close ...

Second up, (even) regardless of the white balance setting I always seemed to
end up with all shots being under exposed by a full stop (or even a little
more) - very consistent.

When I looked at the pre-processing suggestions for a RAW file for the first
time it was always suggesting a number of fairly heavy corrections (I'm
assuming that it gets this from the parameters that would have been passed
through to the Digic processor for a JPEG). If I tell PS NOT to use these
parameters HEY PRESTO almost immediately I get an almost perfect exposure
(in fact my uncorrected RAWs look better than the JPEGs have even after I've
spend 1/2 hour on PS tweaking the JPEGS.

It's almost like the camera has been taking a correct exposure all along,
but passing incorrect compensation parameters to the Digic processor.

Has anyone had any experience with this?

The only thing I could think of was "could the fact that the colour
temperature is so low be affecting the light metering"?

Cheers,

CC

 

Yes, RAW is definatly the way to go if you want any degree of control.
Just like a darkroom without the mess.

No. I would like to look at this. Are you able to email me a RAW file?
(Remove NOSPAM from my reply address)

-Mike

 

[..]

You probably mean "could the fact that the colour temperature is so
high", compared with normal incandescent light - or "so low" compared
with daylight? In either case, not a colour temperature the camera is
calibrated for.

I don't know about your camera model specifically, but I have read
that many digital cameras (including DSLRs) use the green luminance
channel only, not an average of the RGB channels, to compute and
display the exposure histogram. If that data is then used to compute
exposure compensation values to store in a RAW file, then the answer
is yes. A high-Kelvin light source (containing a lot of blue and
green) might cause the camera to think it has overexposed and try to
compensate by reducing the exposure factor after taking the shot. I
don't know if the sensors used to compute the exposure value _before_
taking the picture have a flat spectral response and/or if they
compensate for different light temperatures (I would expect they do).
If so, they also can give a biased exposure with light of a
temperature they are not calibrated for.

 

I might be getting myself confused here, but I was under the impression that
halogen lights would be best suited to a white balance setting of
"tungsten" - the same as incandescent lights? The halogen lights appear to
have a spiralled tungsten coil in the bulb which starts off glowing red and
slowly turns "white" as I crank up the voltage on my variac. In any case,
regardless of which white balance setting I used I always ended up with
quite a severe red colour cast - which I understand means "low colour
temperature" - it appears to be considerably lower than any other light
source.

Only the custom correction totally compensated - the trial shot of a sheet
of A4 paper was done on a tungsten setting and it came out with a very
distinct colour cast - a follow up shot with the custom white balance came
out (more or less white) - different as chalk and cheese.

 

I've been doing some more testing - and I think I made a wrong assumption.
When I imported the RAW file into PS I assumed that the parameters entered
by the system into the "pre-processing" dialog boxes were recovered from the
RAW file as specific values - but on closer examination it appears as though
PS is simply "looking at" the histogram and making recommendations based on
that (unfortunately they're waaaaaay out).

I've still got an exposure issue to sus out, but I'll do some more
experiments with AEB and metering first.

Thanks for your help.

CC

 

I am under the impression that halogen photoflood lamps for
photography use have a higher temperature (i.e. a bluer cast) than
normal incandescent bulbs. However, if you use a different type of
halogen lamp, anything is possible. Using a variac to change the lamp
voltage does affect the temperature of the lamp and its colour cast,
so you have to recalibrate the white balance every time you adjust the
voltage. Darkroom enlargers for colour prints used to have a
stabilized voltage/current source because even a tiny change in the
lamp voltage would throw the white balance off.

 

Yes, exactly. I have a set of older studio lights with a calibrated
transformer so I can change them from 3200K (tungsten) to 3400K (type
A photoflood). Any time you change the voltage you change the colour
temperature. Genuine photo type bulbs burn at approximately the same
colour temperature throughout their life but there can be quite a
range of colour temperatures for bulbs not specifically designed for
photography.

 

I'm wondering if it's possible to get bulbs for the work lamps that are the
right colour temperature? These ones are definately red red red. I don't
actually use the variac for my main lights - saving it instead for a
backlight, so the main ones are always run at full voltage (which in all
fairness probably varies a little).

I'm wondering if, in our modern days of digital photography and tools like
photoshop if things like white balance aren't anywhere near as critical -
I'm thinking of including a black and white card in the corner of each shot
to provide a reference for photoshop when removing a colour cast, then
simply crop it out for the final "print".

 

Set up the lights, then take a shot of a Macbeth Color Checker, or, on the
cheap, a gray card or white piece of paper or something at the subject
position. Then just shoot. Do your adjustments on the test shot, then
apply them to the entire series shot under the same light.

 

Thanks for that.

What's the basic technique from within photoshop for applying the same
correction to a series of photos?

 

In CS2, there are a couple of ways. The first is to make your adjustments
on your test shot in Camera Raw, and then go into Bridge, do "copy Camera
Raw settings" on that image, and then "paste Camera Raw settings" onto all
the other images; it will give you a chance to tell it which settings you
want to apply.

The second would be to load up all of the images at once into Camera Raw,
make the adjustments on the first, then click "select all" and then
"synchronize". This, too, will let you choose which settings to apply.
Obviously this probably isn't the best choice if there are a large number
of images.

In CS2 as well as previous versions, you can do them one at a time as you
load them; first do your adjustments on the test image. Then, load up
one of the other images and choose "previous conversion" from the drop-down
menu, which will apply the settings from the last picture you did. If you
do that every time, you can easily do some small number of pictures.

You can also save the Camera Raw settings from the test image, from within
Camera Raw. The little pop-up on the right side above the adjustment
controls has two functions for this, "save settings" and "save settings
subset". The latter will let you save only some of the settings. You
can then do "load settings" with your other images to apply those settings
to those images.

 

In message <>,

I doubt that the weighting for the histogram is used for anything but
the histogram. In fact, if you shoot flat, evenly illuminated colored
surfaces with AE, on my Canons, different colors wind up at different
parts of the histogram, so the weighting is clearly independent of the
metering sensitivity by color.
--

 

In message <mcxIe.322$>,

Right for what? If you want lighting that renders grey objects such
that the RAW values for red, green, and blue are the same, that would be
magenta for many cameras, and the only way to get magenta light is with
filters, as magenta is not a color temperature.

I use magenta filters over my flashed when I am using all or mostly
flash, and the shadows are much better than I get with any normal light
temperature.

Daylight and tungsten are both arbitrary, non-natural light sources for
RAW data.
--

 

In message <mcxIe.322$>,

It's actually very critical, if you care about the qulaity of shadows,
but as I said in another post, the best light is usually magenta with
some (maybe most?) RGB Bayer cameras.
--

 

Black does not help you to get a white balance (does "all cats are
black at night" suggest anything?). Gray does, of course. White is not
usable if overexposed. Gray also allows you to calibrate the exposure.
You don't need to have the card in every shot, just set the camera's
white balance on manual, put the card in a first calibration shot,
then continue shooting without card until you change something in the
lighting setup that might affect the white balance (primarily the lamp
type/voltage and the positions of reflectors/diffusers if they are not
white/silver). A calibrated neutral gray card is far preferable to a
homemade one, which could have all sorts of strange colour tinges that
might even change depending on the angle of illumination.

 

In message <mcxIe.322$>,

B&H carries these little pads of papers with a white, 18% grey, and
black on each. Don't know how accurate the grey is, though. Next time
I'm at B&H, I'll buy a pad and test them against the Gretag-Macbeth
color checker.
--