White Balance & Canon D-60

[shamrock838]

White Balance and Canon D-60:

I’m not very knowledgeable about the intricacies of white balance…so please bear with me…

What is the difference(s) between using the Auto WB option and being sure the WB is set to each changing lighting condition? Are the differences that critical?

Please elaborate. Many thanks.

M I k e

[Longwatcher]

They are somewhat critical.
My experience with the D60 has shown that it depends on what kind of lens I have and my lighting/flashes.

Up until two weeks ago I was using 28-135mm IS lens and a 75-300mm IS lens, which are both non-L series lens. With the "L" series lens (16-35mm) I am getting much better WB on auto.

Examples:
- I get good WB on auto, when using two Home Depot work lights and a 550EX flash, even better when using my "L" series lens.

- If on auto with just the work lights I get a noticeable orange cast to the color, so I need to set to indoor lights to get the correct color balance, still a slight orange shift with non-L lenses

- On the flip side with just the flash I get a ever so slight cyan tint to the image unless I am using my "L" series lens, in which case it gets it right on. (note: major blue shift if I stupidly leave the WB on indoor with the flash)

- I have seen no appreciable difference (nothing consistant anyway) when shooting outdoors.

Lastly, when in auto WB, I have never had a shot I could not correct the color, if I go to a specific mode and I am using the wrong one, sometimes it gets too far off to correct. Thus unless I am trying for something specific, I default my camera to Auto WB when putting it away, just in case.

Just my experience

[Timo Autiokari]

What is the difference(s) between using the Auto WB option and being sure the WB is set to each changing lighting condition?

Hello Mike, autoWB does it either by an algorithm that tries to guess the color temperature of the illumination by analyzing the captured data or the camera could have a coarse WB sensor. I do have the D60 but I do not know what method it is using.

We have the whitebalancing control for two reasons:

1) the human vision is said to have a property called chromatic adaptation. It means that what ever "white" the illumination is, as long as it is on the blackbody radiation curve so that it can be expressed in Kelvin degrees, it is said that the vision will compensate the colors so that we see no color-cast... e.g. a white paper always appears to us as the same white paper no matter what the color temperature of the illumination is. This theory is not very good if correct at all, in my opinion we just are not so very aware about the color of the illumination, in a room that is illuminated by incandescent light bulb everything there has the dirty yellow cast if you just pay attention to it. Some people claim that the chromatic adaptation of the human vision only works within a narrow range from 5000K to 6500K, in my opinion this is not correct either, the real reason is that we are less sensitive to color-temperature changes within that range.

2) The tri-chromaticity of the digital camera (the spectral properties of the color filter array on top of the sensor) can be so much different from the tri-chromatic properties of the human vision that correction is needed, this means that un-white-balanced shots could have a weird color cast so they must be wb corrected. Even the early high-end D-SLRs had this problem, the D60 is rather good in this respect.

Now, whitebalancing the shots means that say when we shoot a scene when it is illuminated by mid-day sun and the same scene again when it is illuminated by incandescent light bulbs we will get two photos that are quite similar in regads to colors (the form of the shadows sure appear differently since the sun is a little more far away than where we can place the incandescent light bulbs) but by colorimetry the images appear similar.

So if we always whitebalance the shots we get images that all appear to be shot under the very same illuminant, the sunlight. In other words whitebalancing will remove the information about the illumination completely away. Sometimes this is beneficial/required, e.g. when shooting product shots or when documenting something where the properties of the illumination has absolutely no value. For normal photography the characteristics of the illumination give a lot to the images and for this reason I shoot all the time he camera set to daylight and I post-process the white-balance in Photoshop only when it is needed. using the daylight setting captures the scene pretty much like it appeard for the human vision, e.g. incandescent and tunsten illumination appears yellowish, and "white" on a CRT screen that is set to 9000K appears bluish etc.

[transco]

If time permits, I would recommend taking a picture of something known to be white and setting the custom white balance before photographing your subject. If you don't have time for this, at least toss a white sheet of paper on the ground and snap a shot before moving on. I carry a couple 3 x 5 cards in my pocket and gadget bag for this purpose (I use the same cards, strapped to my flash with a rubber band as bounce reflectors). Personally I seldom have time to drill down through the menu to set the custom white balance before the shoot so usually shoot the white card when finished and "apply it" in post. I've found the best results is shooting in raw mode, then applying the white balace during RAW-to-TIFF conversion.

I learned my lesson during a "one chance only" shoot a while back. Only having the on-camera flash with me, I bounced it off what I was sure was a white ceiling. In post it became obvious that the ceiling was actually ivory. The subject I was photographing didn't have a speck of white anywhere to use as a reference. I hate to tell you how many hours I wasted in post trying to fix those pictures. Never did get it right.

[gsrossano]

The Canon bodies do not do white balance by analyzing the captured image. They make a WB measurement and apply it to the images. If the lighting changes between the measurement and the taking of the image (on a time scale of a few msec) the color will not be correct. If the lighting changes durring the exposure you will get a non-uniform exposure and color. It also depends on the shutter speed. If the exposure time is long enough variations in the lighting will be averaged out. Generally slower than about 1/125 sec. The effect is pronounced for the D30 and D60 under older type arc lights. It is also present in the 1D, but greatly reduced. It appears the time delay between white balance measurement and exposure is shorter for the 1D compared to the others. I find that AWB and custom white balance tend to give the best results when shooting in the same lighting enviornment for extended periods of time. The 1D allows setting the color temperature of the lighting, but if you don't have a color meter it is just trial and error.

[gsrossano]

I don't agree with the practice of shooting daylight white balance all the time and then correcting in Photoshop.

1. The color casts are extreme under artificail lighting, which makes it difficult if not impossible to get correct color over the full visible spectrum by post processing. It is better to get as close as possible in the camera and then only tweak it in post processing.

2. I currenty shoot about 100,000 frames a year. Up to 4,000 frames some days. There is just no time to be fiddling with every image under those circumstances - especially when things have to be turned around right away. My clients want to see the images immediately, and they are not going to buy them if the color is way off, or if I have to tell them they have to wait until I re-process them.

[hmhm]

shamrock838 wrote:

What is the difference(s) between using the Auto WB option and being sure the WB is set to each changing lighting condition? Are the differences that critical?



An in-camera exposure meter sees a certain amount of light
coming in through the lens, it can't tell whether that's due to
bright light hitting a dark subject or dim light hitting a bright
subject, so it "guesses" that the subject is about "average"
tone. If the subject isn't "average", the results will be off.
Getting the right answer either requires use of an incident
light meter or a reflective metering off a grey card. White balance
works the same way.

An in-camera meter can't tell whether greenish light is coming
from white light hitting a green subject or greenish flourescent
light hitting a white subject. AWB has to make assumptions about
the reflective properties of the subject, and results will always be
off if the subject's properties deviate from that assumption.

The 1D and 1Ds have a separate sensor on the front of the
camera body that allows them to measure light hitting the
camera. This is better than "through the lens" metering
(assuming no filters on the lens), as it can average reflected
light over a larger area (e.g. the entire area in front of the
camera instead of just what the lens is looking at). This can still
be fooled, of course, if you're standing a couple feet in front of
a wall painted green, the camera is going to see a greenish cast
to the light that isn't characteristic of the light that would strike a
person standing in front of that green wall. The D60 doesn't have
such a sensor, it can only look at what comes in the lens.

As for exposure metering, "getting it exactly right" requires either
an incident meter or a reflective shot of a grey card. These
shots can then be applied to CWB, and "exact" color balance
achieved.

The presets provide samples of "typical light". Obviously if
there's one preset for "sunny" and another preset for "overcast",
a "partly sunny" day must be somewhere in the middle. Whether
the presets are good enough or not depends on your own
view of "good enough".

Personally, I tend to use presets most of the time. Shooting in raw
is good, because you can override the camera's setting during
conversion, which is good if you tend to forget to change the
setting from shot to shot.

By the way, this isn't mentioned very often, but exact color balance
isn't necessarily what you want. If you're shooting at sunset,
do you really want to remove that sunset "color cast"? Not necessarily.
But people standing under those icky green flourescent lights
that terrorize offices shouldn't end up looking like martians.
-harry

[soumya63]

gsrossano wrote:
I currenty shoot about 100,000 frames a year. Up to 4,000 frames some days.

This may be totally off topic, but you are the right person to ask about shutter reliability of Canon bodies.

Has any D30 survived a year with you? I have clocked only 5000 in a year. A thumsup from you will definitely add to my peace of mind.

:D

[Timo Autiokari]

transco wrote:
Only having the on-camera flash with me, I bounced it off what I was sure was a white ceiling

How do you bounce the on-camera flash off from the ceiling? On my D60 it does not seem to bend to that direction and if it did the power is soooo weak that I believe it would not help any.

[gsrossano]

soumya63 wrote:
gsrossano wrote:
I currenty shoot about 100,000 frames a year. Up to 4,000 frames some days.

This may be totally off topic, but you are the right person to ask about shutter reliability of Canon bodies.

Has any D30 survived a year with you? I have clocked only 5000 in a year. A thumsup from you will definitely add to my peace of mind.

:D

In general the D30 has been trouble free for the year+ I have used it. I have lost count, but I think I am up over 100,000. My D60 shutter died after 25,000 frames over 7 seven days. Since being repaired I have put another 50,000 on it. It has other problems though. My experience is that the D30 is a lot more reliable than the D60.