Auto WB and Daylight WB, totally different

Ok so i was sitting down at a local railfan spot just waiting for a train to come. Started to just fool around with settings and what not. I shot a pic using auto white balance then took subsequent shots using the rest of the white balance settings. It was bright and sunny and knew the rest of the shots would suck interms of color and what not but it was just for fun.

Anyway i was a bit surprised at how much better the Auto WB setting looked overall in the pic compared to the daylight WB. Settings were exactly the same and focused on the same spot. What is the Auto WB so much brighter overall in the shadows compared to the daylight wb?

Auto whitebalance

HTTP response: 404 | MIME changed to 'image/gif' | Byte size: ZERO | PHOTOBUCKET ERROR IMAGE

HTTP response: 404 | MIME changed to 'image/gif' | Byte size: ZERO | PHOTOBUCKET ERROR IMAGE

Those pictures aren't a fair comparison because it's obvious that the light was changing.

In #1 it's sunny, as you can tell by the well-defined shadows. In #2 it's not sunny - no shadows. I wouldn't expect Daylight WB to be appropriate for #2 - Shade WB or Cloudy WB ought to be much better.

(#2 is also quite under-exposed, BTW.)

Auto white balance is, of course, trying to make sense of the subject matter in the image it sees. There's no way for AWB to "understand" what you want out of it under all circumstances.

I have found that, with my 20D, AWB is wrong more often than it is right. Thus, I always try to set the camera for the dominant type of lighting in any situation.

I also shoot in RAW mode all the time (actually RAW plus Large JPG). This is because it is extremely simple to correct white balance during the RAW conversion process. You can select another standard selection just as if you're selecting it in the camera or you can do other things to correct the white balance. Then, you can apply the correction to a group of images in a batch conversion.

There are two factors in play here. The first is, as Stewart said, the changing light conditions. Since you were using Evaluative metering the exposure was strongly influenced in both cases by the bright sky. In the first shot the platform was illuminated by sunlight coming from the side, so the difference between sky and platform was not too great for the DR of the camera. In the second shot a cloud has moved in front of the sun but the sky remained the same and, therefore, the Evaluative metering has failed to adjust the exposure and gave you an underexposed foreground.
In order to understand the second factor, you have to understand how white balancing is accomplished. In order to obtain the correct color balance the red and blue channels are magnified linearly while only a very small adjustment or none is applied to the green. The Daylight preset which assumes a color temperature of roughly 5200 degrees does an adjustment of approximately 2.0 x R and 1.5 x B. The Auto WB, on the other hand, sought to average the sunlit areas and the shaded ones (which probably had a temperature of around 7000). If it set a correction of, for instance, 6200, that might be a correction of 2.5R and 1.3B. In other words, a slight net increase in luminance (and would explain the warmer tone of the pavement.)

SkipD wrote in post #3618283
I have found that, with my 20D, AWB is wrong more often than it is right. Thus, I always try to set the camera for the dominant type of lighting in any situation.

And I have found that AWB is by far the best bet for any lighting condition including flash, but excluding incandescent indoor lighting.

Once I had set the WB to daylight because it was daylight and sunny. Then when I was developing my RAW files some months later, I could understand why my pictures were so red. Then I realized that daylight WB was used. Use AWB and it'll get it right more times than switching or forgetting to switch it manually.

RedHot wrote in post #3622335
And I have found that AWB is by far the best bet for any lighting condition including flash, but excluding incandescent indoor lighting.

Once I had set the WB to daylight because it was daylight and sunny. Then when I was developing my RAW files some months later, I could understand why my pictures were so red. Then I realized that daylight WB was used. Use AWB and it'll get it right more times than switching or forgetting to switch it manually.

If you shoot in RAW mode, it makes little to no difference if you choose the wrong white balance setting on the camera before taking a shot. You can merely click on the correct setting during RAW conversion.

However, if you shoot using AWB, there's a good chance that every shot in a series will be somewhat different.

SkipD wrote in post #3618283
I have found that, with my 20D, AWB is wrong more often than it is right. Thus, I always try to set the camera for the dominant type of lighting in any situation.

Funny, because with my 20D I find just the opposite is true. Most of the time, the AWB setting is closer to the way my eye registered the scene than what the appropriate WB setting gives.

Ok i see some are saying a cloud rolled in and changed the shot. I will say this whether you believe me or not, no cloud changed the subject at all. I thought i was seeign things on my viewfinder so i took more test shots using daylight and auto wb and the samething happened again with the shadows. I know it sounds crazy but its true, the sun did not get hidden behind the clouds.

Anyway, so in a situation where i'll have a bright sky but shadowy forground, i should not use evaluative metering? I usually always use it.

If a cloud didn't get between the sun and the scene, where are the hard shadows from the posts and the bench in the second shot? Something happened to them. It is entirely possible that the illumination of the underside of the canopy was better lit in the first shot because of the light hitting the sidewalk in a way that it apparently isn't in the second shot.

Many lighting situations are way to tricky in my opinion to use camera automation for exposure control. I always use the "partial" metering setting in my 20D, and at times even that is not good enough for some tricky situations. That's why I also have and use a handheld meter (with a 1° spotmeter attachment for the really odd situations).

Try this: On the first one shot with AWB, open it in ACR and change the white balance to Daylight and then post both results from the same shot.

Here's my solution to a good starting point for exposure.
Need an exposure crutch?

As for AWB... Gray card: Why your meter may be lying to you!
Notice that the very last exposure in the 2nd group of tests was of a gray card.

i need to crack open the books i have at home.

as for the pics, wb/settings and what not. I wasn't trying to shoot anything. It actuallay just came about as i was just fooling around while waiting for some train action.

SkipD wrote in post #3622489
However, if you shoot using AWB, there's a good chance that every shot in a series will be somewhat different.

Yes, but closer to the actual color temperature that was out there. And trying to tune it to the right WB during RAW developing can be tedious because "what looks right" as many different color temperatures can look fine for a particular setting.