I'm still working through some shots from my trip this summer, and last night I spent some time on this image form RMNP. This first post is what came out of my original processing work.
Then I just decided to try an auto-color adjustment in CS3 and the image changed considerably, and I started to wonder if my original shot had some color cast to it.
What do you think?
Posted by Todd Caudle on October 8, 2008 at 9:25 AM
I'd say there was a slight yellow color cast in the first one, but very, very subtle. Keep in mind these things are highly subjective, and you'd probably poll 100 people about which they prefer and come away with an even split. I certainly couldn't pick one over the other with any confidence, even looking at them side by side.
I agree about chosing one over the other, partially why I posted on here to see what people thought.
I seem to find a bit of color cast in many of my images, does that mainly come in from the white balance?
Posted by Phillip Noll on October 8, 2008 at 9:42 AM
The first has a definite yellow-greenish cast on my monitor here at work. I think the "auto" adjustment removed the cast adequately. I also like the increased contrast of the 2nd post. The first one looks a little flat.
Posted by John Mumaw on October 8, 2008 at 9:57 AM
My views are in line with Todd.
Using the grey eyedropper in either the Curves or Levels dialog and clicking on a grey or neutral color (e.g. the mountain tops) confirm that a color cast is present.
I always use Auto Color as a quick check, and find that it has a pretty high success rate.
See Bruce Fraser's short essay on using Auto Color, which includes slight tweaks to the default parameters.
When I use Auto Color, I also use it via the Curves or Levels dialog box as described in the essay because then you can implement it as an adjustment layer, which doesn't damage the original image data.
Posted by Rick Dunn on October 8, 2008 at 10:29 AM
I see a mild yellow cast as well, but agree that it's not a big deal here. I rarely have success with the auto-color feature in PS. I usually use it as a quick check, and then almost always undo the change and manually manage the color balance.
The light is fairly neutral here, so I'd expect the auto feature to work reasonably well. But in the warm light of sunrise/sunset, or the cool light of open sky, I find the auto feature tends to remove a cast that should actually be there given the lighting conditions.
Where does a color cast like that generally come from?
Posted by Rick Dunn on October 8, 2008 at 10:47 AM
Probably from your white balance setting. If you're shooting in auto WB, the camera is selecting the color temp and it will vary between shots. I don't know much about the algorithms used in auto WB, but even with two exposures of the same scene in short succession, it's not uncommon to end up with two different color temps and noticeably different casts.
It can also come from filters, particularly cheaper ones.
Thanks, I was thinking it was coming from the white balance. When I go to import my RAW files using the ACR utility the images always come up at 6700 for the temp. Seems pretty high for many of the images.
Posted by Mel Ladewig on October 8, 2008 at 11:14 AM
If you opened the file from RAW, it is possible that you (or ACR) chose an incorrect color temp and tint combination. If you used an in-camera AWB setting, its possible that the camera was fooled on the white balance.
As Rick indicated, the risk in using an auto adjustment is that it probably won't recognize when a color cast is supposed to be there, in cases where its either sunrise or sunset. The gray dropper in levels is fine if there is something in the scene that you absolutely know is neutral gray, but that may not always be the case. There are other cases where a color cast in an object that could be seen as gray are desirable like the deep blue underside of a summer thunderhead.
One way to get around this is to use ACR with settings known to give a daylight-like result. Chose a color temp between 4900 and 5500. The tint might need an adjustment, but it seems to vary by camera and color temp. If you've set the camera to daylight at the time of capture, Canon raw files default to 5100K and +5 tint (purple) when you open them in ACR. ACR defaults the color temp to 5500K and +10 tint for its daylight own setting. Both are reasonable, but for my XT I typically prefer a point in between. I usually use 5300K and +8 tint, though I do often make slight tweaks to this white balance.
I have found the shade and cloudy defaults for ACR to be too warm for my tastes and typically in these situations I use a color temp around 6000K or lower.
By deliberately choosing a white balance that emulates daylight, you are forcing your camera/file to act like a daylight balance slide film.
Also I've read recently that Adobe have done their testing with cameras under laboratory controlled tungsten and daylight lighting. This explains to me why the ACR profiles for my camera are OK for daylight and tungsten, but unreliable for fluorescent, cloudy and shade.
Posted by Phillip Noll on October 8, 2008 at 12:44 PM
I've always thought that, when shooting raw, the White Balance setting in the camera has no effect. Is that not the case?
Posted by Grant Collier on October 8, 2008 at 1:19 PM
The white balance has an effect in raw because the setting you select in the camera is the setting that is initially used in ACR or Lightroom. With raw, you can always change those settings in ACR, so choosing the wrong white balance in camera is not too big a deal...it just saves you time, because if you initially have the right setting in camera, the initial image you see in ACR should be much closer to what you want.
Kevin, if the initial white balance in ACR always shows 6700K, it probably means you have your white balance set at shade in your camera and you haven't been adjusting it based on the shooting conditions. So that would lead to a yellow color cast for any cloudy images and even more of one for sunny images. If you change the white balance in camera to match the shooting conditions, the initial color temp in ACR will be closer to what you want.
Posted by Mel Ladewig on October 8, 2008 at 2:22 PM
I've always thought that, when shooting raw, the White Balance setting in the camera has no effect. Is that not the case?
Grant is exactly right. The camera just tags the image with a white balance giving you a starting point (and affecting the embedded jpeg), but it doesn't cause any change to the raw pixel data.
Remember that when a camera is outputting a raw file, it is basically dumping the pixel data straight. It doesn't apply the Bayer matrix algorithms, so its up to the raw editor to determine what the color is supposed to be. That is why ACR (Lightroom and Photoshop) and DPP give significantly different looking results.
Most of the raw formats embed a small jpeg into the image. This is used to provide the preview image on the back of the camera when you shoot, but it cannot be extracted from the raw file and doesn't affect the raw conversion.
Posted by Phillip Noll on October 8, 2008 at 2:45 PM
OK. So the WB setting does not affect the pixels in a raw file (just as I thought) but it is used as a starting point in ACR/Lightroom (which I wasn't aware of). That makes sense! Thanks for the clarification guys!
Thanks for all the information everyone, this helped clear some things up. I was thinking I must have something set on my camera that everything is so warm, I will double check the settings tonight and see where I'm at. I didn't mind making the adjustments in ACR, I just wasn't sure why they always looked this way to start with..
Good discussion... Out of curiosity, how many of you actually manually adjust WB settings while you are out shooting? I always leave mine on Auto, but there have certainly been instances when it's way off. I guess the only downside is more time in front of the PC trying to get things balanced properly when Auto gets fooled...
Posted by Kalin Wilson on October 8, 2008 at 4:36 PM
I almost always choose daylight or cloudy based on the condition. Only when I'm really lazy do I use AWB. I also really enjoy using my ExpoDisk. It can be fooled but in the majority of the situations it gives me spot on white balance.
If you shoot panoramas, it's really important to not use AWB, choose a fixed WB and stick with it across the scene.
Posted by John Mumaw on October 8, 2008 at 4:38 PM
I leave mine on Auto because Canon's AWB setting almost always nails it for me under all but artificial light.
On another note:
"Most of the raw formats embed a small jpeg into the image. This is used to provide the preview image on the back of the camera when you shoot, but it cannot be extracted from the raw file..."
Sure it can. Check out this thread I posted in the Digital forum, which leads you to a sweet (and free) little utility that can do just that.
Posted by Phillip Noll on October 8, 2008 at 5:31 PM
I leave mine on "Auto" and it does get fooled sometimes. For instance, the photos I took in the rain forests in the Columbia River Gorge and Olympic NP all have a strong yellow cast to them and I had to cool them down considerably to get back the deep greens of the foliage in these areas. Sunsets/sunrises are most often fooled too and require a bit of work in ACR to get the colors back to where they should be. I don't mind making these adjustments as I consider it just another part of my workflow and usually the adjustments are based on aesthetics anyway.
Posted by Nat Coalson on October 8, 2008 at 9:15 PM
Outdoors, I always use Daylight white balance as the starting point, then fine tune the WB for every image in Lightroom. Never Auto WB.
I've found that I'm not looking for "neutral" colors as much as I'm trying to preserve the look of the scene as it was shot. Whether the light is warm or cool, Daylight preserves it.
If you want to always have neutral colors, without color casts, start by using the WB presets on your camera for each lighting condition. (You'll find that you're changing WB a lot.)
If you need truly "accurate" color shoot a Color Checker and use the eyedropper to set WB on the second-lightest square. You can then sync the rest of the shots in the sequence.
Rick pointed out a really important point - "neutral" density filters aren't always really neutral. Cokins are very brown and vary depending on the direction of illumination.
Posted by Jesse Speer on October 10, 2008 at 5:52 PM
Never Auto WB? Why Nat? (considering a RAW workflow)
I shoot nothing but Auto WB, unless I'm shooting images for stitching. Just curious. I haven't noticed any impact on the quality of my results. Sometimes I have used preset WB settings, but it didn't seem worth the effort.
Discussion
21 repliesI'm still working through some shots from my trip this summer, and last night I spent some time on this image form RMNP. This first post is what came out of my original processing work.
What do you think?
I agree about chosing one over the other, partially why I posted on here to see what people thought.
I seem to find a bit of color cast in many of my images, does that mainly come in from the white balance?
Using the grey eyedropper in either the Curves or Levels dialog and clicking on a grey or neutral color (e.g. the mountain tops) confirm that a color cast is present.
I always use Auto Color as a quick check, and find that it has a pretty high success rate.
See Bruce Fraser's short essay on using Auto Color, which includes slight tweaks to the default parameters.
When I use Auto Color, I also use it via the Curves or Levels dialog box as described in the essay because then you can implement it as an adjustment layer, which doesn't damage the original image data.
The light is fairly neutral here, so I'd expect the auto feature to work reasonably well. But in the warm light of sunrise/sunset, or the cool light of open sky, I find the auto feature tends to remove a cast that should actually be there given the lighting conditions.
It can also come from filters, particularly cheaper ones.
As Rick indicated, the risk in using an auto adjustment is that it probably won't recognize when a color cast is supposed to be there, in cases where its either sunrise or sunset. The gray dropper in levels is fine if there is something in the scene that you absolutely know is neutral gray, but that may not always be the case. There are other cases where a color cast in an object that could be seen as gray are desirable like the deep blue underside of a summer thunderhead.
One way to get around this is to use ACR with settings known to give a daylight-like result. Chose a color temp between 4900 and 5500. The tint might need an adjustment, but it seems to vary by camera and color temp. If you've set the camera to daylight at the time of capture, Canon raw files default to 5100K and +5 tint (purple) when you open them in ACR. ACR defaults the color temp to 5500K and +10 tint for its daylight own setting. Both are reasonable, but for my XT I typically prefer a point in between. I usually use 5300K and +8 tint, though I do often make slight tweaks to this white balance.
I have found the shade and cloudy defaults for ACR to be too warm for my tastes and typically in these situations I use a color temp around 6000K or lower.
By deliberately choosing a white balance that emulates daylight, you are forcing your camera/file to act like a daylight balance slide film.
Also I've read recently that Adobe have done their testing with cameras under laboratory controlled tungsten and daylight lighting. This explains to me why the ACR profiles for my camera are OK for daylight and tungsten, but unreliable for fluorescent, cloudy and shade.
Kevin, if the initial white balance in ACR always shows 6700K, it probably means you have your white balance set at shade in your camera and you haven't been adjusting it based on the shooting conditions. So that would lead to a yellow color cast for any cloudy images and even more of one for sunny images. If you change the white balance in camera to match the shooting conditions, the initial color temp in ACR will be closer to what you want.
Grant is exactly right. The camera just tags the image with a white balance giving you a starting point (and affecting the embedded jpeg), but it doesn't cause any change to the raw pixel data.
Remember that when a camera is outputting a raw file, it is basically dumping the pixel data straight. It doesn't apply the Bayer matrix algorithms, so its up to the raw editor to determine what the color is supposed to be. That is why ACR (Lightroom and Photoshop) and DPP give significantly different looking results.
Most of the raw formats embed a small jpeg into the image. This is used to provide the preview image on the back of the camera when you shoot, but it cannot be extracted from the raw file and doesn't affect the raw conversion.
If you shoot panoramas, it's really important to not use AWB, choose a fixed WB and stick with it across the scene.
On another note:
Sure it can. Check out this thread I posted in the Digital forum, which leads you to a sweet (and free) little utility that can do just that.
I've found that I'm not looking for "neutral" colors as much as I'm trying to preserve the look of the scene as it was shot. Whether the light is warm or cool, Daylight preserves it.
If you want to always have neutral colors, without color casts, start by using the WB presets on your camera for each lighting condition. (You'll find that you're changing WB a lot.)
If you need truly "accurate" color shoot a Color Checker and use the eyedropper to set WB on the second-lightest square. You can then sync the rest of the shots in the sequence.
Rick pointed out a really important point - "neutral" density filters aren't always really neutral. Cokins are very brown and vary depending on the direction of illumination.
I shoot nothing but Auto WB, unless I'm shooting images for stitching. Just curious. I haven't noticed any impact on the quality of my results. Sometimes I have used preset WB settings, but it didn't seem worth the effort.