Drowning in WB confusion

[divamum]

No, not the principles which (I think) I fully understand, but a seeming inability to adjust a WB/colour cast properly in post!

I'm going through all of Rutt's wonderful posts on this and LAB adjustments (not sure I understand it all yet, but I'm trying to assimilate the information and figure it out), but I still feel I'm missing something more basic than that, and am more than frustrated since so many of my processed images don't seem to be quite nailing this simple but important element.

I'd welcome general pointers (tutorials, info, links, suggestions, good karma, handholding etc) but also have a few very specific questions:

- how do you determine what part of an image is 18% gray, and how do you use that in LR or CS3 to adjust a WB?

- I understand colour temp in theory, but don't entirely see the relationship between it and the other "tint" slider. I'm trying to eyeball things (my monitor is calibrated) but I just never seem to get it right.

- (possibly a slightly diffferent topic, but what the heck - I'll include it here as a related) since in theory flash is daylight balanced, when do you need to take a custom wB using flash outdoors, when do you need to gel the flash with a CTO and when do you just fix it in post? (optimum scenarios, of course - I know that this will seldom be the case, but it's nice to know what would be ideal!)

Thanks in advance. I'm getting SO frustrated with processing at the moment and am determined to figure out a useable - and succcessful - workflow; I'm getting really tired of reprocessing the same images over and over for basic problems that I feel should be an automatic part of my processing habits.

Thanks!

[arodney]

Are you shooting Raw? Because WB is super easy and efficient handling this process at the Raw rendering stage. Trying to fix a rendered image later, not easy, not effective.

[divamum]

Originally Posted by arodney

Are you shooting Raw? Because WB is super easy and efficient handling this process at the Raw rendering stage. Trying to fix a rendered image later, not easy, not effective.

Absolutely - should have mentioned that. I shoot RAW, and have a calibrated monitor. And I'm still struggling

[arodney]

Originally Posted by divamum

Absolutely - should have mentioned that. I shoot RAW, and have a calibrated monitor. And I'm still struggling

OK good. First thing, don't WB on gray! Try a non specular white. Half of all the data in a Raw linear capture is in the first stop of highlight. What converter are you using?

Next, a "proper" WB may not produce an ideal color appearance! I often WB in Lightroom and ACR, then tweak the tint/temp sliders a tad as I usually find the results a bit too cool for my tastes. Also, if you have the proper WB target, that can help as something that's not spectrally neutral will affect the WB. Something like this:

http://www.babelcolor.com/main_level/White_Target.htm

But sometimes, you can't pop this into the shot and you just have to look for something that's lighter than 18% gray, click a few times and, if using an Adobe converter, move the temp/tint slider a tad. Remember, its all non destructive.

[divamum]

I upload my raw images from camera to computer through Picasa (I find its cataloging interface easier than LR so I've just continued to use that even though I now have LR). I shoot Canon, if that makes any difference, so the files are imported as CR2.

I then have both CS3 and LR2 which I can use. I like the easy, non-destructive adjustments I can do on my raw file in LR as well as the many terrific presets out there for it, but if I do my basic processing through LR without fail I'm disappointed when I look at them in another program and have come to the conclusion that if I make my first edits with LR I always wind up redoing them. So I'm now taking to viewing them in Picasa to cull and rename keepers, then opening those keepers individually in CS3 and processing them using the ACR there, and then fine-tuning them in LR.

Short version: my workflow sucks ... which is one of the reasons I'm asking this question because as I'm starting to do more crucial shoots (read: stuff that's for other people rather than just for me) I'm getting overwhelmed and confused. I simply HAVE to figure out a better way of doing this (especially one that doesn't require redoing everything over and over!!)

Thanks so much for your speedy responses - I really appreciate your replies (and look forward to digging into this more)

ETA: Yes, non-specular white. I get that. But sometimes I don't have one... I'll use threshold in CS3 to find white and black points, but sometimes even that doesn't seem to work since when I use that white point in curves it goes NUTS and way off the colour I want. I can't believe something this simple and basic is giving me this much of a headache - I understand it... I just don't seem to do it very well!!

[arodney]

Originally Posted by divamum

I like the easy, non-destructive adjustments I can do on my raw file in LR as well as the many terrific presets out there for it, but if I do my basic processing through LR without fail I'm disappointed when I look at them in another program and have come to the conclusion that if I make my first edits with LR I always wind up redoing them.

OK first off, you're seeing a difference in rendered images OR between Raw processors. The former isn't good. The later is to be expected as all parametric (instruction based) Raw edits are Raw processor specific. Only ACR and Lightroom can share between themselves because they use the same processing engine. All the instructions you make there are not visible in any other Raw processor and vise versa.

If you render say a TIFF and you're seeing a difference, well that's bad and means that other application isn't color management aware.

[divamum]

Originally Posted by arodney

OK first off, you're seeing a difference in rendered images OR between Raw processors. The former isn't good. The later is to be expected as all parametric (instruction based) Raw edits are Raw processor specific. Only ACR and Lightroom can share between themselves because they use the same processing engine. All the instructions you make there are not visible in any other Raw processor and vise versa.

If you render say a TIFF and you're seeing a difference, well that's bad and means that other application isn't color management aware.

We're usually talking LR or CS3, so should'nt those look more or less the same? Hmmmm.. have to think about whether it's all files or just rendered/exported ones.... I usually export from LR as jpgs so that I can upload them - haven't much got into TIFF since theonly time I tried it, when I tried to REopen it later I just got mangled data (saved and re-opened in CS3). But if I open that same exported image in CS3 - which I have always assumed would display exactly as LR since they have the same engines (don't they?) it looks different. Not hugely different, but ... different.

[TheSuede]

Hi :) Try to make sure that all is well with your LR installation and settings... What you say sounds really odd to me. There are plenty of Adobe "settings" tutorials out there for raw-workflow scenarios. Now for the questions... :)

First of all... Basic colour temperature determines the TILT between the three primaries. Low temperatures on the slider amplifies the Blue channel and decreases the Red channel (keeping the Green channel the same). High colour temperatures does the same, only the other way around (+Red and -Blue).
The "TINT" slider keeps the balance between the Blue channel and the SUM of the Red/Green, but alters the balance between Red & Green. This is useful as the light you want to adjust the WB after isn't always "balanced" like daylight (fairly proportional amounts of R, G and B), one channel might be unproportionally strong or weak! - and there you have the need for two sliders.

Secondly... WB should always be done as well as you can IN THE RAW-CONVERSION! This is the only place in the workflow where you have the primaries in "linear mode" intensity. You can create a custom RGB in Photoshop with Gamma 1, but that's rather unwieldy, and also too late - you've already gone through a colour-transform in the raw-conversion.
This is fairly important - if you do WB on a Gamma-corrected colourspace like sRGB or aRGB the shadow- and highlight-zones will not react identically, and this will give you a colour-cast in either of the regions. Lab isn't really ideally suited either, if you wan't to be picky - doing WB in Lab will not give you the same inter-colour lightness relationships as doing it "the right way" will.

Thirdly... Try to explore and research some of the "keyword" functionality in the LR-package, keeping your workflow to a minimum number of applications is very often a good idea. The use of keyword hirearchies in LR is actually quite powerful once you've started to use them.

[arodney]

Originally Posted by divamum

We're usually talking LR or CS3, so should'nt those look more or less the same?

In Develop module, at 1:1, yes, they should match.

But if I open that same exported image in CS3 - which I have always assumed would display exactly as LR since they have the same engines (don't they?) it looks different. Not hugely different, but ... different.

You need to be clear here about what in CS3 you're talking about. It comes with Camera Raw. Camera Raw and Lightroom, when both are the same versions are the same processing engine. LR 2.0 and CS3's version of ACR are not the same, there is newer functionality in LR. If you have the same versions (say CS4 or ACR 5.X and LR 2.0) the two are on parity and you can move from LR to ACR and back.

Exported images are rendered images. That is, you've processed the Raw into a TIFF or JPEG etc. Now you're done with Raw (parametric) editing and you're pixel editing.

So we need to be clear here. ACR is a Raw converter plug-in that ships with Photoshop. Lightroom is also a Raw converter. Photoshop is a pixel editor. You might have the same engine (ACR/LR) or you might not (CS3, LR2).

[pathfinder]

Originally Posted by divamum

- how do you determine what part of an image is 18% gray, and how do you use that in LR or CS3 to adjust a WB?

I agree with Andrew, that using a neutral gray is not ideal, even though it is frequently talked about. There is even a video tutorial showing how to find that neutral gray in addition to your black and white points - http://www.youtube.com/watch?v=Se3htZz2noo

The problem I have is that using this gray to set the center of my curve often does not look any better at all- so this is an interesting skill to know, and may be useful in those images without an obvious black or white - but I really rarely use it. Goes to show one thing I have said here about Photoshop rather often - It is good to know more than one way to skin a cat.

Andrew's Babel Color bracelet does work well - I used it in my discussion of white balancing tools here, seen on the left side of the first image -
http://www.dgrin.com/showthread.php?t=90438 Note that the Babel Color disc is the whitest area in that image, but it is not blown out, and this image is an "straight out of the camera" jpg, not a RAW file that has been edited.

(possibly a slightly diffferent topic, but what the heck - I'll include it here as a related) since in theory flash is daylight balanced, when do you need to take a custom wB using flash outdoors, when do you need to gel the flash with a CTO and when do you just fix it in post? (optimum scenarios, of course - I know that this will seldom be the case, but it's nice to know what would be ideal!)

Most flash is rated at 5600 Kelvin give or take, so out of doors if it is used as fill on a sunlit face, that should be reasonably close.

If the face is in the shade, then the flash may need to be a little bluer to match the ambient blue light from the sky which is definitely cooler ( bluer ) than straight sunlight. Some folks prefer to use a warmer flash anyway for the more pleasant warmer color. What you really have to watch for is shooting faces laying down in the grass - they will be getting a lot of green reflected light. Shoot enough brides on a golf course and you will notice this, I guarantee it!

Use a CTO gel to make the flash match tungsten lighting indoors, or to more closely look like late sunlight from a setting sun. Not always a full CTO but 1/4 or 1/2 CTO gel. For a setting sun you may need even more.

This was shot with a CTO gelled flash in a Lumiquest softbox at 9 am - note the direction of the light and the shadows



A 1/4 CTO can even be used in daylight, with the camera set to a tungsten setting. This will give nice warm color to the flash lit face and drive the blue sky to a very deep dark blue.

You can always shoot a frame with one of the color balancing tools mentioned in the link I posted about color balancing tools - a WiBal card, a Babel color bracelet or a BalanceSmarter reflector. All of these will facilitate color balancing in ARC or LR.

Ultimately color balancing is what you determine it is - not what is "correct" but what you like.

I frequently balance an image, and the as a final step in my editing may make a subtle adjustment in the color balance to make the image look "better" more pleasing to me. Not right, but "better" as defined by my eye!

I think Andrew was saying the same thing about color balance - ultimately it is about creative control.

[arodney]

Originally Posted by pathfinder

I frequently balance an image, and the as a final step in my editing may make a subtle adjustment in the color balance to make the image look "better" more pleasing to me. Not right, but "better" as defined by my eye!

I'd submit, its both better and right.

[pathfinder]

I think we both agree that finally it really is about creative control.

That is easy for me to say as I only have to please myself, and I think my color sense has gotten a bit better over time as I continue to look at and process images.

If I were a wedding shooter, of course, then it is not only my eye that has to be pleased, but the bride, or perhaps, the mother of the bride that has to be satisfied. Whether my eye agrees or not!!

That is why including a color balance tool periodically in the workflow can be helpful and easily pays for itself in speed of processing.

[divamum]

Thank you all for your very helpful comments. Some more questions...

- is there a good basic workflow technique you all could recommend? I realise this is a HUGE subject with as many variants as people, but some generalized, standard approaches might help me figure out where I'm getting lost.

- a couple of examples (part of what started me thinking about all this.

Here's a shot from a couple of weeks ago, taken in a gym. Everybody who saw them here at dgrin said they looked green to them. Despite having a calibrated monitor, I just don't SEE them as green... because that actually is what the gym light looked like. For the record, to me they're slightly too yellow, but the feedback I got was that they are really GREEN. This was processed in LR, using the white sleeve in the bg for WB:



Based on feedback, I corrected it to this... which ALSO used the white sleeve for WB, just a different part of it (based on the feedback about the green, I hunted around to see which parts did what) and addition of the magenta slider:



For me, the silver of the steel is too pink in this second shot, but people preferred the skintones.

Which is why I say I'm confused and drowning because I feel like I can't trust my eyes, my monitor or the software!

Thanks again for continued input - I'm determined to figure out how to do this better

[arodney]

I agree, you probably want to split the difference here. Top is too green, bottom a tad too magenta. Of the two, bottom is preferable but needs a tiny tweak IMHO.

[pathfinder]

Shots in a gym are frequently quite yellow due to reflections from the varnished yellow gym floor, or because of Sodium lights in the ceilings, or ( horrors ) both.

I agree with you and Andrew the first image is definitely yellow. If you use the eye dropper tool in Photoshop you can read the image pixel numbers. When I read the pixel data in the LAB color space, the horn is quite a bit more positive in the b than the a channel and this is definitely yellow, and a silver horn should be near equal in the a and b channels.

I find the second image much better, but possibly a bit magenta as Andrew said, as the horn is now higher positively in the a than the b in the highlights. Interestingly, in the shadows the horn is much more minus in the b channel suggesting the shadows are too blue. Images with different color casts in highlights and shadows can be challenging to the best of image editors, so go easy on your self.

Sometimes when it is hard to decide what is going on with an image after looking at the pixel numbers, it is helpful to wander into Image > Adjustments > Variations which will open an spread sheet style with your current image in the center and arrayed around it, your image with variations in tonality of red, green yellow, blue, cyan, and magenta. You can make any of these stronger or weaker and continue to work your way towards your goal. You can also adjust the image to display highlights, midtones, or shadows to help evaluate these as well.

Variations is an old tool in Photoshop that does not get much discussion, but can still be helpful with some thorny images.

Another trick I use, is to adjust beyond what I know is needed, and then blend it back with the orginal image like Andrew mentioned. In Lightroom I will frequently make note of the As Shot numbers, as well as the Temp and Tint that I find through the White Balance Eye dropper, and then finally adjust the Temp and Tint somewhere in between for what I finally select.

[Thunder Rabbit]

Howdy.

To get Lr and Ps to play nice, you need to go into Lr and Bridge preferences and make sure they both have the proper settings.

In Lr library module, Menu>Edit>Catalog Settings (Ctl+Alt+,). Select "Metadata" tab.

Make sure the "Include Develop settings in metadata inside JPEG, TIFF, and PSD files" box is checked.

Make sure the "Automatically write changes into XMP" box is checked.

In Bridge, Menu>Edit>Camera Raw Preferences. In "DNG File Handling", make sure the "Ignore sidecare ".xmp" files" box is unchecked.

Bridge should now read and render changes made in Lr, if you save (Ctl+S) the changes. I know there's no crying in baseball and no saving in Lr, unless, you want other apps to read the changes immediately. Lr will see changes made in Bridge ACR only if you synchronize the folder in Lr or set Lr to watch the folder.

Looks like you've got plenty of help on the wb issue. But getting Lr and ACR to synchronize properly should end one level of confusion.

Good luck.

[rutt]

Here's what I do:

1. Try to find something neutral in the image to use as the target for the "grey" dropper in either ACR or in a curves layer in PS. Often this will get me on the right planet, but it won't work if the light is very different in this point than the most important parts of your image.
2. Sometimes there is no obvious neutral area to balance on. Then you have a sort of mystery story to figure out. Perhaps there will be a blue horse and a face that's too magenta. That's a good clue that the light is cooler than you thought. It sounds hard at first, but eventually it becomes so natural that people like Andrew do it without thinking about it at all.
3. But if the photograph has a person I check to make sure that the fleshtones are credible. If they are not, it doesn't matter how neutral the tuxedo and wedding dress are, it won't look right. Worry about the fleshtones first.
4. If there is interesting light, you often won't be able to get all the fleshtones right. For example, in theater light, there may strange light. But at least some of the flesh should be on the map.
5. Sometimes there will be mixed casts from different kinds of light. For example, looking into or out of a window, or when window light mixes with indoor artificial light. This is both good news and bad news. It presents a problem which you have to resolve somehow or another. But it also presents an opportunity for dramatic results. Do you want to show the mixed casts or suppress them? It's up to you.

I played with your gym image a little. It does have a mixed cast. You can see this by looking at the boy's hair. Look at all the different colors there. Just open a curves layer and use the neutral dropper on different points in his hair. You'll get very different results (just as you did with different points on the sleeve.) I'm not saying that his hair is black (as opposed to brown.) We don't know that. But I think it is reasonable to assume that his hair is of uniform color and that the difference are differences in light.

So what to do? You did the right first step by getting the face right. I got onto the right planet with the sleeve and then used selective color to add some magenta (remove some green) and remove some cyan (add some red) to the yellows. This made the flesh tones look pretty good to me. Does the silvery brass bother you? You can fix it if you like, but I think you'll have to make some sort of selection or use a brush of some sort to do it. If this were mine, I'd stop before that unless I had a client who insisted.

[divamum]

You guys are nothing short of AMAZING. I need to read through this when I'm fresher than I am after a long day and can assimilate all of this, but you are already starting to get my understanding on-track here. I have no doubt I'll be back with more questions as soon as I've parsed the information you've given me, but for now..... Have I mentioned recently how much I love this place?