White balance question??

[Deleted User]

I shoot in raw and normally use PSE 5 raw converter. I have a couple of question on white balance;

a) I beleive (not sure if I'm correct) that if I have something white (not an over exposed section) in the picture that I should use the eye dropper to set the white balance in the picture from that?

b) Assuming a) is correct what do I do when I have no white, select one of the inbuilt level which match the picture the best?

c) if a) is wrong what should I be doing??

Sorry if this sounds a stupid question

Tim

ifocus

This is not stupid at all. You may want to pick an area as close as possible as neutral gray (18%). Darker green grass will be typically very close. However, I would not worry too much about it, just adjust it manually to level that looks good to you if the camera default is not pleasing. Trust you instinct and your artistic talent.

arodney

draggin wrote:

I shoot in raw and normally use PSE 5 raw converter. I have a couple of question on white balance;

a) I beleive (not sure if I'm correct) that if I have something white (not an over exposed section) in the picture that I should use the eye dropper to set the white balance in the picture from that?

b) Assuming a) is correct what do I do when I have no white, select one of the inbuilt level which match the picture the best?

c) if a) is wrong what should I be doing??

You're on the right track here. First off, in Raw you white balance, not Gray balance so don't go looking for gray! This is linear encoded data, half of all the data in a Raw document is in the first stop of highlight exposure. You want to WB on something that's shy of a specular white. And this is totally subjective and non destructive. You can click and if so desired, move the other sliders (I'm assuming you're using ACR so that would be Tint and Temp) to ultimately produce a color appearance you so desire. Some images will be ineffectively handled with a WB eyedropper (think black cat on coal). Ultimately, just using the tint and temp sliders may do the job as easily as trying to click balance on white.

jfriend

draggin wrote:

I shoot in raw and normally use PSE 5 raw converter. I have a couple of question on white balance;

a) I beleive (not sure if I'm correct) that if I have something white (not an over exposed section) in the picture that I should use the eye dropper to set the white balance in the picture from that?

b) Assuming a) is correct what do I do when I have no white, select one of the inbuilt level which match the picture the best?

c) if a) is wrong what should I be doing??

Sorry if this sounds a stupid question

Tim

You may also want to know that all pictures shot in the same light should end up with the same WB setting. So, if you have only one shot in the light that has a good white reference, you can use the eye dropper on that shot and then use that temperature and tint setting on all the other shots that were shot in the same light.

And, technically accurate WB is not always the best looking for your photos. An image taken in the shade probably will look more natural if it's slightly blue (cooler than neutral). An image taken in the late afternoon sun probably will look more natural if it's slightly warmer than neutral. So, I often use a neutral reference and then use my visual judgement on whether the image looks better with a slight tweak from there.

[Deleted User]

Thanks for the comments guys, now have a better understanding.

Tim

claudermilk

arodney wrote:

You're on the right track here. First off, in Raw you white balance, not Gray balance so don't go looking for gray! This is linear encoded data, half of all the data in a Raw document is in the first stop of highlight exposure. You want to WB on something that's shy of a specular white. And this is totally subjective and non destructive. You can click and if so desired, move the other sliders (I'm assuming you're using ACR so that would be Tint and Temp) to ultimately produce a color appearance you so desire. Some images will be ineffectively handled with a WB eyedropper (think black cat on coal). Ultimately, just using the tint and temp sliders may do the job as easily as trying to click balance on white.

That's quite different from what I usually hear. My understanding is that you are looking for a neutral target (i.e., R=G=B) and the luminosity shouldn't matter. I typicaly use a WhiBal G5 and target the lighter of the gray cards. Works perfectly for me. It seems looking for specular white could still give you a bad WB if you happen to pick a blown point as it will be neutral yet not be able to balance the image properly.

arodney

claudermilk wrote:

That's quite different from what I usually hear. My understanding is that you are looking for a neutral target (i.e., R=G=B) and the luminosity shouldn't matter. I typicaly use a WhiBal G5 and target the lighter of the gray cards. Works perfectly for me. It seems looking for specular white could still give you a bad WB if you happen to pick a blown point as it will be neutral yet not be able to balance the image properly.

Again, since half of all the data is contained in the first stop of linear Raw data, here's where you want to WB (and NOT on a specular white). The 2nd white patch on the Macbeth color checker is about the right.

A light gray in a 1.0 gamma space (level 128 in an 8-bit values) is about an Lstar value of 76. The 2nd white patch on the Macbeth is a bit lighter than this but the 3rd gray is quite a bit darker. So the 2nd white patch is far closer to the ideal mid gray in a linear encoded color space. The 4th gray is closer to the old Kodak 18% gray card and its way, way too dark.

claudermilk

I understand how the data is distributed, but I must be thick here as I don't see the significance for WB. Neutral is neutral. If the point you pick is 240=240=240 it achieves the same thing as say 64=64=64, does it not? I know I don't want to go looking in the lower end as noise levels start coming up and throw things off, but so long as I'm more-or-less in the middle of the histogram, I figure I should be ok. Right?

arodney

claudermilk wrote:

I understand how the data is distributed, but I must be thick here as I don't see the significance for WB. Neutral is neutral. If the point you pick is 240=240=240 it achieves the same thing as say 64=64=64, does it not? I know I don't want to go looking in the lower end as noise levels start coming up and throw things off, but so long as I'm more-or-less in the middle of the histogram, I figure I should be ok. Right?

There's a reason why its called "White balance" and not "Gray Balance" ;-)

You CAN click on a gray if you want to push the issue. When you white balance in ACR/LR, you're controlling the colorimetric definition of the RGB primaries, they lay near white, not gray for one. This is just one reason why White balancing should be done after exposure moves because such moves play a role on the WB. Lastly, and maybe most importantly here, when Thomas Knoll builds his two profiles (D65 and Illuminate A) he used the 2nd white patch on the Macbeth as the sample points! If you attempt to WB on a white that's too white (one or more channels blown out), ACR will actually pop a warning. So again, not specular white, not gray (its too dark, has to few tones in a linear gamma encoding).

dogwood

arodney wrote:

There's a reason why its called "White balance" and not "Gray Balance" ;-)

This isn't entirely accurate. On my 5D, Canon recommends shooting a white target to white balance. However, on my IIN (and my old 20D's and 10D), Canon recommends shooting a gray target to "white" balance.

In fact, quoting from the Canon 40D manual: "Instead of a white object, an 18% gray card (commercially available) can produce a more accurate white balance."

arodney

dogwood wrote:

This isn't entirely accurate. On my 5D, Canon recommends shooting a white target to white balance. However, on my IIN (and my old 20D's and 10D), Canon recommends shooting a gray target to "white" balance.

In fact, quoting from the Canon 40D manual: "Instead of a white object, an 18% gray card (commercially available) can produce a more accurate white balance."

Which only affects JPEGs (gamma corrected), not Raw (suggestion only). And the OP asked specifically about ACR.

arodney

dogwood wrote:

"Instead of a white object, an 18% gray card (commercially available) can produce a more accurate white balance."

A white object (like any gray object) may or may not be spectrally neutral, a so called "Commercially available" Gray card is supposed to be. Had they said "a white card" (implying one that's built for WB), that be a different story. Lots and lots of white objects "look white" but are not as your eye adapts to white.

claudermilk

I'm probably looknig to split hairs, but feel I'm still missing the significance. I do undertand that you don't want to get too dark with the gray for setting WB. But still don't see why a lighter gray is any less usable than a very light, near-white gray. I guess I'll just have to pull out the WhiBal & try it out on the different cards--I've just been using the lighter gray one to this point. Again, I'm probably falling into the overthinking trap I chide others about.

arodney

claudermilk wrote:

I'm probably looknig to split hairs, but feel I'm still missing the significance. I do undertand that you don't want to get too dark with the gray for setting WB.

There's the distinct possibility you'll get poor results!

Look, if you don't want to take my word, purchase and download the ACR tutorial on Luminous Landscape and you'll learn a lot for a little:

http://www.luminous-landscape.com/videos/LR-V1.shtml

Chapter 3, White Balance. Jeff clearly describes why (as I've attempted to) you don't want to use gray!

"The Optimal answer is to use Camera Raw's WB tool, and you notice I selected the 2nd brightest white from the left, in the color checker.... Its not as good to use a gray card, because remember a capture is linear, and there's a lota data up here (pointing to 2nd white), but there's a lot less data in the deeper tones, and you may get skews in your color by trying to do a white balance on a gray".

But look, if you're set in your ways to use the tool incorrectly, despite the advise of those who actually built the tool (Thomas) and other's who have some expertise in this topic, knock yourself out.

ziggy53

claudermilk wrote:

I'm probably looknig to split hairs, but feel I'm still missing the significance. I do undertand that you don't want to get too dark with the gray for setting WB. But still don't see why a lighter gray is any less usable than a very light, near-white gray. I guess I'll just have to pull out the WhiBal & try it out on the different cards--I've just been using the lighter gray one to this point. Again, I'm probably falling into the overthinking trap I chide others about.

If image sensors were perfectly linear in how they measure each channel of color at each level, then it wouldn't matter, as long as the gray in use was truly neutral in tonality.

The awful truth is that cameras and sensors tend to be calibrated better at the higher levels than at lower levels. You probably notice that in the deep shadows, colors can be mis-represented. That is a exaggerated manifestation of the problem.

As such, a bright white, but not too bright white (not blown at standard exposure) tends to be more accurate for measuring "white" balance.

If you are using the sample for just getting close to proper WB, but you will adjust further according to other tones, absolute WB accuracy may not be that important and a gray will do nicely. Certainly some known value is better than nothing.

If absolute color accuracy at all levels is important ("color" balance), then you should include, in your shot, a reference chart like the Kodak chart here which includes color tones at different levels as well as a differential gray scale. Even then, concessions may have to be made.

http://www.d.umn.edu/tma/MungerSite/PicturesSmall/Target.jpg

(BTW, this is a scanned image from someone else and not usable for "your" system. You would need to purchase the chart for yourself.)

arodney

ziggy53 wrote:

If absolute color accuracy at all levels is important ("color" balance), then you should include, in your shot, a reference chart like the Kodak chart here which includes color tones at different levels as well as a differential gray scale.

But I wouldn't use that target an IT8. That's intended for profiling scanners, comes in different emulsion flavors and its real glossy. The gloss alone could be an issue. Also, the colors, and the grays don't have to be spectrally neutral. When you build a scanner profile with this target, a TDF (Target Description File) is supplied that has the measured values. The gray or whites can be off neutral but that's OK as long as the colors and the TDF are in sync.

A Macbeth color checker is ideal if you want color and grays because the grays and whites are spectrally neutral (well pretty much). The pigments used are for the use in photography, not for building profiles.

This is the most spectrally neutral target I've measured (and I've got a pile of white cards and such from manufacturers):

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

And yes, its very white and neutral for a reason based on its intent (to WB Raw).

dogwood

arodney wrote:

Which only affects JPEGs (gamma corrected), not Raw (suggestion only). And the OP asked specifically about ACR.

Well of course it doesn't affect the raw file. Isn't that point of a raw file? It's just the data recorded by the camera without being processed into an image. The WB isn't set until the image is processed. And that could be a tiff or a jpg or a psd, etc.

Anyway, there are plenty of folks (including the camera manufacturers) who disagree with you about using a white target (instead of a gray target) for a custom WB. The link below is for an entire website dedicated to gray targets -- and before you dismiss it as a sales gimmick, take a look at some of the folks involved in the research. Will Crockett is a very well known commercial photographer as are many of the others involved in that site. I've taken week long seminars from Will and can assure you he is a top-notch professional photographer.

http://www.balancesmarter.com/

Anyway, I know what works for me and that's to follow the camera manufacturers advice-- which in many cases is to use a gray target. Just my two cents, but I suspect both Canon and the balancesmarter folks have spent a whole lot more on R&D on this issue than anyone here, including myself.

arodney

dogwood wrote:

Well of course it doesn't affect the raw file. Isn't that point of a raw file?

It affects the quality of the rendered image, most certainly in ACR as asked by the OP (see above).

Anyway, there are plenty of folks (including the camera manufacturers) who disagree with you about using a white target (instead of a gray target) for a custom WB.

Suite yourself. If the logic provided, based on the captured data, the way in which the profiles were built used in ACR/LR and the advise of myself, Jeff Schewe and Bruce Fraser (who discuss this in Real World Camera Raw), let along Thomas Knoll who wrote the applications are not worth listening to in respect to this Raw converter, but instead you just feel you've got a more compelling technical reason NOT to WB on white, go for it.

Will Crockett is a very well known commercial photographer as are many of the others involved in that site. I've taken week long seminars from Will and can assure you he is a top-notch professional photographer.

Please, don't start with me by quoting Will Crockett who used to advise people to set their display profile as their working space in Photoshop or has such preposterously incorrect suggestions about sRGB to the point that his articles disprove him and only suggest he doesn't understand how to read a gamut map!

I'm not here to debate religious beliefs. I've provided sound technical reasons why you should WB, certainly in a product asked about by the OP. As someone who's an alpha tester for both Photoshop and Lightroom, as someone who's spent far more time with Mr. Knoll and company than you or Mr. Crockett and as someone who's done the research, I'm happy to continue to discuss this from the standpoint of sound, technical perspectives but if you wish to simply go by a belief system that makes no technical sense, you're wasting my time and others.

arodney

Which Macbeth from a Raw was WB using the white square and which the gray one?

There are RGB numbers for Gray to help....

dogwood

arodney wrote:

you're wasting my time and others.

Gee-- sorry about that. I thought this was an open forum meant to discuss ideas. Sure didn't mean to waste your time.

But I do find it interesting that you're recommending a different WB method than the folks who actually created the raw formats for their own cameras. To each his own, I guess. I'll continue to follow Canon's advice-- which is to use a gray target with most of their cameras.

arodney

Gee-- sorry about that. I thought this was an open forum meant to discuss ideas. Sure didn't mean to waste your time.

Look, the OP asked about WB in a fixed product, then some came out of the woodwork to talk about gray balancing. That's NOT best practices for all the reasons I've defined and demonstrated. There are all kinds of ideas and methods people use, many based on incorrect assumptions about what's either happening under the hood or they get ill informed advise from people like Will who have a product to sell or don't have elbow time with the people who actually design the software products discussed. They go out and "teach" and the poor student doesn't know they are getting a pile of either marketing BS or just misinformation. So if you want to discuss ideas, lets discuss those based on sound and correct imaging concepts.

There are indisputable points here. One has to do with how the Raw data is encoded. If someone who disagrees can point out a reliable scientific discussion that points out how the majority of data isn't in the first stop of white, and hence is just one compelling reason to white balance there, I'm all ears.

Next is the fact that ACR and its WB tool and the profiles built are based on the 2nd patch of the Macbeth. If someone has some information that disputes what Thomas Knoll has said (or that Bruce Fraser and Jeff Schewe discussing what he said) is wrong, again I'm all ears.

Third, the tool is called White Balance. The Adobe engineers don't name things to confuse users. Now even the Adobe help says you can click on a white or gray. And so far, I've never said you can't click on a gray, I said the best practices are to click on a non specular white. And Jeff discuses why clicking on a gray may skew colors and that's what you're seeing in the Macbeth images above.

IF you charge by the hour, or you like jumping through hoops, or you like clicking lots of times when less clicks should be possible, click on a gray. Its not going to ruin your image (no one said it would). But its not the best place to start, which is what the OP was asking about.

But I do find it interesting that you're recommending a different WB method than the folks who actually created the raw formats for their own cameras. To each his own, I guess. I'll continue to follow Canon's advice-- which is to use a gray target with most of their cameras.

If you can find somewhere they split up the discussion between getting a JPEG or getting a Raw, I'd like to see what they have to say. The WB discussed is aimed at JPEG capture (because it doesn't affect the Raw data at all). Nor do they discuss specifically white balancing in a product called Camera Raw do they? Considering what a WB as defined from the camera manufacturer does when shooting JPEG and what it does (or doesn't provide) shooting Raw, I highly suspect they discriminate the two and if so, lets see that discussion. Nor do they discuss anything I've tried to point out here in terms of the data distribution in Raw, the way profiles were built in ACR or what is pretty clear to be best (certainly better) practices.

Now if you or anyone else can demonstrate when and where gray balancing is going to be better, in terms of how the data is distributed, or how ACR/LR or for that matter any Raw converter is concerned, I'm willing to continue if you'll provide data points based on sound empirical digital imaging concepts. So far, no one has. When one person tried to point out why gray balancing isn't best practices, backed up by sound digital imaging concepts, or the design of a product, or other real experts like Schewe, Fraser or even Danny Pascale @BabelColor, a guy who's actually a color scientist, but instead is given a web site by what appears to be a used car salesman hocking a product I can guarantee isn't spectrally neutral, I get a bit impatient.

Balls in your court. You want to gray balance in ACR or LR, go for it. If you want to advise people the same, it would be useful if you had something a bit more concrete in the recommendation aside from Will or the fact that the camera manufacturer says its OK but isn't at all clear where and on what kind of data.

KED

arodney wrote:

The 2nd white patch on the Macbeth color checker is about the right.

I've read all through this thread (which is 2 pages past this by now), but if you will excuse an aside, how does one reference (or better yet, cross-reference) that 2nd white patch on the Macbeth color checker while, for example, trying to fix WB in real time in Lightroom? Does this have anything to do with LR's pop-up grid when the WB eye-dropper is activated?

Sorry if this is an incredibly stupid question!

arodney

KED wrote:

I've read all through this thread (which is 2 pages past this by now), but if you will excuse an aside, how does one reference (or better yet, cross-reference) that 2nd white patch on the Macbeth color checker while, for example, trying to fix WB in real time in Lightroom? Does this have anything to do with LR's pop-up grid when the WB eye-dropper is activated?
!

Assuming you've got a Macbeth, the process in LR is the same (they share the same processing engine). If you don't have a Macbeth (which is par for the course), the point is, try to WB on something that's close to that value. IOW, not a specular white (you'll generate a warning anyway) but not gray but something that's just short of a pure white.

claudermilk

Man, someone got out of the wrong side of bed. Sorry for asking a innocent question honestly trying to understand WHY. I wasn't arguing, just trying to understand why one practice was better than another. In any case, don't worry, I won't bother wasting your valuable time any more.

arodney

claudermilk wrote:

Man, someone got out of the wrong side of bed. Sorry for asking a innocent question honestly trying to understand WHY.

I said why in post #3, 7 and 9 and 14, along with processed examples. One, maybe two posts should have been enough, hence the issue about "time".

[Deleted User]

Now now, I only asked question, I don't want this turning into hand bags at dawn................

Can we all take one step backward and breath in deeply..........................and out, right now lets go and take some photo

Tim

arodney

draggin wrote:

Now now, I only asked question, I don't want this turning into hand bags at dawn................

You asked a good and appropriate question about white balancing.

Some however wish to "debate" or question the logic of white balancing as defined by creators of the product you asked about and instead feel that doing this on a gray is superior, preferable, or recommended by a camera manufacturer and a guy selling a questionably useful product in "defense" of what is the described best practice (explained in no less than four posts).

Can we all take one step backward and breath in deeply..........................and out, right now lets go and take some photo

Sure. Or those who feel doing this on gray can let us know why although I suspect and would hope they've moved past that concept....