Setting White Balance from a Target

[SCoombs]

I am not a beginner, but this seems like an appropriate place for this topic. I have been pretty confused trying to get a good white balance by using a white balance target, either using a white balance dropper tool in processing or setting a manual white balance in camera using a target. It has very consistently given me results that are much warmer than it should. I've gone out and specifically tested this several times over the years to the same results each time, but here I will present only a few examples of photos I still have easily accessible.

First, I would say that on the rare occasion it does seem to work decently.

Here is a photo I took to test at one point, this one with the white balance adjusted by me so the photo looks accurate to life:





Here is the white balance set using the dropper tool on the target. It is very, very close to what Lightroom's Auto WB comes up with:



This is perhaps slightly off from reality to my eye, but it's okay.

In most other cases, the accuracy is not at all what I would expect.

Here is a photo with WB adjusted by me to look accurate:



Here is what I get from the ColorChecker WB, either in camera or using the dropper afterwards:



In this case it's not necessarily awful outside of any context, but it's definitely not accurate to reality, which the first version is much closer to.

Again, my manual WB set to reality:





From the colorchecker:





At one point I had seen several videos suggest using a crumpled up piece of printer paper to set the white balance, so I had tried that.

My edit:





Going off of paper:





Here is the paper I used to set this, with the "correct" WB (to my eye) vs. the WB taken from the paper:







Another attempt during one of my testing efforts:









This is the most egregious example. The others are at least... close in SOME way, maybe they would work for a given style, for instance. Here we just get something completely orange.

Please note: in all these photos I am showing the "official" WB target of the colorchecker, but I have many times done all of the same things using the tabs with the color patches and some additional WB targets there and gotten the same sort of results.

One other thing to add here: in all of the examples other than the boxes of tea, the white balance shown in the "correct" shots was really just Lightroom's auto white balance which I then tweaked slightly, OR the camera's auto WB which I tweaked slightly. This is another whole level to the confusion I have, then: one of the most common - maybe THE most common - bits of advice about photography that can be found from experienced users and experts and educators out there is to not use auto white balance, which they all say will get it wrong, and that it should be set using either the "named" WB (daylight, cloudy, shady, etc.) or ideally using a pre-set white balance read from a target or neutral item in the scene.

Yet in my experience over many tens of thousands of photos, the auto WB gets it very close the vast, vast majority of the time and any time I have ever tried to set using a target or a gray or white element in the scene, I get something that ranges from too orange to much too orange.

I appreciate any thoughts here.

 

[tclune]

Have you calibrated your monitor?

 

[SCoombs]

Have you calibrated your monitor?

Yes, but I'm sure that you can see the difference here in what I'm posting so you really don't have to rely on my own judgment coming from my own monitor.

Thinking about it a different way, I'm trying to take the white balance here off of a specifically designed white balance target here and so even if my monitor were as bad as one could possibly imagine, theoretically I should still be able to achieve a proper white balance by doing this. That's the point of a white balance target: so a person can get an accurate white balance without needing to rely on his own judgment. Yet when I use a white balance target (and not just one, but any number of different ones including but not limited to the colorchecker I've shown here) I frequently get something that is just orange on any device I pull the photo up on - the back LCD of the camera, my calibrated monitor, my other calibrated monitor, my phone, other people's phones, my other computers, other people's computers, etc.

 

[dsjtecserv]

I am not a beginner, but this seems like an appropriate place for this topic. I have been pretty confused trying to get a good white balance by using a white balance target, either using a white balance dropper tool in processing or setting a manual white balance in camera using a target. It has very consistently given me results that are much warmer than it should. I've gone out and specifically tested this several times over the years to the same results each time, but here I will present only a few examples of photos I still have easily accessible.

First, I would say that on the rare occasion it does seem to work decently.

Here is a photo I took to test at one point, this one with the white balance adjusted by me so the photo looks accurate to life:



Here is the white balance set using the dropper tool on the target. It is very, very close to what Lightroom's Auto WB comes up with:



This is perhaps slightly off from reality to my eye, but it's okay.

In most other cases, the accuracy is not at all what I would expect.

Here is a photo with WB adjusted by me to look accurate:



Here is what I get from the ColorChecker WB, either in camera or using the dropper afterwards:



In this case it's not necessarily awful outside of any context, but it's definitely not accurate to reality, which the first version is much closer to.

Again, my manual WB set to reality:



From the colorchecker:



At one point I had seen several videos suggest using a crumpled up piece of printer paper to set the white balance, so I had tried that.

My edit:



Going off of paper:



Here is the paper I used to set this, with the "correct" WB (to my eye) vs. the WB taken from the paper:





Another attempt during one of my testing efforts:





This is the most egregious example. The others are at least... close in SOME way, maybe they would work for a given style, for instance. Here we just get something completely orange.

Please note: in all these photos I am showing the "official" WB target of the colorchecker, but I have many times done all of the same things using the tabs with the color patches and some additional WB targets there and gotten the same sort of results.

One other thing to add here: in all of the examples other than the boxes of tea, the white balance shown in the "correct" shots was really just Lightroom's auto white balance which I then tweaked slightly, OR the camera's auto WB which I tweaked slightly. This is another whole level to the confusion I have, then: one of the most common - maybe THE most common - bits of advice about photography that can be found from experienced users and experts and educators out there is to not use auto white balance, which they all say will get it wrong, and that it should be set using either the "named" WB (daylight, cloudy, shady, etc.) or ideally using a pre-set white balance read from a target or neutral item in the scene.

Yet in my experience over many tens of thousands of photos, the auto WB gets it very close the vast, vast majority of the time and any time I have ever tried to set using a target or a gray or white element in the scene, I get something that ranges from too orange to much too orange.

I appreciate any thoughts here.

To my eye, the images in the first pair or indistinguishable. Note that this is the only sample taken indoors.

One thing to keep in mind is that color is not a linear thing where all the wavelengths are present in equal proportion. The mix of colors (wavelengths) from different light sources will be different, even when the "color temperature" is the same. Outdoor pictures, especially those taken in the shade on a blue sky day or on a cloudy day, are dominated by blue light. Simply normalizing a patch to white or gray will tend to shift everything radically toward yellow and red (orange). I suspect that is what is happening here.

Assuming you are shooting raw, the choice of color profile you make when you bring it into the raw converter will make a difference. The different "named" profiles may give you better or worse results, so try them out to get the best starting point.

I'd suggest you post this in the Retouching forum, if you haven't already. They would be better equipped to explain the nuances of color rendition in the digital process.

Dave

--

Galleries by David S. Johnston

www.pbase.com

 

[SCoombs]

I am not a beginner, but this seems like an appropriate place for this topic. I have been pretty confused trying to get a good white balance by using a white balance target, either using a white balance dropper tool in processing or setting a manual white balance in camera using a target. It has very consistently given me results that are much warmer than it should. I've gone out and specifically tested this several times over the years to the same results each time, but here I will present only a few examples of photos I still have easily accessible.

First, I would say that on the rare occasion it does seem to work decently.

Here is a photo I took to test at one point, this one with the white balance adjusted by me so the photo looks accurate to life:



Here is the white balance set using the dropper tool on the target. It is very, very close to what Lightroom's Auto WB comes up with:



This is perhaps slightly off from reality to my eye, but it's okay.

In most other cases, the accuracy is not at all what I would expect.

Here is a photo with WB adjusted by me to look accurate:



Here is what I get from the ColorChecker WB, either in camera or using the dropper afterwards:



In this case it's not necessarily awful outside of any context, but it's definitely not accurate to reality, which the first version is much closer to.

Again, my manual WB set to reality:



From the colorchecker:



At one point I had seen several videos suggest using a crumpled up piece of printer paper to set the white balance, so I had tried that.

My edit:



Going off of paper:



Here is the paper I used to set this, with the "correct" WB (to my eye) vs. the WB taken from the paper:





Another attempt during one of my testing efforts:





This is the most egregious example. The others are at least... close in SOME way, maybe they would work for a given style, for instance. Here we just get something completely orange.

Please note: in all these photos I am showing the "official" WB target of the colorchecker, but I have many times done all of the same things using the tabs with the color patches and some additional WB targets there and gotten the same sort of results.

One other thing to add here: in all of the examples other than the boxes of tea, the white balance shown in the "correct" shots was really just Lightroom's auto white balance which I then tweaked slightly, OR the camera's auto WB which I tweaked slightly. This is another whole level to the confusion I have, then: one of the most common - maybe THE most common - bits of advice about photography that can be found from experienced users and experts and educators out there is to not use auto white balance, which they all say will get it wrong, and that it should be set using either the "named" WB (daylight, cloudy, shady, etc.) or ideally using a pre-set white balance read from a target or neutral item in the scene.

Yet in my experience over many tens of thousands of photos, the auto WB gets it very close the vast, vast majority of the time and any time I have ever tried to set using a target or a gray or white element in the scene, I get something that ranges from too orange to much too orange.

I appreciate any thoughts here.

To my eye, the images in the first pair or indistinguishable. Note that this is the only sample taken indoors.

One thing to keep in mind is that color is not a linear thing where all the wavelengths are present in equal proportion. The mix of colors (wavelengths) from different light sources will be different, even when the "color temperature" is the same. Outdoor pictures, especially those taken in the shade on a blue sky day or on a cloudy day, are dominated by blue light. Simply normalizing a patch to white or gray will tend to shift everything radically toward yellow and red (orange). I suspect that is what is happening here.

Assuming you are shooting raw, the choice of color profile you make when you bring it into the raw converter will make a difference. The different "named" profiles may give you better or worse results, so try them out to get the best starting point.

I'd suggest you post this in the Retouching forum, if you haven't already. They would be better equipped to explain the nuances of color rendition in the digital process.

Dave

Yes, it seems to work "as advertised" indoors, at least for the most part.

The first pair do look indistinguishable to me here now on my phone. On my calibrated monitor, they look very similar but definitely different.

I can post over there. I had thought that since this was a question not per se about post processing that it may be inappropriate. If a mod wants to move it there, this would be great. Otherwise I will repost later when I have the more efficient controls of my computer.

 

[Ellis Vener]

Which profile are you using in your raw processing application or if shooting JPEGs, which “picture style”?



that can affect color saturation, contrast and biasing towards one color cluster or another. I ask because mostly what i think I am seeing is increased color saturation and color contrast

 

[SCoombs]

Which profile are you using in your raw processing application or if shooting JPEGs, which “picture style”?

that can affect color saturation, contrast and biasing towards one color cluster or another. I ask because mostly what i think I am seeing is increased color saturation and color contrast

I have only ever used the standard, or neutral profiles.

With Lightroom set to import with camera settings, it sets the saturation and contrast of these to zero on import. This isn't saturation - it's just white balance. For instance, here is the most egregious of these with Lightroom's auto white balance and the contrast and saturation then both increased to 50:



Definitely a bit too contrasty and saturated for my liking, but clearly not suffering from the extremely orange color that the white balance target yields (either using Nikon's pre-set white balance or Lightroom's white balance dropper).

 

[Ellis Vener]

The saturation and color contrast I referred to is not a slider setting: it’s part of how the profile tells the program to render the color.

since you have a ColorChecker passport, have you tried creating a custom profile for your camera?

-
Ellis Vener
To see my work, please visit http://www.ellisvener.com
I am on Instagram @EllisVenerStudio
If you like my question or response, please give a thumbs up. My ego needs the strokes.

 

[SCoombs]

The saturation and color contrast I referred to is not a slider setting: it’s part of how the profile tells the program to render the color.

since you have a ColorChecker passport, have you tried creating a custom profile for your camera?

-
Ellis Vener
To see my work, please visit http://www.ellisvener.com
I am on Instagram @EllisVenerStudio
If you like my question or response, please give a thumbs up. My ego needs the strokes.

This behavior is the same for any and all profiles one could choose. This includes all of the camera matching profiles, the Adobe profiles, and any custom profiles created with the Colorchecker. This also happens the same way in other software such as DxO PhotoLab which do not use profiles in the same way. All of this makes perfect sense to me since any of the mapping that a color profile does makes is always going to start off with the colors determined by the white balance as its inputs.

You seem resistant to the idea that what I am demonstrating could actually be happening. I would be happy to share any number of RAW files to help you understand the phenomenon for yourself.

 

[Ellis Vener]

The saturation and color contrast I referred to is not a slider setting: it’s part of how the profile tells the program to render the color.

since you have a ColorChecker passport, have you tried creating a custom profile for your camera?

-
Ellis Vener
To see my work, please visit http://www.ellisvener.com
I am on Instagram @EllisVenerStudio
If you like my question or response, please give a thumbs up. My ego needs the strokes.

This behavior is the same for any and all profiles one could choose. This includes all of the camera matching profiles, the Adobe profiles, and any custom profiles created with the Colorchecker. This also happens the same way in other software such as DxO PhotoLab which do not use profiles in the same way. All of this makes perfect sense to me since any of the mapping that a color profile does makes is always going to start off with the colors determined by the white balance as its inputs.

You seem resistant to the idea that what I am demonstrating could actually be happening. I would be happy to share any number of RAW files to help you understand the phenomenon for yourself.

I am not resistant to what you are demonstrating. I am trying to eliminate as many variables as possible so I can better understand the problem. I do not at all doubt that what you are seeing is what you are seeing.

Since you are seeing the problem occur when processing the same files with different processing software, either the problem is with the target, or it's a difference between what is pleasing color for you and what the target and your processing software programs think is accurate color. I don't see a problem with that. If one way of setting color reliably fails to please, you need to find a way that does.

Along those lines, have you also tried using the color biasing target in the CCPP, which Calibrite calls the Creative Enhancement Target? Calibrite describes there function the following way:

Two rows of warming and cooling patches in the middle serve as a creative guide for you to create pleasing and repeatable edits. Creatively refine the color of your image by adding warmth to skin tones or boost the deep greens or blues in a landscape. Simply click and sync your selection to other images that were shot under that same lighting to create pleasing edits – it’s that simple!

A row of clipping patches across the bottom serve as a visual reference for judging, controlling and editing images for shadow details or highlight clipping. The clipping patches are separated into two groups: light and dark. Each patch is 1/3 F-stop difference between them with the exception of the last black patch. The exposure difference between the darkest and next darkest patch is approximately 1/10th of a stop, and the dynamic range of the target is about 32:1 (5 stops).

The top HSL (hue, saturation, lightness) row includes 8 spectrum patches to ensure color fidelity across all hues so that you can evaluate and edit for any color shifts.

I would be happy to share any number of RAW files to help you understand the phenomenon for yourself.

If you think it would help me better understand, I'll be happy to take a look. If you use WeTransfer or a similar service, send them to [email protected]

I also use the ColorChecker Passport II target set as well as the Calibrite ColorChecker Digital SG target, https://www.xrite.com/categories/calibration-profiling/colorchecker-digital-sg

 

[Michael Fryd]

It's important that the white balance target be lit the same as the subject.

Consider a photo of a girl outside, with the white balance target tilted up towards the sky. The girl may be lit by indirect light from the surroundings. The white balance target may have more of a component of direct sunlight. This might be enough of a difference to throw off the white balance.

 

[SCoombs]

The saturation and color contrast I referred to is not a slider setting: it’s part of how the profile tells the program to render the color.

since you have a ColorChecker passport, have you tried creating a custom profile for your camera?

-
Ellis Vener
To see my work, please visit http://www.ellisvener.com
I am on Instagram @EllisVenerStudio
If you like my question or response, please give a thumbs up. My ego needs the strokes.

This behavior is the same for any and all profiles one could choose. This includes all of the camera matching profiles, the Adobe profiles, and any custom profiles created with the Colorchecker. This also happens the same way in other software such as DxO PhotoLab which do not use profiles in the same way. All of this makes perfect sense to me since any of the mapping that a color profile does makes is always going to start off with the colors determined by the white balance as its inputs.

You seem resistant to the idea that what I am demonstrating could actually be happening. I would be happy to share any number of RAW files to help you understand the phenomenon for yourself.

I am not resistant to what you are demonstrating. I am trying to eliminate as many variables as possible so I can better understand the problem. I do not at all doubt that what you are seeing is what you are seeing.

Since you are seeing the problem occur when processing the same files with different processing software, either the problem is with the target, or it's a difference between what is pleasing color for you and what the target and your processing software programs think is accurate color. I don't see a problem with that. If one way of setting color reliably fails to please, you need to find a way that does.

As I can always adjust any photo's WB to achieve a color that I find more pleasing, I am not really concerned with my own preferences. I am more concerned with accurate color.

Many times I have gone ahead and spent 30 minutes to an hour carefully studying scenes to get a good sense of what they look like, setting the WB off of a target in camera, taking a photo, then right then and there comparing the photo off of the rear LCD to what I am looking at, then comparing the EVF to what I am looking at, then putting it on my calibrated monitor right away while the correct look is still fresh in my mind and looking at it there. In doing so, I always see the same drastic differences from accurate color in the photo, whereas if I use auto white balance I see something that looks very much like reality.

I have also taken products from major brands which have strict pantone standards for their brand colors and photographed these in different light/WB so that there would be a sure frame of reference for people to know what the original colors really were. Unfortunately I can't find any of those right now and I may have recently deleted them. If I am able to find the time today or tomorrow I may do a few more such shots so I can share some photos of things that people will know the real color of.

It's important to be clear here that this is not just my Calibrite target, but any of several white balance targets I have or using any number of countless white or neutral gray objects as a target.

Along those lines, have you also tried using the color biasing target in the CCPP, which Calibrite calls the Creative Enhancement Target? Calibrite describes there function the following way:

Two rows of warming and cooling patches in the middle serve as a creative guide for you to create pleasing and repeatable edits. Creatively refine the color of your image by adding warmth to skin tones or boost the deep greens or blues in a landscape. Simply click and sync your selection to other images that were shot under that same lighting to create pleasing edits – it’s that simple!

A row of clipping patches across the bottom serve as a visual reference for judging, controlling and editing images for shadow details or highlight clipping. The clipping patches are separated into two groups: light and dark. Each patch is 1/3 F-stop difference between them with the exception of the last black patch. The exposure difference between the darkest and next darkest patch is approximately 1/10th of a stop, and the dynamic range of the target is about 32:1 (5 stops)

The top HSL (hue, saturation, lightness) row includes 8 spectrum patches to ensure color fidelity across all hues so that you can evaluate and edit for any color shifts.

Yes, I have used those as well. At one time I took a large number of photos specifically testing each patch. Again, I am not really concerned here with aesthetic. I think these patches offer an interesting and useful way to try to achieve different aesthetics. Unfortunately, they are all dependent on the "correct," neutral WB being correct.

Indoors and in direct sunlight, these also work as advertised. In other kinds of light, these all uniformly give me very orange results as well.

I would be happy to share any number of RAW files to help you understand the phenomenon for yourself.

If you think it would help me better understand, I'll be happy to take a look. If you use WeTransfer or a similar service, send them to [email protected]
I also use the ColorChecker Passport II target set as well as the Calibrite ColorChecker Digital SG target, https://www.xrite.com/categories/calibration-profiling/colorchecker-digital-sg

 

[SCoombs]

It's important that the white balance target be lit the same as the subject.

Consider a photo of a girl outside, with the white balance target tilted up towards the sky. The girl may be lit by indirect light from the surroundings. The white balance target may have more of a component of direct sunlight. This might be enough of a difference to throw off the white balance.

True, but in this case the color of the shirt would for example be correct, which it isn't in the too-orange output. Additionally, I have used these targets next to the face with the same results.

 

[Michael Fryd]

It's important that the white balance target be lit the same as the subject.

Consider a photo of a girl outside, with the white balance target tilted up towards the sky. The girl may be lit by indirect light from the surroundings. The white balance target may have more of a component of direct sunlight. This might be enough of a difference to throw off the white balance.

True, but in this case the color of the shirt would for example be correct, which it isn't in the too-orange output. Additionally, I have used these targets next to the face with the same results.

Is it possible that your targets are not actually neutral? Perhaps it's time to try a number of different "neutral" objects in a frame, and see how they compare.



When using white paper (or clothing), be aware that these likely contain optical whiteners. These glow with a blue tint under UV light (sunlight and some strobes do contain some UV light).



Many laundry detergents contain optical brighteners to make your whites "whiter than white".

 

[bobnn]

It's important that the white balance target be lit the same as the subject.

Consider a photo of a girl outside, with the white balance target tilted up towards the sky. The girl may be lit by indirect light from the surroundings. The white balance target may have more of a component of direct sunlight. This might be enough of a difference to throw off the white balance.

I think that, for this image at least, that's what's happening. Looking at the RGB numbers in RawTherapee, the white target and especially the gray have extra blue in them. (The white probably would too, but it's clipped, blue tops out at 100%.) That is not the case for the window frames, nor for the whites of her eyes. Those things are pretty much right on (excluding the stained horizontal member of the leftmost window), and using them with the dropper doesn't have the orange cast, it doesn't actually change much.

For what it's worth, the camera AWB and/or the OP have done a great job.



I do understand that OP has many other cases of this with varying targets and target positions.



BTW, very cute little girl.

 

[Ellis Vener]

You might be interested in trying the Datacolor LightColor Meter. At $399.00 it’s a relatively inexpensive colorimeter (relative to the ~$1700 Sekonic c-800U spectrometer) designed for reading light sources from ambient daylight to various artificial light sources, including flash, photographers and film/video makers will encounter.

so far I have been using one to match RGB LED panels to indirect daylight, and to create in-camera custom WB settings, including the various auto WB settings, for my Nikon Z6III camera. My tentative conclusion is that by setting the camera’s WB to the color temp and M/G tint for the key (primary) light reading from the LCM, and adjusting the RGB panel, which I read separately, to also match the key light, I feel like I am getting both more accurate and more pleasing color as judged by eye on my calibrated and (using a Calibrite Display HL) BenQ SW display. .



one of my next steps will be to see how closely those readings correlate or differ from using a WB target illuminated by various types of single lights (flash, LED daylight white, Bi-RGB, and daylight) and which I prefer.

Color is truly one of the the areas where the science of the devices meets the expressive aims and ends of photographic art.

--
Ellis Vener
To see my work, please visit http://www.ellisvener.com
I am on Instagram @EllisVenerStudio
If you like my question or response, please give a thumbs up. My ego needs the strokes.

 

[Bill Ferris]

and adjust in post, as needed, also by eye.

 

[jthomas39]

(sorry, I haven't read all the posts above.)

I imported these two images, one with white balance via the paper, the other edited to your satisfaction. Neither paper surface is strictly white, but both are fairly close.

I used darktable, and deleted all the normal processing that's applied to incoming images (using the history stack) and used it's nice Color Picker tool. That tool can either be a spot tool or an averaged drawn rectangle.

~~~

Your preferred white balance image, cropped showing the white balance selection area.

This selected area is red 231, green 233, blue 236. Normally, selecting white balance from a white or gray spot will have all three values the same. This has more blue than that.

Foliage in the background: slightly blue-green. Skin tones seem a little too cool for me.



~~~~

The automatic white balance version:

the area is red 237, green 236, blue 233. Closer to a pure white, but a little less blue.

The foliage in the background: more olive-yellow, which I like. but the skin tones aren't quite right here.

 

[jthomas39]

White balance card

I don't normally use a white balance card. A portrait in shade has bluish light from the sky, or a greenish cast from light filtering through the trees. One in late day sunlight has more red, of course. I don't usually want them both to look like they were illuminated by a white photography floodlight. (I will likely warm up the shade scene a little, though.)

I might use a custom white balance in the camera if it's a night shot with streetlights that have a strong color cast. Or if I ever wanted to record an accurate object color for "documentation" purposes.

While editing, if the colors look off, I sometimes use the color picker to see what a white balanced scene looks like, picking a neutral gray spot in the image. That's often a good starting point for some minor white balance changes.

My Nikon often shows foliage in sunlight with a slight blue-green cast. Increasing the white balance temperature moves it toward olive-yellow, and that's what I see by eye if I'm looking carefully at the scene.

(The extreme example is shooting a tabletop scene by candlelight. A white balance card makes the scene look like white LED lighting. Kind of weird to see that in the viewfinder!)

I often alter my white balance slightly, but it's by taste for what I like there.

 

[A74Me]

I never use a WB card outside ,there is to much reflective colour. i only use WB card in my studio setting up different lights. my studio has no reflective colours just neutral whites and greys. your edits look good to me.