White Balance Temperature and Camera Profiles

[Steve Petermann]

I just finished doing some camera profiles using the X-Rite ColorChecker Passport system. I did them for direct sun, shade, and studio flash. Now, for the studio flash I think this should be consistent. However, I did this several times for the outdoor lighting and found some variation, I think because of the changes in the light. So, I was wondering if, to be more precise, I could just run the test several times under various lighting conditions and try to get a good sample of tests over the various color temperatures for outdoor light. Then when I do the white balance in Lightroom, I could just use the color temperature it showed to decide which camera profile was the closest. I was wondering if this would work.

[Manfred M]

I just finished doing some camera profiles using the X-Rite ColorChecker Passport system. I did them for direct sun, shade, and studio flash. Now, for the studio flash I think this should be consistent. However, I did this several times for the outdoor lighting and found some variation, I think because of the changes in the light. So, I was wondering if, to be more precise, I could just run the test several times under various lighting conditions and try to get a good sample of tests over the various color temperatures for outdoor light. Then when I do the white balance in Lightroom, I could just use the color temperature it showed to decide which camera profile was the closest. I was wondering if this would work.

I run with two profiles only, one with studio lights and one with outdoor lighting. Both give me near identical results. I've never found a need to do additional profiles.

White balance should not really play into this at all as the software analyses raw data (in *.dng format), and raw data has not WB data associated with it. What you are really doing is profiling sensor performance under different light sources; the wavelengths that sunlight produces are there, regardless of lighting conditions / time of day. If I were shooting under a different light source (LED or KinoFlow fluorescents) , I would run a profile of each of those as well.

[Steve Petermann]

Ok, here's what's confusing me. From my reading it is suggested that for outdoor light, a profile should be created for both direct sun and open shade. So, I did that. Now if I open a ColorChecker image in Lightroom, do the white balance, and then select the camera profile what I see is a noticeable shift in colors depending on whether I choose the direct or shade profile. So, unless I'm misunderstanding this, the camera sensor reacts somewhat differently depending on the lighting conditions.

Now maybe I'm being too nit-picky but I thought I'd try to get the colors as close as possible as a starting point. I've read that professional photographers use the Passport on each shoot presumably so they can do a profile for that particular lighting situation. Now, I won't be doing that but since I'll be doing a lot of outdoor photography, it did make me wonder how all this works.

[Manfred M]

Steve - you are confusing two different but related functions of the ColorChecker. If you look at your ColorChecker, you will notice there are three different targets. Two of them are used when setting your white balance and the third one (the one with 24 equal sized sample, is used to profile your camera in conjunction with your ColorChecker software.

1. Creating a camera profile - this is somewhat like using an i1 or ColorMunki to create a profile for your computer screenThe run the software and create a profile that your editing software uses. Once you've done that, you will have calibrated and profiled your screen. Do that once a year or so (to account for colour changes as your screen ages).

Profiling a camera is a similar operation. . This means putting the target in the light source, taking a shot in raw and convert it to a *.dng file. The ColorChecker software will use that shot to create a profile for your camera. When using either Adobe Camera Raw (ACR) or Lightroom (LR), choose the appropriate camera profile, rather than one of the Adobe Standard ones. You can see I use two profiles that can be seen on the dropdown box:



That is the process I wrote about in my previous post. All you are doing there is comparing how your camera's sensor performs under different light sources.


2. The second use of the ColorChecker uses one of the other targets, either the solid middle gray or the card that is visible beside the profiling target. These are used by photographers to do a white balance. This is what you are writing about when you ask about photographers using them in their shoots. This is where they place the target near the subject so that the same light that is hitting the subject also hits the target and use this as their reference shot. Every time the light setup is changed, another reference shot is taken. This technique is used by both portrait and product photographers, but not landscape photographers.

The middle gray target is easiest to explain. When you open your target shot in ACR or Lightroom, click the target and that gives you a neutral white balance. The other card is used very much the same way, but gives you the ability to use a WB that is not 100% neutral. I do that in my portraiture work, as I tend to prefer a very slightly warmer white balance when shooting people. That's what the two central rows of this card give you - standard off-white balance.

Just to be clear on the two targets in the image below - the one on the right is used for creating your camera profile and the one on the left is used for tweaking your white balance.




I hope this clarifies things a bit

[Steve Petermann]

Right, I'm aware of all that. That's exactly how I did my profiles and shoot. When I start a shoot I use the white balance card for white balance. Then in Camera Raw I'll pick that photo and set the white balance. Then I'll pick the camera profile and sync all the images for that situation. However, I've seen some photos of photographers using the color panels on a shoot like these even on the x-rite site:
http://www.xritephoto.com/images/pro...shoot_roof.jpg
http://blog.xritephoto.com/wp-conten...1/NurKasih.jpg

I don't understand why they would be doing that if they weren't going to create a profile. I guess they could be using some of the patches to make adjustments but it doesn't seem like a pro would need that.

[Manfred M]

I don't understand why they would be doing that if they weren't going to create a profile. I guess they could be using some of the patches to make adjustments but it doesn't seem like a pro would need that.

The camera profile maps how the camera responds to a particular light source. Sunlight emits a very specific spectrum, a flash tube emits a very specific spectrum, tungsten light emits etc. This spectrum is nothing more than than a map of the quantity of light at each specific frequency. It doesn't matter at what time of day you shoot, whether it is cloudy, sunny, just before sunset or just after sunrise, the light that the sun emits does not change. Atmospherics, the Raleigh effect, etc. affect the colour temperature of the light, but that's. These effects can filter out parts of the spectrum, but if the frequency wasn't there to begin with, it cannot be added back. If you have a proper calibration for your camera for the full spectrum, you can be reasonably sure that the colours have been mapped correctly by the profile you have created. Put a filter on your lens or shoot at twilight, the colour response will still be accurate.

A similar story can be made for any other light source, with one notable exception. All artificial light sources change over time. Flash tubes yellow, tungsten light bulbs spectral characteristics change as their filaments thin over time, so these light sources need to be recalibrated from time to time.

Could you do a new profile every time you shoot, of course you can, but I think you will find that in many cases it is a complete waste of time as the differences are so small that other issues (notably colour temperature), are going to make a much bigger difference in your image quality.

The two shots you link to, I strongly suspect the one with the woman would work well with an existing daylight profile, but there might be something to be gained in the one with the two men in the "unusual" lighting situation, but I suspect that any gains would be quite insignificant.

Let's also put this into context. If you are a commercial photographer shooting editorial or advertising material that is going to print, this much attention to detail is going to be totally irrelevant. Portrait customers are going to want a specific look. Editorial images are going to be colour graded for a particular look and feel. If you are shooting for a commercial client, the colours of the product had better be bang in. That can of Coca Cola better match the Pantone specs exactly and the colour of the brown liquid had better be bang on too.

If you are working on those commercial shoots, the edits will be done in a dimly lit room and on an Eizo screen with a full hood to ensure nothing interferes with getting the colours bang on. There will be a colour managed workflow (of which camera calibration and shooting a reference target will take you to a calibrated and profiled screen and if the output is printed, the work will be done against either a custom printer / paper profile or one that uses one supplied by a third party (like the paper manufacturer).

If the output is going to be the internet - who cares. The minority of the viewers with their low end commodity sRGB screens that are set up to be really bright will ever notice if the colours are 100% accurate, because their equipment can't reproduce it that way.


I do use my profiles, but frankly the difference between what I get between the daylight one and the studio light one (using my Paul C Buff Einstein 640 lights) is negligible. The colours are definitely "brighter" than what I get when I use the standard Adobe profile in ACR.

All I can suggest is that you try different profiles and see which ones make a tangible difference to your work. When I first started I had about 6 profiles I made, but now I am down to two. If I ever use my studio lights in action mode rather than colour mode, I will probably run another profile just to see if it makes any difference.

[Steve Petermann]

Wow! Fantastic info. Thanks so much.

[mikesan]

It doesn't matter at what time of day you shoot, whether it is cloudy, sunny, just before sunset or just after sunrise, the light that the sun emits does not change. Atmospherics, the Raleigh effect, etc. affect the colour temperature of the light, but that's. These effects can filter out parts of the spectrum, but if the frequency wasn't there to begin with, it cannot be added back. If you have a proper calibration for your camera for the full spectrum, you can be reasonably sure that the colours have been mapped correctly by the profile you have created. Put a filter on your lens or shoot at twilight, the colour response will still be accurate.

I would question the assertion that the light conditions at the time you shoot the CC card are totally without influence when creating a daylight profile. However, if one uses the Adobe DNG Profile Editor to create the profile, these variables are automatically "corrected".

From Adobe Labs:
When running the Chart Wizard, PE will automatically use the light gray patch in row 4, column 2 of the ColorChecker Chart to set the white balance before creating color adjustments.

I don't know if this is also the case when using X-rite software to create the profile, but would expect it is.

[Manfred M]

With Adobe DNG editor (and I suspect the XRite software too), it is possible to create a "dual illuminant" profile which encompasses two profiles,

That would be a "yes".


The problem with any dual illuminant is that you are looking at a situation where you are choosing a "synthetic" value that is incorrect for either of the two illuminants. Is it better to get one correct and the let the other fall where is does?

We really are splitting hairs here and the easiest solution for many of the issues is shoot in a studio where this becomes a non-issue. I'm pretty sure this works well except for the most extreme cases.

[mikesan]

A note to beginners, particularly those shooting outdoors: Having a satisfactory daylight dcp profile for your camera does not obviate the need for proper white balancing of individual shots (or group of shots all taken under the same daylight conditions). As Manfred mentions, the ColorChecker middle grey card, when included in the shot, makes that very easy.

[DanK]

An excellent thread. My question picks up on Manfred's last post. I routinely use a spectrally neutral card (I use a whiBal) for white balance, but I have never used a color checker to profile my cameras (currently a 5D III and a 7D). I wonder how much difference it makes. I understand that it is pearls before swine for web viewing, but I do print, and so I wonder whether I should bit the bullet and profile my cameras. Any thoughts?

[Manfred M]

An excellent thread. My question picks up on Manfred's last post. I routinely use a spectrally neutral card (I use a whiBal) for white balance, but I have never used a color checker to profile my cameras (currently a 5D III and a 7D). I wonder how much difference it makes. I understand that it is pearls before swine for web viewing, but I do print, and so I wonder whether I should bit the bullet and profile my cameras. Any thoughts?

Dan - I profiled my camera because I wanted to see how much of a difference it would make to the final product. Being a "techie" I tend to do things like this.

Most people (including a number of commercial photographers I know) have never done so and most Photoshop users probably never look at anything other than the default "Adobe Standard" profile and have been pumping out lots of great looking images over the years.

In my view it has a minor, but noticeable positive impact on the image colour.

Here are two identically processed image - all of the edits are "native" ACR / Lightroom edits with no additional work in Photoshop and no plug-ins. White balance is identical in both.


Adobe Default :





Custom Daylight Profile:

[Donald]

Just to add another dimension into the mix, which I don't think has been mentioned above and is pretty irrelevant to the discussion. But I'll mention it anyway.

I profiled my Canon 5DS and 7DMkII for the purposes of establishing their dynamic range, with the profiles being loaded into my Sekonic light meter. I have found that since paying more attention to dynamic range and the where the clipping points are set within the cameras, I have hardly ever resorted to using a GND. As I previously reported, I also find that my exposures are now much more accurate and I don't engage in any trial and error type exposure activity.

[DanK]

Manfred,

Thanks. Very interesting. I can clearly see the difference on a calibrated monitor. It looks like one of the effects is a shift on the magenta-green axis.

I think I will go ahead, buy a color checker, and create a few profiles. It's not a lot of work to do.

Dan

[Steve Petermann]

White balance and Camera Colour profiling are two distinct effects - WB deals with variation in spectral distribution of different light sources whereas Camera Colour profiling deals with the response of the individual colour channels of the camera.

From what I read the white balance deals with the blue/yellow shifts due to the lighting conditions. Apparently the Tint adjustment usually isn't relevant unless the light source is a gas type like florescent.

[Manfred M]

Manfred,

Thanks. Very interesting. I can clearly see the difference on a calibrated monitor. It looks like one of the effects is a shift on the magenta-green axis.

Dan - there is something very strange going on in those two images. I see virtually no colour differences when I look at them on my calibrated and profiled wide gamut screen, before and after I convert them from ProPhoto (my default colour space) to sRGB. However, once Flickr finishes processing them, there is a very distinct colour difference there. I get the same issue with TinyPic.

When I put both images on the same canvas (top one is Adobe Standard and bottom one is Custom Profile), there is significantly less difference between the two images.



Obviously there is something going on behind the scenes in the online image processing causing this. This is closer to what I see on screen.

[Steve Petermann]

Another thing I'm starting to do routinely is shoot the white balance card for a particular lighting situation (outdoors mainly) and then set the custom white balance on my camera. Then if I shift to another lighting condition, I do it again. I think this makes post processing a bit easier because I don't have to scroll through the pictures to sync up the ones for the lighting situation. I can just use the "as shot" white balance.

[DanK]

Dan - there is something very strange going on in those two images. I see virtually no colour differences when I look at them on my calibrated and profiled wide gamut screen, before and after I convert them from ProPhoto (my default colour space) to sRGB. However, once Flickr finishes processing them, there is a very distinct colour difference there. I get the same issue with TinyPic.

When I put both images on the same canvas (top one is Adobe Standard and bottom one is Custom Profile), there is significantly less difference between the two images.



Obviously there is something going on behind the scenes in the online image processing causing this. This is closer to what I see on screen.

Manfred,

That's really a puzzle. I don't use TinyPic, and I stopped using Flickr, and I never looked into the processing in the latter case. I post to Smugmug, but the only thing I have found about their processing is that they apply some sharpening unless you set the site not to.

Dan

[Dave Humphries]

Hi Manfred,

I am wondering what exactly you/we're seeing, or what you/we think you/we're seeing ...

When I alternately view the two images above (hosted on Flickr), in LyteBox, when I flick between them, my eyes instantly see a difference in the green-magenta axis, but the difference fades away after a second. This will be my eye adjusting to the new 'colour temperature' (the human eye is good at this), or it could also partly be the way the transition works in the LyteBox.

What I am saying is that what you perceive as a difference will also be affected - and by how much depends upon many factors; the speed and method of image transition, which may well vary between the fade we see in LyteBox and the jump-cut you may (?) be using when switching between on your computer in PS or whatever.

We must all recognise the 'afterglow/shadow' effect when viewing a high contrast simple subject, then changing our eye's position to view a blank white or black surface - we can still see the outline of the image and if the high contrast simple subject was a bold colour, we'll see the inverse of this colour - well the same thing happens even when the subject isn't simple and the contrast isn't high, but mostly our brains don't recognise it.

As for trying to spot the difference in the two images viewed simultaneously and moving your eye, I suspect that so much else is changing, it masks the real difference.

I feel this is a situation where measurement is required; I assume that if you sample the RGB values of a single pixel image point in the above two photos, you will get a different set of numbers. Whether they'll be different depending upon where the image is hosted is debatable, since there are so many other variables that may also be changing.

That's probably all I can add to this thread

Cheers, Dave

[dje]

An excellent thread. My question picks up on Manfred's last post. I routinely use a spectrally neutral card (I use a whiBal) for white balance, but I have never used a color checker to profile my cameras (currently a 5D III and a 7D). I wonder how much difference it makes. I understand that it is pearls before swine for web viewing, but I do print, and so I wonder whether I should bit the bullet and profile my cameras. Any thoughts?

Dan my experience with my D610 has been similar to that described by Manfred in post 13 and I quote

I profiled my camera because I wanted to see how much of a difference it would make to the final product. Being a "techie" I tend to do things like this.

Most people (including a number of commercial photographers I know) have never done so and most Photoshop users probably never look at anything other than the default "Adobe Standard" profile and have been pumping out lots of great looking images over the years.

In my view it has a minor, but noticeable positive impact on the image colour.

One thing to bear in mind is that there is an Adobe Standard Camera Profile for every camera supported and the one relevant to the camera used for the image being edited is automatically selected by LR/ACR. So the camera profile you produce yourself accounts for variations between your camera and the one of the same model that Adobe used for their measurement. It would seem that there may not be a lot of variability from one camera to another of the same model but that may depend on the brand and the model.

It is still just a starting point for your editing too. You may want to manipulate the colours a bit for "creative reasons".

I can't really see that it should have any bearing on printing. The main issue with colour management and printing is that you get on the print what you expect based on what you see on the monitor screen.

One other thing to remember about Adobe profiles is that they contain a tone curve which gives the image more "life". This is really quite separate to the colour calibration side of things but it is included in the dcp profile. This tone curve can be modified in the Adobe DNG profile editor (free software). I actually use a slightly toned down version because i feel that the default version tends to make the images look a bit "washed out".

Dave