Daylight White Balance - 5000 or 5500 on Kelvin Scale

[Peter Ryan]

I have been shooting for some time now, in RAW and processing in PS CC. I have studied White Balance and understand it. Shooting RAW I feel by Nikon's auto white balance is a bit cool and generally need to add about 200 on the Kelvin scale to images shot on one light source.

What I had not done was look to see what Daylight setting was in camera verses PS CC. I find that my Nikon considers daylight to be 5000 while PS CC sets it at 5500.

I looked into this further and found that 5000 is the Kelvin reading for daylight at the horizon while 5500 - 5800 is the Kelvin reading for daylight overhead.

So by assuming you took a shot in daylight using the auto white balance setting and just choosing the PS daylight setting does render a considerably different colour outcome than what you would capture with the daylight setting in camera.

[AB26]

Hi Peter,

You are correct in what you have found. There seems to be no real daylight WB that is correct. Depending on the season and time of day the Kelvin temp. will differ.
I have found “daylight WB” to vary from 4800K to 5300K in winter in SA. It also varies depending on the environment you are shooting in.

Good old experience is needed to judge the correct WB. I do depend a lot on the histogram and “chimping” to get the WB to what I want it to be.

I do not think there is a “correct” WB to use – it is dependant on what you want it to be.

[DanK]

Peter,

Interesting. Thanks for posting this information. I didn't know this.

Well, one could split the difference. My Canon 5D3 uses 5200 for daylight. I rarely use it. Like you, I shoot raw. I worry about it in post. If I think I will have trouble later, I take a shot with a whiBal.

Andre, I'm confused by your suggestion. The histogram represents luminance, not color temperature. The appearance of the image on the LCD would give you an idea of the color temperature if you are shooting jpeg, although I don't know how accurate they are.

Dan

[Mike Buckley]

The histogram represents luminance, not color temperature.

True, but luminance is affected by color temperature. That explains why changing the white balance also changes the histogram.

The appearance of the image on the LCD would give you an idea of the color temperature if you are shooting jpeg, although I don't know how accurate they are.

It also gives you and idea if you are shooting RAW. That's because the image displayed in the LCD is the JPEG that is embedded in the RAW file. Even so, the quality of the LCD and the lighting conditions that exist so often when reviewing the LCD are far from optimal. That explains why I would never use the LCD display to critique in a highly discriminating manner anything pertaining to the white balance.

[Manfred M]

Peter - in many shooting circumstances (i.e. pretty well anytime you are not shooting in a studio), any WB is really just a starting point. We've all seen that green colour cast when pictures of people standing under a tree or the reddish cast introduced by the sun bouncing of a red brick building that show up regardless of what the camera or software says....

I find that I will start with the Auto White balance that I get from the camera and then apply different pre-sets from the RAW converter). I use this as the starting point and often end off a few hundred K, as I try to get a look that I like. Sometimes I get lucky and I actually like the WB when I open up the RAW file, and will keep that setting.

Like anything else, WB is not written in stone, and I guess I have been doing this long enough to know what works for me and what does not.

[DanK]

True, but luminance is affected by color temperature. That explains why changing the white balance also changes the histogram.

Good point. However, I still don't see how this would be helpful.

I just took 4 shots of an area that is heavily shaded. There are also heavy clouds. I used 4 different WBs: daylight, shade, cloudy, tungsten. The choice among the first three had a minor effect on the overall histogram. What changing among these settings did is change, as it should, is the relative position of the peaks of the three color channels. Changing from daylight to cloudy, for example, increased red (it shifted that histogram to the right) and decreased blue a bit (shifting that histogram a bit to the left). The net effect on the total histogram, however, was pretty trivial. The change to the tungsten setting was more extreme and shifted the overall histogram appreciably to the left.

However, I don't see how this would help. The position of the overall histogram does not vary much, and I can't see how the modest differences in histograms would tell you anything about correct white balance. The relative position of the peaks of the three color-specific histograms clearly is related to color balance, but what is "correct" in any image depends on the image. In this particular image, which was of a forest with pine needles on the ground, the correct balance had red to the right, but had most of the image been blue, for example, that wouldn't have been true. So even with this additional information, I am left with the conclusion that the histogram is not a useful tool for evaluating white balance.

[William W]

It occurs to me that you were expecting when your camera is in Auto White Balance Mode - AND - you are shooting in daylight that the camera will always render the same Kelvin reading: it won't and it is not designed to do so.

*

. . . So by assuming you took a shot in daylight using the auto white balance setting and just choosing the PS daylight setting does render a considerably different colour outcome than what you would capture with the daylight setting in camera.

Yes. Often it does, because the White Balance is set to 'Automatic'.

*

. . . Well, one could split the difference. My Canon 5D3 uses 5200 for daylight. I rarely use it. Like you, I shoot raw. I worry about it in post. If I think I will have trouble later, I take a shot with a whiBal. . .

I didn't interpret the Original Post to mean that he was using a Manual White Balance setting on the "Daylight" selection, but rather that he was always using Auto White Balance and then noted and compared the Kelvin temperatures of the camera’s automation, when the files were subsequently opened in Photoshop.

Perhaps the exact meaning of the OP could be clarified.

WW

[Mike Buckley]

the histogram is not a useful tool for evaluating white balance.

I wouldn't know. I carefully evaluate white balance only when viewing the image being displayed on my computer monitor. If I decide that a change is needed, even then I never use the histogram to determine that or to evaluate the white balance once I have made the change.

[DanK]

" I carefully evaluate white balance only when viewing the image being displayed on my computer monitor. If I decide that a change is needed, even then I never use the histogram to determine that or to evaluate the white balance once I have made the change."

Exactly what I do too.

[ajohnw]

Peter - in many shooting circumstances (i.e. pretty well anytime you are not shooting in a studio), any WB is really just a starting point. We've all seen that green colour cast when pictures of people standing under a tree or the reddish cast introduced by the sun bouncing of a red brick building that show up regardless of what the camera or software says....

I find that I will start with the Auto White balance that I get from the camera and then apply different pre-sets from the RAW converter). I use this as the starting point and often end off a few hundred K, as I try to get a look that I like. Sometimes I get lucky and I actually like the WB when I open up the RAW file, and will keep that setting.

Like anything else, WB is not written in stone, and I guess I have been doing this long enough to know what works for me and what does not.

Personally I think this hits the nail on the head. It's an odd subject really. Daylight overcast sky is reckoned to be 6500K at noon. Horizon 5000K or clear blue sky directly towards the poles 15-27000K. I'm not agreeing with these numbers just quoting them. Others include clear sky noon 6500K and 5,800K, shade 7000K. Those relate to reflected colour temperatures. To top it all screens are usually set at 6500K but this isn't a true black body type colour temperature only an approximation.

I too feel that a cameras auto white balance is a good starting point and often under even lighting with an even distribution of colour it will get it close to correct. Not sure that an even distribution of colour is the right way of putting it. Sometimes for instance a colour in partial shade will be bang on within the limits of the camera where as the directly lit areas will be way out. That is an entirely different problem to camera white balance setting. Our eyes/brain don't have a problem as they automatically adjust to the scene as we look around. Cameras can't. I think this aspect explains the growing use of some styles of PP. Getting things as they actually were can be rather difficult.

John
-

[PhotomanJohn]

Reading threads like this makes me wonder if I need some help with my thinking about auto white balance. I basically never use auto white balance. My assumption is that to perform auto white balance the camera evaluates the scene based on some algorithm and determines the white balance based on one or more reference scenes within the the algorithm. I assume it is somewhat like when the camera determines the exposure by averaging the scene to neutral which is an over-simplification of how modern cameras determine exposure.

In the case of auto exposure, if one is taking shots of couples all in the same light the camera will use a different exposure for the grooms party in black tuxes versus the brides party all in white dresses. We either handle this with exposure compensation or shooting in manual as I generally do.

Moving this thinking over to auto white balance, I assume that the camera will vary the white balance of a couple wearing wearing red outfits as compared to a couple wearing blues outfits even though the light hasn't changed so the white balance shouldn't change. If this is really the case then the resulting photos will need different compensation in post to correct for the camera using different white balances which would be a pain.

In my case, I always shoot using a fixed white balance and make a custom white balance (and generally a custom profile) when I really want an accurate starting point for post processing. Does this make sense? If not, what is wrong with my thinking? What advantages of auto white balance am I missing.

Thanks,

John

[Mike Buckley]

I assume that the camera will vary the white balance of a couple wearing wearing red outfits as compared to a couple wearing blues outfits even though the light hasn't changed

Theoretically, that should not happen.

In the case of auto exposure...Moving this thinking over to auto white balance

The two processes and their results are not analogous because the two processes are very different. When using auto exposure, a reflective light meter built into the camera is used in conjunction in some cases with with the placement of the auto focus point. When using auto white balance, the reading of the scene is compared with a "database" of scenes and the closest match is assumed to be the proper white balance.

[PhotomanJohn]

I did some reading including the CC tutorial on white balance and everything confirms what I believed to be the case. One of the most concise statements is from Canon:

" Auto White Balance: You can use this setting as a default in most straightforward lighting conditions. Auto White Balance works by evaluating the scene and deciding the most appropriate white point in it. The setting works reasonably well if the colour temperature of the ambient light is between 3,000-7,000K. However, if there is an abundance of one colour in the image, or if there is no actual white for the meter to use as a reference, the system can be fooled, resulting in an image with a colour cast."

So I am still not sure what the real value of auto white balance is for those of us who post process our images.

John

[Mike Buckley]

Sorry, John. Now that I have seen your quote from Canon, I realize that I misunderstood the main gist of your first post explaining your concern about auto white balance.

So I am still not sure what the real value of auto white balance is for those of us who post process our images.

We have to start with some white balance, whether it is auto, preset or custom-defined. If the auto white balance almost always gets us in the ballpark as mine does, that starting point is as good as any other white balance when shooting a scene when we can't control the light source, such as in a landscape scene.

[PhotomanJohn]

From Mike Buckley - [QUOTE=We have to start with some white balance, whether it is auto, preset or custom-defined. If the auto white balance almost always gets us in the ballpark as mine does, that starting point is as good as any other white balance when shooting a scene when we can't control the light source, such as in a landscape scene.[/QUOTE]


Thanks Mike - My point was consistency between shots is important in some situations like wedding photos. I agree with your statement for a general landscape scene. From one shot to the next auto white balance should give you consistent results and if you are only processing a couple of images it really won't matter anyway. For me using a fixed white balance provides a good starting point and assures consistent (whether "right" or "wrong") white balance in every case.

Over the last few years I have become a true believer in custom white balance and then making a color profile using a Color Checker Passport which I always have with me. It only adds a couple of minutes when taking an important shot and saves me more then that time fooling around in the different color channels because some of the colors didn't come out as I remembered them. I think it is definitely worth the $100 investment and something worth trying for those who have not used one.

John

[Mike Buckley]

For me using a fixed white balance provides a good starting point and assures consistent (whether "right" or "wrong") white balance in every case.

Exactly, especially in the situation you described.

As another example, when I'm shooting in my makeshift studio using bulbs rated 5500K, it really doesn't matter what the white balance setting is in my camera. That's because it's so easy to change to a 5500K setting in the first step of my post-processing workflow. It's actually easier to change it in that software than in the camera. The main point, though, is that the starting point is consistent and appropriate for the situation.

[William W]

- My point was consistency between shots is important in some situations like wedding photos. . . etc

I understand that point and I concur

I capture raw + JPEG(L) and from the get go, when we cut the studio over to digital from film I set the White Balance manually, for Weddings and also Studio Portraiture.

I think that there were two main reasons for me doing this.

The first was habit - I was used to shooting wedding in 'lighting sets' and then controlled the Lab Technician's Colour Balance when printing each of those sets so that, especially the B'Maids and Mother's dresses were the same, throughout the Album.

The second was (by accident at first) - I realized that the same principle applied to Digital Post Production - that is time is money - and at first, when we were inexperienced at PP, we employed a specialist and if we gave her batches of files shot in 'lighting sets' she would fly through the Post Production much more quickly and attain continuity of the Colour Balance across the whole wedding shoot very quickly.

I often use a manual white balance now even for casual shooting as it puts THE SAME starting point for Colour Temperature and Tint when I open the raw file of a ‘set of’ images.

I also use AWB. And for the main AWB gets it pretty close and is very useful for ‘one off shots’ – but I do think about whether I want consistency of WB across a set of images in the same lighting conditions, before I select AWB.

WW

[AB26]

Andre, I'm confused by your suggestion. The histogram represents luminance, not color temperature. The appearance of the image on the LCD would give you an idea of the color temperature if you are shooting jpeg, although I don't know how accurate they are.

Dan

Dan,

Mike has explained it.

Thanks Mike.

[Peeshan]

Concerning the "let's care about WB in RAW post-processing": I also used to do that, but actually there is a lot of situations - typically landscapes with grey clouds for example - where finding a nice white balance is almost impossible afterward. I really noticed the difference in a lot of shots between autoWB and custom. Also think about the environemental factor: our brain has it's own "white balance system", so if you're post-processing in the evening with electric lighting - usually "warm white" +/- 3000K - the result may look too warm the next morning in natural light.

For what it's worth, I would advise to consider WB in the same way as aperture or focus that you always check and try to find the more suitable - especially with modern digital SLR, it's rather easy to compare and adapt.

[Glenn NK]

I wouldn't know. I carefully evaluate white balance only when viewing the image being displayed on my computer monitor. If I decide that a change is needed, even then I never use the histogram to determine that or to evaluate the white balance once I have made the change.

That's what I do. And because of this I have never bothered changing the WB settings from Auto. I found that it wasn't worth the bother - I had to change WB in PP.

We've discussed this issue (even beat it to death once or twice). The "correct" WB can very much be a matter of taste - this is particularly true for sunset images. Has anyone ever tried to adjust WB on a sunset using a WB card?

One other comment on a calibrated monitor - it must be used in the light that was used to calibrate it. If you calibrate in the daytime and then do PP in the evening with artificial light, the calibration will be off.

Glenn