Daylight White Balance — What works for you?

[Mark S Abeln]

At midday, in full sun, I usually shoot with my Nikon's Daylight preset, and the white balance of the photos usually look great.

Well OK… they sometimes have a slight cyan cast, and so I will occasionally adjust the white balance in post processing to make them look a bit warmer, or I'll do a manual white balance. But it's good enough.

As I do nearly all my shooting near my home, I've never given any thought to the Daylight preset. It usually works well for me. But the other day I read this thread:

http://www.dpreview.com/forums/post/51975303

where the OP wrote the following:

I was reading Bryan Petersons' Understanding Exposure for the first time, & was surprised to read that he uses the "Cloudy" setting for the majority of his outdoor photography.

My initial gut reaction was that Mr. Peterson must be wrong, or that his computer monitor must be too blue. But I reconsidered when other folks say that they use high color temperature settings routinely. Although now it seems obvious in retrospect, typical midday color temperatures increase with latitude — the lighting is more blue and less yellow, since the sunlight becomes more attenuated and the coverage of the blue sky is larger.

What works for you? I'd be curious if there is a clear pattern or unusual conditions that greatly alter white balance.

• What camera do you use?
• Where do you live?
• What white balance settings do you use at midday?
• What software do you use, or do you use in-camera JPEGs?
• Do you have any other observations on daylight shooting?

 

[Mark S Abeln]

Mark Scott Abeln wrote:

• What camera do you use?

Nikon DSLRs

• Where do you live?

Saint Louis, USA

• What white balance settings do you use at midday?

Nikon's Daylight preset, or I manually adjust in post.

• What software do you use, or do you use in-camera JPEGs?

In Adobe Camera Raw, I'll often use that software's Daylight setting, which is slight different from Nikon's, or I'll set it to a color temperature of about 5500 K and 15 tint.

• Do you have any other observations on daylight shooting?

Someday I'll take more careful records of white balance throughout the day. At sunset I'll often use the Daylight preset to preserve the color of the sunlight reflected off of the clouds. I'll even use Daylight at dusk, to preserve the blue colors of the shadows.

 

[brianj]

Mark Scott Abeln wrote:



What works for you? I'd be curious if there is a clear pattern or unusual conditions that greatly alter white balance.

• What camera do you use? canon SX260
• Where do you live? Australia
• What white balance settings do you use at midday? AUTO WB (5300K)
• What software do you use, or do you use in-camera JPEGs? I use in camera jpg and PSP9 to PP
• Do you have any other observations on daylight shooting? The AUTO WB is close enough to be able to forget about it at all times of the day. That was not the case with my fuji F100fd (cyan cast) or samsung EX1 (greenish skies), so I think canon have got this sorted out.

Brian

 

[Brev00]

Mark Scott Abeln wrote:

• What camera do you use?

Nikon D90

• Where do you live?

New Jersey, one hour west of Manhattan

• What white balance settings do you use at midday?

I keep my wb at daylight pretty routinely and adjust in post. I read Bryan Peterson's book and gave his suggestion a try. It didn't stick for me. My wb moves around to my taste. I often use the white balance dropper in ACR to adjust as well. In general, I keep it cooler than most around maybe 5300 as a guess.

• What software do you use, or do you use in-camera JPEGs?

I process my raw files in the ACR of Elements 9. I find the picture controls have certain color renditions that I adjust with the wb and tint, often adding magenta to Adobe Standard to correct what I see as a green cast.

• Do you have any other observations on daylight shooting?

I find warming up the daylight preset a bit in ACR to be appealing but I usually do not go so far as the cloudy preset, either Nikon or Adobe, unless I am looking for that particular warmth in the shot especially if we are talking specifically daylight and blue sky shooting. Still like Peterson. Reading his guide to composition now. Take what you want and leave the rest.

 

[gollywop]

Mark, I'm not so sure differences due to latitude are so great as to have the kind of effect you suggest on the "daylight" temperature. To be sure, latitude would have some effect, but perhaps not all that much. I can offer a small, and certainly not a systematic, bit of evidence in that direction.

I just got back from a month's cruise that went from New York, to Southampton, Hamburg, Greenock (near Glasgow), Dublin, Liverpool, Invergordon, Stavanger (Norway), and back to Southampton and New York.

I took over 450 shots including lots of WhiBal references. On the sunny days, the daylight temperature was in the 5100° range in Southampton, and closer to the 5200-5300° range in Greenock with one 4950 reading in Liverpool. Unfortunately, our time in Stavanger (the furthest north) enjoyed no sun, but the readings on the cloudy days there were in the 5800° range, far lower than the E-M5's "cloudy" preset of 6500°.

These temperature values are those determined by ACR from the WhiBal reference card. ACR is not necessarily the most meaningful interpreter of light temperature, but its readings are in a proper ballpark. I checked the 4950° ACR Liverpool reading in PhotoNinja, and it gave a 5100° reading. PN tends to read about 100° higher than ACR.

I find that 5050°-5100° is a pretty common daylight range here in the Boston area and is similar to my New York readings. The north of the UK seemed to be only a few hundred degrees higher -- maybe -- but nowhere near the difference that results from using the "cloudy" preset.

--
gollywop

 

[GodSpeaks]

Mark Scott Abeln wrote:

What works for you? I'd be curious if there is a clear pattern or unusual conditions that greatly alter white balance.

• What camera do you use?

Panasonic GX1, Sony A850, Nikon D800E

• Where do you live?

Bangkok

• What white balance settings do you use at midday?

I set the WB on all my cameras to Daylight. Never change it. I shoot only raw.

• What software do you use, or do you use in-camera JPEGs?

Capture One, never shoot jpg.

• Do you have any other observations on daylight shooting?

Slower lenses work just fine at base ISO.

 

[Detail Man]

Mark Scott Abeln wrote:

At midday, in full sun, I usually shoot with my Nikon's Daylight preset, and the white balance of the photos usually look great.

Well OK… they sometimes have a slight cyan cast, and so I will occasionally adjust the white balance in post processing to make them look a bit warmer, or I'll do a manual white balance. But it's good enough.

As I do nearly all my shooting near my home, I've never given any thought to the Daylight preset. It usually works well for me. But the other day I read this thread:

http://www.dpreview.com/forums/post/51975303

where the OP wrote the following:

I was reading Bryan Petersons' Understanding Exposure for the first time, & was surprised to read that he uses the "Cloudy" setting for the majority of his outdoor photography.

My initial gut reaction was that Mr. Peterson must be wrong, or that his computer monitor must be too blue. But I reconsidered when other folks say that they use high color temperature settings routinely. Although now it seems obvious in retrospect, typical midday color temperatures increase with latitude — the lighting is more blue and less yellow, since the sunlight becomes more attenuated and the coverage of the blue sky is larger.

What works for you? I'd be curious if there is a clear pattern or unusual conditions that greatly alter white balance.

• What camera do you use?
• Where do you live?
• What white balance settings do you use at midday?
• What software do you use, or do you use in-camera JPEGs?
• Do you have any other observations on daylight shooting?

Hi Mark,

Use a DMC-GH2 and DMC-LX3. Live at Latitude 47.661. Record in RAW. Daylight (on LX3). UniWB on GH2. Do not use my graycard in scenes for later reference in processing. Typically shoot late afternoon or early evening in non-direct natural lighting. Color temperature thus should always be higher than around 5500 Degrees Kelvin. Tweak WB Temp and Tint to taste in DxO Optics Pro.

Typically use no WB Tint in processing settings (or just plus/minus 1 or 2, which equates to higher numberical values in LR/ACR). My Color Temp preferences range from around 4700 DK (as reported by DxO Optics Pro, see notes below) for dusky light and dusky light look (as long as that does not result in overemphasized Blue hues). Will raise to around 5300 DK if Blue is causing Green or Red hues to appear to have too much Blue mixed in.

May take Color Temp up to 5500 DK or around 6000 DK closer to mid-day light (which is almost always still indirect, not first, reflections). Am not fond of the common Yellowish look or the over-emphasized Reds that are often associated with using higher Color Tempearture references. Sometimes shady environments (in mid-day) warrants up to around 6500 DK. That's about as high as I find myself (usually) going.

Whenever Auto WB modes are used with cameras, what values the camera may come up with would seeem to be quite variable and uncertain. See:

http://www.dpreview.com/forums/post/42346863

Iliah Borg appears to state in this post:

http://www.dpreview.com/forums/post/29765442

... that AWB algorithms use "look up tables" which "guess" as a result of image-data derived and averaged from sets of "typical" image-scenes that various camera manufacturers have used in research. Therefore, it seems that the "Color Temp" that cameras will come up with (and sometimes report in image-file meta-data) when using an Auto WB mode is likely highly variable in many respects.

While cameras may try to somehow approximate the "Planckian locus" applicable to black-body radiators (including the sun) where it comes to Color Temp, and the Green/Magenta Tint may try to approxinmate "isotemperature lines" on an x,y chromaticity chart, (it seems to me that we may) depend upon the camera's applied color matrices (?) and what the particular camera (by whatever means employed) report in the meta-data. It seems that different camera models are likely going to differ in what they report in terms of Color Temp - even in the case where a mode is used where the user can (ostensibly) "dial up" a reference Color Temp ?

Another important point regarding the ("As Shot" or "Camera Settings") WB Temp and Tint that RAW processors report varies. When I set DxO Optics Pro to "as shot" WB, it typically reports a Color Temp value that is around 500 DK lower value than does Lightroom 3.6 (for the same camera settings, and around 5000 DK ranges). Yet RAW Therapee will report yet a third unique Color Temp value (which, according to the data linked to below, appears to be much higher in value).

Color Temps (and Tints) reported by individual RAW processors appear likely to be "all over the map". Timur Born posted this table a while back:

http://www.dpreview.com/galleries/3262748529/photos/2150271/e-m5-wb-comparison

... showing differences between an E-M5's reported Color Temp, and LR, DOP, and RT Color Temp and Tint values reported in various situations in his personal testing. Note how the numerical relationships between DOP and LR "flip" between around 5000 DK and 6000 DK. Nothing simple about what is going on there.

.

So, I am doubting that trying to determine variations (based upon latitude, time of day, as well as the phase of the seasonal cycles at some particular time of the year) is likely to be successful - as the data is likely going to be "swamped" by the (above discusssed) "noise" (that is, the numerical uncertainty) of various camera and processor reported Color Temperature reference values.

DM ...

 

[Detail Man]

Detail Man wrote:

Use a DMC-GH2 and DMC-LX3. Live at Latitude 47.661. Record in RAW. Daylight (on LX3). UniWB on GH2. Do not use my graycard in scenes for later reference in processing. Typically shoot late afternoon or early evening in non-direct natural lighting. Color temperature thus should always be higher than around 5500 Degrees Kelvin.

Actually, I suppose that the average Color Temp of the framed and recorded scene (itself) could indeed be reduced below 5500 DK in the above described times of day and conditions when a lot of Red exists (such as in some of my flower shots), or around sunset in cases where the sunlight is near the horizon and/or reflecting off (or being refracted/diffused by) cloud formations and other atmospheric components.

Would be interested in any resident knowledge on anybody's part that may be informative about such situations (as well as situations where afternoon sunlight is diffused by multiple reflection paths, but not really in what might be termed as being strictly restricted to "shade" conditions).

I find this from a quick search (which indicates that average scene Color temp does indeed drop below 5500 DK in the kind of late afternoon and early evening light that I typically shoot in:

http://voices.yahoo.com/the-color-temperature-lighting-conditions-the-8449595.html

This rings some (admittedly rusty) bells in my head - and also helps me to understand why I often start with an (barring overly Bluish hues) seem to prefer (an usually start at) a WB Color Temp of around 4700 DK when RAW processing in DxO Optics Pro. I tend to like the "Bluish" hues of that (indirect, and nearing dusky) light - and will sometimes use what may be a lower reference WB Color Temp than the scene might measure at. I find that it seems to increase the visibility of fine foliage detail (as long as the Greens, or any Reds do not appear too "Bluish").

Some of my shots include surfaces of water in them (and a few include sky), and I also like to emphasize the "Bluish" hues found there (not being much of a fan of using polarizers, as they seem to impart an unnatural kind of "plasticine" look in nearby foliage details, and also remove reflections from water surfaces that leave shallow creek water looking quite uninteresting).

Tweak WB Temp and Tint to taste in DxO Optics Pro.

 

[Mark S Abeln]

Detail Man wrote:

Whenever Auto WB modes are used with cameras, what values the camera may come up with would seeem to be quite variable and uncertain. See:

http://www.dpreview.com/forums/post/42346863

Iliah Borg appears to state in this post:

http://www.dpreview.com/forums/post/29765442

... that AWB algorithms use "look up tables" which "guess" as a result of image-data derived and averaged from sets of "typical" image-scenes that various camera manufacturers have used in research. Therefore, it seems that the "Color Temp" that cameras will come up with (and sometimes report in image-file meta-data) when using an Auto WB mode is likely highly variable in many respects.

Auto White Balance is not really the issue here, although it is an important if frustrating function. There are lots of AWB algorithms, which all have particular strengths and weaknesses. Now I have no idea what algorithms cameras use, although I bet some have been reverse engineered and so this is worth digging into. Iliah appears to be referring to the 'color by correlation' method which analyzes the colors in an image and attempts to determine what kind of light source could possibly have generated it. I'm pretty sure ACR uses a version of it, but I've not looked at my Nikons' images to see if they might be using it also. I'm not too sure of that, since it always fouls up incandescent lighting, which should be easy for the color by correlation method.

You can tell if a raw converter uses color by correlation if it only produces a fixed small set of white balances and nothing in-between.

While cameras may try to somehow approximate the "Planckian locus" applicable to black-body radiators (including the sun) where it comes to Color Temp, and the Green/Magenta Tint may try to approxinmate "isotemperature lines" on an x,y chromaticity chart, (it seems to me that we may) depend upon the camera's applied color matrices (?) and what the particular camera (by whatever means employed) report in the meta-data. It seems that different camera models are likely going to differ in what they report in terms of Color Temp - even in the case where a mode is used where the user can (ostensibly) "dial up" a reference Color Temp ?

I don't think the cameras produce any additional metadata other than the balances of the various color channels along with the WB preset. Color Temperature — I think — is always reverse engineered by the raw converter. If Tint is along the isotemperature lines, that would be excellent because then it would be an orthogonal coordinate to color temperature, which is as it should be.

Another important point regarding the ("As Shot" or "Camera Settings") WB Temp and Tint that RAW processors report varies. When I set DxO Optics Pro to "as shot" WB, it typically reports a Color Temp value that is around 500 DK lower value than does Lightroom 3.6 (for the same camera settings, and around 5000 DK ranges). Yet RAW Therapee will report yet a third unique Color Temp value (which, according to the data linked to below, appears to be much higher in value).

Sadly this appears to be the case, although it would be consistent within a particular raw converter with a particular camera, which would be an important fact to build on.

Color Temps (and Tints) reported by individual RAW processors appear likely to be "all over the map". Timur Born posted this table a while back:

http://www.dpreview.com/galleries/3262748529/photos/2150271/e-m5-wb-comparison

... showing differences between an E-M5's reported Color Temp, and LR, DOP, and RT Color Temp and Tint values reported in various situations in his personal testing. Note how the numerical relationships between DOP and LR "flip" between around 5000 DK and 6000 DK. Nothing simple about what is going on there.

Yep. For this reason, I asked for what software was being used.

So, I am doubting that trying to determine variations (based upon latitude, time of day, as well as the phase of the seasonal cycles at some particular time of the year) is likely to be successful - as the data is likely going to be "swamped" by the (above discusssed) "noise" (that is, the numerical uncertainty) of various camera and processor reported Color Temperature reference values.

DM ...

The more data the better! I love lots of data, as long as it is easily accessed.

 

[Charles2]

Often use one wb for shadows and another for lit areas. Done in post processing, of course (Picture Window Pro).

Mystery creek

 

[Mark S Abeln]

Detail Man wrote:

Actually, I suppose that the average Color Temp of the framed and recorded scene (itself) could indeed be reduced below 5500 DK in the above described times of day and conditions when a lot of Red exists (such as in some of my flower shots)

The adjusting of color temperature with red flowers may undoubtably happen, but that would be with automatic white balance. I'd assume that better overall color rendering would not take into account the colors of the subject — that is, by using a manual or preset white balance based only on the color from the sky.

Would be interested in any resident knowledge on anybody's part that may be informative about such situations (as well as situations where afternoon sunlight is diffused by multiple reflection paths, but not really in what might be termed as being strictly restricted to "shade" conditions).

I find this from a quick search (which indicates that average scene Color temp does indeed drop below 5500 DK in the kind of late afternoon and early evening light that I typically shoot in:

http://voices.yahoo.com/the-color-temperature-lighting-conditions-the-8449595.html

Yeah, sunset is a special case. My early experience with sunsets was disappointing, because AWB subtracted so much of the red color from the image, but I quickly learned to use the Sunset WB mode. On later cameras I just used the Daylight preset. My D7000 has a Sunset mode, but it too uses the Daylight white balance.

And midday, hardly any photos are taken with the sun directly at the back of the camera — since the sun is so high in the sky. Undoubtably that would be a bit warmer than most photos which are taken with a mix of sun and sky lighting. And so we can say that the color temperature at sunset will be much lower, that is only true if measuring an object directly facing the sun. If we were to measure the color temperature at a subject pointing 90 degrees from the sun, undoubtably it would be much higher.

This rings some (admittedly rusty) bells in my head - and also helps me to understand why I often start with an (barring overly Bluish hues) seem to prefer (an usually start at) a WB Color Temp of around 4700 DK when RAW processing in DxO Optics Pro. I tend to like the "Bluish" hues of that (indirect, and nearing dusky) light - and will sometimes use what may be a lower reference WB Color Temp than the scene might measure at. I find that it seems to increase the visibility of fine foliage detail (as long as the Greens, or any Reds do not appear too "Bluish").

Some of my shots include surfaces of water in them (and a few include sky), and I also like to emphasize the "Bluish" hues found there (not being much of a fan of using polarizers, as they seem to impart an unnatural kind of "plasticine" look in nearby foliage details, and also remove reflections from water surfaces that leave shallow creek water looking quite uninteresting).

I'm undecided to what color temperatures to use during dusk, and how much blue to retain in the image. Too much, and I'm left with very little detail in the shadows, which needs to be brought up in Photoshop to an extreme degree. Completely corrected white balance looks rather ordinary — and it turns the skies too much towards white. I have the same problem on overcast days if I include the sky in the photo. I have to do an extreme amount of Photoshop to bring out texture in the sky if I don't want it to be a featureless white area, and I have to do a lot of work to bring out detail in the shadows. If I exclude the sky, I fully correct the white balance and the photos turn out great.

--

The Refracted Light

A website about the Art and Science of Photography

therefractedlight.blogspot.com

 

[Detail Man]

Mark Scott Abeln wrote:

Detail Man wrote:

Whenever Auto WB modes are used with cameras, what values the camera may come up with would seeem to be quite variable and uncertain. See:

http://www.dpreview.com/forums/post/42346863

Iliah Borg appears to state in this post:

http://www.dpreview.com/forums/post/29765442

... that AWB algorithms use "look up tables" which "guess" as a result of image-data derived and averaged from sets of "typical" image-scenes that various camera manufacturers have used in research. Therefore, it seems that the "Color Temp" that cameras will come up with (and sometimes report in image-file meta-data) when using an Auto WB mode is likely highly variable in many respects.

Auto White Balance is not really the issue here, although it is an important if frustrating function. There are lots of AWB algorithms, which all have particular strengths and weaknesses. Now I have no idea what algorithms cameras use, although I bet some have been reverse engineered and so this is worth digging into. Iliah appears to be referring to the 'color by correlation' method which analyzes the colors in an image and attempts to determine what kind of light source could possibly have generated it. I'm pretty sure ACR uses a version of it, but I've not looked at my Nikons' images to see if they might be using it also. I'm not too sure of that, since it always fouls up incandescent lighting, which should be easy for the color by correlation method.

You can tell if a raw converter uses color by correlation if it only produces a fixed small set of white balances and nothing in-between.

I mentioned this because (in at least the case of the DMC-GH2) a specific Color Temp value tag exists in the "Maker Notes section of the image-file meta-data. My thought then was that some readers might be reporting such a (camera calculated and derived) Color Temperature value.

While cameras may try to somehow approximate the "Planckian locus" applicable to black-body radiators (including the sun) where it comes to Color Temp, and the Green/Magenta Tint may try to approxinmate "isotemperature lines" on an x,y chromaticity chart, (it seems to me that we may) depend upon the camera's applied color matrices (?) and what the particular camera (by whatever means employed) report in the meta-data. It seems that different camera models are likely going to differ in what they report in terms of Color Temp - even in the case where a mode is used where the user can (ostensibly) "dial up" a reference Color Temp ?

I don't think the cameras produce any additional metadata other than the balances of the various color channels along with the WB preset. Color Temperature — I think — is always reverse engineered by the raw converter. If Tint is along the isotemperature lines, that would be excellent because then it would be an orthogonal coordinate to color temperature, which is as it should be.

I see (using ExifTool) that the E-M5 "Maker Notes" contain only the usual RAW RGB coefficients (as well as the coefficients of the applied color matrix). My DMC-LX3 "Maker Notes" (when shot using "Daylight" fixed WB mode) has a a tag called "ColorTempKelvin" (but which has a value of zero in LX3 JPGs and RW2s). Since Timur Born lists E-M5 WB Color Temp values in his data tables referenced, I assume that used a WB mode where the user is able to set that reference value.

However, the "Maker Notes" section of my GH2 JPGs and RW2s has the same "ColorTempKelvin" tag, and does indeed assign a unique numerical value (in addition to listing the usual RAW RGB WB coefficients).

This is the case for GH2 JPGs and RW2s recorded in Auto WB mode ("Standard" Film Mode).

This is also the case for GH2 JPGs and RW2s recorded in my UniWB setup. WB mode is set to on of the 4 "Custom WB" settings - calibrated on a magenta-hued calibration target that results in essentially equal RAW RGB WB coefficients (UniWB). Standard B&W Film Mode used in that case (to avoid the "green meanies" in the preview screens).

How the number is derived, or how accurate is may be, I do not know. I don't pay much attention to its value. RAW Therapee 4.x came up with a much higher ("Camera Settings") Color Temp number (and also comes up with Color Temps that are higher than DOP and LR, as well).

Another important point regarding the ("As Shot" or "Camera Settings") WB Temp and Tint that RAW processors report varies. When I set DxO Optics Pro to "as shot" WB, it typically reports a Color Temp value that is around 500 DK lower value than does Lightroom 3.6 (for the same camera settings, and around 5000 DK ranges). Yet RAW Therapee will report yet a third unique Color Temp value (which, according to the data linked to below, appears to be much higher in value).

Sadly this appears to be the case, although it would be consistent within a particular raw converter with a particular camera, which would be an important fact to build on.

Color Temps (and Tints) reported by individual RAW processors appear likely to be "all over the map". Timur Born posted this table a while back:

http://www.dpreview.com/galleries/3262748529/photos/2150271/e-m5-wb-comparison

... showing differences between an E-M5's reported Color Temp, and LR, DOP, and RT Color Temp and Tint values reported in various situations in his personal testing. Note how the numerical relationships between DOP and LR "flip" between around 5000 DK and 6000 DK. Nothing simple about what is going on there.

Yep. For this reason, I asked for what software was being used.

Just mentioned that because it seeems to make the particular RAW processor application used in deriving the numbers essential to (also) know (specific WB Tint as well as Temp information).

So, I am doubting that trying to determine variations (based upon latitude, time of day, as well as the phase of the seasonal cycles at some particular time of the year) is likely to be successful - as the data is likely going to be "swamped" by the (above discusssed) "noise" (that is, the numerical uncertainty) of various camera and processor reported Color Temperature reference values.

The more data the better! I love lots of data, as long as it is easily accessed.

 

[Detail Man]

Mark Scott Abeln wrote:

Detail Man wrote:

Actually, I suppose that the average Color Temp of the framed and recorded scene (itself) could indeed be reduced below 5500 DK in the above described times of day and conditions when a lot of Red exists (such as in some of my flower shots)

The adjusting of color temperature with red flowers may undoubtably happen, but that would be with automatic white balance.

I was referring (above) to the average color temperature of the scene itself (presumably more accurately measured by instrumentation more reliable than in-camera Auto WB widgets).

I'd assume that better overall color rendering would not take into account the colors of the subject — that is, by using a manual or preset white balance based only on the color from the sky.

Perhaps so. However, in many of my flower shots the flower itself fills a relatively large portion of the image-frame - whereas the background is usually just surrounding floral greenery, etc:

Would be interested in any resident knowledge on anybody's part that may be informative about such situations (as well as situations where afternoon sunlight is diffused by multiple reflection paths, but not really in what might be termed as being strictly restricted to "shade" conditions).

I find this from a quick search (which indicates that average scene Color temp does indeed drop below 5500 DK in the kind of late afternoon and early evening light that I typically shoot in:

http://voices.yahoo.com/the-color-temperature-lighting-conditions-the-8449595.html

Yeah, sunset is a special case. My early experience with sunsets was disappointing, because AWB subtracted so much of the red color from the image, but I quickly learned to use the Sunset WB mode. On later cameras I just used the Daylight preset. My D7000 has a Sunset mode, but it too uses the Daylight white balance.

I gave up on all Auto WB widgets many years ago back in my JPG-shooting days - when it resulted in so many botched JPGs that were hard to try to correct for after the fact in JPG image-editors. Switched at the time to "Sunny" ("Daylight") only all the time in all cases. Now that I shoot RAWs only, it is a non-issue (except where the look of the LCD preview screen is concerned on my LX3). EVF/LCD previews are sich a joke anyway where it comes to color accuracy (preview/review/playback). They saturate the crap out of the color and boost the Blues.

And midday, hardly any photos are taken with the sun directly at the back of the camera — since the sun is so high in the sky. Undoubtably that would be a bit warmer than most photos which are taken with a mix of sun and sky lighting. And so we can say that the color temperature at sunset will be much lower, that is only true if measuring an object directly facing the sun. If we were to measure the color temperature at a subject pointing 90 degrees from the sun, undoubtably it would be much higher.

I was referring to my occasional forays into attempting to photograph sunset scenes (looking into the sun). Even then, I have stuck with "Daylight" (for JPGs). Never tried "Sunset" mode. Whereas it matters not for RAWs, this DMC-LZ5 JPGs was shot using "Daylight" WB settings (and tweaked a bit using PSP's RGB Histogram adjustment - which allows the individual adjustment by "geometric strecthing" of the individual R, G, and B histograms to balance colors):


DMC-LZ5 JPG processed using PSP 9.01

This rings some (admittedly rusty) bells in my head - and also helps me to understand why I often start with an (barring overly Bluish hues) seem to prefer (an usually start at) a WB Color Temp of around 4700 DK when RAW processing in DxO Optics Pro. I tend to like the "Bluish" hues of that (indirect, and nearing dusky) light - and will sometimes use what may be a lower reference WB Color Temp than the scene might measure at. I find that it seems to increase the visibility of fine foliage detail (as long as the Greens, or any Reds do not appear too "Bluish").

Some of my shots include surfaces of water in them (and a few include sky), and I also like to emphasize the "Bluish" hues found there (not being much of a fan of using polarizers, as they seem to impart an unnatural kind of "plasticine" look in nearby foliage details, and also remove reflections from water surfaces that leave shallow creek water looking quite uninteresting).

I'm undecided to what color temperatures to use during dusk, and how much blue to retain in the image. Too much, and I'm left with very little detail in the shadows, which needs to be brought up in Photoshop to an extreme degree. Completely corrected white balance looks rather ordinary — and it turns the skies too much towards white. I have the same problem on overcast days if I include the sky in the photo. I have to do an extreme amount of Photoshop to bring out texture in the sky if I don't want it to be a featureless white area, and I have to do a lot of work to bring out detail in the shadows. If I exclude the sky, I fully correct the white balance and the photos turn out great.

I assume that you are describing actions in RAW processing (as opposed to in-camera JPG recording). I hear you where in somes to sections (clear or overcast) sky in the image-frame. I often wonder how others seem to to well with blue skies (without the use of polarizers). Same kind of thing with blue skies reflecting off water (which I love to see). Both tempt and cause my to boost Bluish hues to the point where it appears overdone, and gives Greenish foliage an unnatural appearance. I love the bluish glow of light near dusk - but it is hard to "pull off" (emerging with all that one would like).

 

[Mark S Abeln]

Detail Man wrote:

This is also the case for GH2 JPGs and RW2s recorded in my UniWB setup. WB mode is set to on of the 4 "Custom WB" settings - calibrated on a magenta-hued calibration target that results in essentially equal RAW RGB WB coefficients (UniWB). Standard B&W Film Mode used in that case (to avoid the "green meanies" in the preview screens).

How the number is derived, or how accurate is may be, I do not know. I don't pay much attention to its value. RAW Therapee 4.x came up with a much higher ("Camera Settings") Color Temp number (and also comes up with Color Temps that are higher than DOP and LR, as well).

I was never able to get my old Nikon D40 to accept a UniWB, and so while I thought it was an intriguing technique, I never pursued it further.

Certainly this whole color temperature and tint business is quite camera and software dependent.

Yep. For this reason, I asked for what software was being used.

Just mentioned that because it seeems to make the particular RAW processor application used in deriving the numbers essential to (also) know (specific WB Tint as well as Temp information).

Indeed.

 

[BackroadsTraveler]

I use a Nikon 7000. I live around 41 degrees north. In daylight, I attempt to set WB to conditions (sunlight when bright, shade or cloudy when appropriate). However, I sometimes forget to reset WB and then need to modify in PP. I use Nikon's View NX2 for most of my exposure related PP, and shoot only RAW.

I discovered that even when the WB was set correctly according to conditions (ie 'Sunlight' for bright or light overcast conditions), I can often get more pleasing and even more realistic color by changing to 'Shady' or 'Cloud'. So, I can see the reasoning behind setting the WB for 'Cloud' when shooting in daylight. If the color modification appears too dramatic, it can always be changed back to 'Daylight'.

 

[Detail Man]

Mark Scott Abeln wrote:

Detail Man wrote:

.... the "Maker Notes" section of my GH2 JPGs and RW2s has the same "ColorTempKelvin" tag, and does indeed assign a unique numerical value (in addition to listing the usual RAW RGB WB coefficients).

This is the case for GH2 JPGs and RW2s recorded in Auto WB mode ("Standard" Film Mode).

This is also the case for GH2 JPGs and RW2s recorded in my UniWB setup. WB mode is set to on of the 4 "Custom WB" settings - calibrated on a magenta-hued calibration target that results in essentially equal RAW RGB WB coefficients (UniWB). Standard B&W Film Mode used in that case (to avoid the "green meanies" in the preview screens).

How the number is derived, or how accurate is may be, I do not know. I don't pay much attention to its value. RAW Therapee 4.x came up with a much higher ("Camera Settings") Color Temp number (and also comes up with Color Temps that are higher than DOP and LR, as well).

I was never able to get my old Nikon D40 to accept a UniWB, and so while I thought it was an intriguing technique, I never pursued it further.

Certainly this whole color temperature and tint business is quite camera and software dependent.

Have had a look at numerous DMC-GH2 shots (and Color Temperature reported in "Maker Notes"):

Standard (B&W) Film Mode: Consistently reports 4700 DK in both Auto WB as well Custom (calibrated for UniWB) modes - completely independent of: the numerical ratios of the Red and Blue WB coefficients relative to Green WB coefficient; as well as the (visually) apparent average color temperature of the scene recorded.

Standard (Color) Film Mode: Consistently reports 4000 DK in Auto WB mode - completely independent of: the numerical ratios of the Red and Blue WB coefficients relative to Green WB coefficient; as well as the (visually) apparent average color temperature of the scene recorded.

I can only conclude that the numerical value of this DMC-GH2 "Maker Notes" meta-data tag named "ColorTempKelvin" is clearly unrelated to the ratio of the Red and Blue WB coefficients relative to Green WB coefficient, seems to be related to the flavor of the "Standard" (color or B&W) Film Mode setting, and is (for all practical purposes) entirely worthless information to rely upon.

Yep. For this reason, I asked for what software was being used.

Just mentioned that because it seeems to make the particular RAW processor application used in deriving the numbers essential to (also) know (specific WB Tint as well as Temp information).

Indeed.

Leaving us with an entirely confusing miasma of differing RAW Processor interpretations of White Balance (Tint as well as) Color Temperature based on (so-called) "As Shot" or "Camera Settings". I think that the saying "so much data; so little information" may here unfortunately apply.

DM ...

 

[happysnapper64]

Hi, Mark. Yes, before I read that quote from B.P. I had never used anything other than AWB, except for a bit of experimenting with the K settings, & of course, flash. I tried a few different modes including the "cloudy" mode, which did work better than AWB on that particular occasion. What I was surprised at though, was how pleasing the use of "flash" WB was. I hope to try it out for real over the weekend.

Camera; 7D

Live in; UK

Use In Camera Jpegs.

PP: Opera, Faststone & Windows Live. I have DPP but not used it yet.

 

[Iliah Borg]

Setting the camera to different "K" values and plotting R/B ratio (based on EXIF) against "K" usually allows to have a spline that can be used "backwards", that is to determine the "intended" K value. Does not work too well for those light sources far from blackbody.

If one has a good camRGB to XYZ transform things are a little more precise. Tables that establish (x,y) for CTs from 2000K to 25000K are easy to find, and not so difficult to calculate too. More advanced methods allow to determine K within 5° across a broad spectrum of light sources, including spiky FLs and LEDs. They are all published.

 

[skyglider]

Mark Scott Abeln wrote:

What works for you? I'd be curious if there is a clear pattern or unusual conditions that greatly alter white balance.

On the digital cameras I've used, the WB changes live on the screen or viewfinder while cycling through the WB settings. So when I enter an environment, I cycle through the WB settings and set the one that looks the best to me at the time.

Strangely, the Panasonic FZ50 has Auto, daylight, cloudy, shade, halogen and flash presets but it does not have tungsten or fluorescent presets. So I set its custom "White set1" for tungsten lighting and "White set2" for fluorescent lighting. These presets allow me to cycle through and find the one that looks the best when entering an environment.

My wife refuses to make any adjustments on her camera. She only wants to point & shoot. So I set her Canon Elph 500HS to Auto WB for shooting in daylight, and to tungsten or fluorescent indoors as appropriate.

Tweaking of WB is done in post if necessary.

Sky

 

[Detail Man]

Iliah Borg wrote:

Setting the camera to different "K" values and plotting R/B ratio (based on EXIF) against "K" usually allows to have a spline that can be used "backwards", that is to determine the "intended" K value. Does not work too well for those light sources far from blackbody.

From this I guess that in addition to Timur Born's published E-M5 and RAW processor data:

http://www.dpreview.com/galleries/3262748529/photos/2150271/e-m5-wb-comparison

... we might want/need to know the spectral nature of the illuminated test-scene photographed. He makes a reference to a "different white target on a different day" on the chart. What kind of test conditions would be required in order that the "dialed-in" CT values would generate a camera output that would cause the resulting ("As Shot" or "Camera Settings") modes of the RAW processors data to as a result be (comparably) informative ?

If one has a good camRGB to XYZ transform things are a little more precise. Tables that establish (x,y) for CTs from 2000K to 25000K are easy to find, and not so difficult to calculate too. More advanced methods allow to determine K within 5° across a broad spectrum of light sources, including spiky FLs and LEDs. They are all published.