Ad
Ad
Ad
Pages: [1] 2 3 »   Bottom of Page
Print
Author Topic: White Balance - Heureka!  (Read 6828 times)
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« on: December 27, 2012, 12:57:15 PM »
ReplyReply

I had a bright moment and have figured out how to set the natural (visually correct) white balance at shooting time. But before I reveal the secret, you have to guess: which one of these 4 shots of this gorgeous scene has the most natural white balance?
#1 (7296) is the AWB
:-) Hening
Well uploading the images does not work. Not even a single one of 550 kB.
« Last Edit: December 28, 2012, 01:41:27 PM by Hening Bettermann » Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #1 on: December 28, 2012, 01:38:33 PM »
ReplyReply

new attempt uploading images
Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #2 on: December 28, 2012, 01:52:11 PM »
ReplyReply

The right number is the one that produces the appearance you desire. Any Kelvin values you find are a range of color. The same values in one converter can look quite different in another. The values are correlated from something that isn't real but theoretical (the black body radiator). YMMV.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #3 on: December 28, 2012, 02:48:10 PM »
ReplyReply

Hi Andrew

thank you for chiming in.
The appearance I desire is the appearance I saw at shooting time, and I exspect my processing pipeline to reproduce that.
The 4 images above were developed in Raw Developer, my camera profile, linear settings.
These below are processed in ACR, Adobe Standard camera profile, linear settings.
It seems to me that the relative difference between white balances persists quite well through the change of raw processor even with different camera profiles.
Good light - and true color.
Hening.

Oh, I forgot to mention that the Kelvin values were those set at shooting time, not in the raw processor.
« Last Edit: December 28, 2012, 02:57:47 PM by Hening Bettermann » Logged

Tim Lookingbill
Sr. Member
****
Offline Offline

Posts: 1201



WWW
« Reply #4 on: December 28, 2012, 03:49:42 PM »
ReplyReply

The bottom one has the most natural WB. What's the secret?
Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #5 on: December 28, 2012, 04:59:54 PM »
ReplyReply

You are right!

Here's the secret/recipe:
Set WB to manual. Take a shot based on an estimate. Step a little bit back from the tripod, so that you view the camera screen image and the scene at the same time. Adjust WB based on comparing the two. Take new shot and repeat until you are satisfied.

It is so ridiculously simple - yet I have not seen it described anywhere, and nobody mentioned it when I asked about how to set the WB manually, on this forum
http://www.luminous-landscape.com/forum/index.php?topic=30688.0;wap2

WB bracketing may be  be helpful. My Canon 5D2 does not display the bracketed values by the numbers, but 3 units seem to equal 500 K.

If the camera had a physical dial to adjust the WB (as well as live view), shots would not be needed.


Good light - and true color!
Logged

stamper
Sr. Member
****
Offline Offline

Posts: 2788


« Reply #6 on: December 29, 2012, 03:07:36 AM »
ReplyReply

Hi Andrew

thank you for chiming in.
The appearance I desire is the appearance I saw at shooting time, and I exspect my processing pipeline to reproduce that.
The 4 images above were developed in Raw Developer, my camera profile, linear settings.
These below are processed in ACR, Adobe Standard camera profile, linear settings.
It seems to me that the relative difference between white balances persists quite well through the change of raw processor even with different camera profiles.
Good light - and true color.
Hening.

Oh, I forgot to mention that the Kelvin values were those set at shooting time, not in the raw processor.

Personally I don't see the point of this. Andrew's response is to me a valid point. However if you think the idea has merit then in ACR Adobe Neutral would be a better choice than Adobe Standard if you wish to make everything neutral. As to viewing the image on the back of the camera then what you are trying to match is a jpeg rendering of your RAW information, so a comparison won't be accurate. Setting Kelvin values in camera is at best a wild guess as to the scene in front of you and there isn't anything at all accurate about the whole exercise hence the reason nobody has admitted trying it before? Smiley
Logged

mac_paolo
Sr. Member
****
Offline Offline

Posts: 423


« Reply #7 on: December 29, 2012, 03:26:15 AM »
ReplyReply

Here's the secret/recipe:
Set WB to manual. Take a shot based on an estimate. Step a little bit back from the tripod, so that you view the camera screen image and the scene at the same time. Adjust WB based on comparing the two. Take new shot and repeat until you are satisfied.
You must be joking!  Shocked
There's a reason why you never read that anywhere.

Camera display...
1) is often less than sRGB
2) is viewed in suboptimal conditions
3) refers to the style that has been cooke by that specific brand

It's no way a good nor precise way to set the WB.
Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #8 on: December 29, 2012, 11:51:06 AM »
ReplyReply

WB bracketing may be  be helpful. My Canon 5D2 does not display the bracketed values by the numbers, but 3 units seem to equal 500 K.

If the camera had a physical dial to adjust the WB (as well as live view), shots would not be needed.

Assuming you're shooting raw, not so. The WB setting has zero effect on raw data. It is simply a metadata suggestion and again, based on a system of values that simply do not define an exact color! There's no reason to bracket the metadata, it would do nothing to the raw data. There's a reason, at least in Adobe raw converters (and others) we have control over Tint/Temp (WB over Magenta/Green and Yellow Blue color axis). No one value will be correct until you, the image creator decides what looks best and NOT using that pretty awful LCD on the camera which only shows you the effect of WB based on a JPEG. Unless you are capturing a JPEG and you use the manufacture’s converter (which may or may not deal with the metadata 'better' than another converter), you're wasting your time on site futzing with WB settings.

Here's where you need to go to understand why a CCT Kelvin value is a range of colors and that depending on a large number of factors, the same set of values will look pretty different depending on what is deciphering those numbers.

http://www.ppmag.com/reviews/200512_rodneycm.pdf

After that, you may wish to read this piece which explains why you are not getting "accurate" color but maybe preferred color:

http://www.color.org/ICC_white_paper_20_Digital_photography_color_management_basics.pdf
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
bjanes
Sr. Member
****
Offline Offline

Posts: 2822



« Reply #9 on: December 29, 2012, 03:28:27 PM »
ReplyReply

Assuming you're shooting raw, not so. The WB setting has zero effect on raw data. It is simply a metadata suggestion and again, based on a system of values that simply do not define an exact color! There's no reason to bracket the metadata, it would do nothing to the raw data. There's a reason, at least in Adobe raw converters (and others) we have control over Tint/Temp (WB over Magenta/Green and Yellow Blue color axis). No one value will be correct until you, the image creator decides what looks best and NOT using that pretty awful LCD on the camera which only shows you the effect of WB based on a JPEG. Unless you are capturing a JPEG and you use the manufacture’s converter (which may or may not deal with the metadata 'better' than another converter), you're wasting your time on site futzing with WB settings.

The white balance setting has no effect on the raw file, but a white balance value is needed to properly render the appearance of the scene. As an example, here is a screen shot of an image shot with UniWB set on the camera. You can adjust the WB to obtain an image that is pleasing to you, but if you want to recover the actual appearance of the flower you need to know a WB setting that will enable an accurate reproduction of the flower.

The takehome point, is that if the scene contains no recognizable neutral values, you should shoot with a WhiBal or similar card in the scene or take a shot of the WhiBal under the same illumination.

Interested users may download the file here. (warning 38 MB)

Regards,

Bill
Logged
Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #10 on: December 29, 2012, 03:31:57 PM »
ReplyReply

Hi again

I have now read your (Andrew's) article on white point and re-read the ICC white paper.

I am aware of that scene-referred data have to be translated (rendered) to human vision, which is non-linear, unlike the camera sensor. This process of translation will necessarily imply individual judgement, in particular w.r.t. the tone curve. This is in my eyes NOT a carte blanche to whatever arbitrariness. We should try to apply a terminology that at least theoretically enables us to describe the difference between a naturalistic, if by necessity subjective, rendering intent and e.g. an intended change of hues. Using the term "pleasing" for both is confusing.

jpeg
I am painfully aware of, that the in-camera jpeg is only very loosely linked to the raw data - because the camera maker does exactly what you advice me to do: just make it look pretty. I am not exspecting or claiming that my "method" gives color accuracy to the 3rd post comma digit. Nevertheless, with a precision level as coarse as 1,000°K, it could produce an image, that was MUCH closer to the scene than the AWB. Just picking the prettiest might have led me to #3.
The bracketing is only intended for the visual comparison, not believing that it will change the raw data. I will trash the shots after I have decided on the WB.

The magenta-green axis
I am aware of that that this is to be considered. But how much does the daylight change during the day along this axis? In another thread on this forum, which I fail to retrieve right now, I was told that it changes very little, and have since set the Tint in the raw converter to zero.

The difference between different (well, 2 so far) raw converters seems to be far less than the differences between the AWB and my "method".

Let's turn it around: Which are my alternatives?
AWB - see above.
Gray card, WhiBal and the like: Will theoretically make every image look like it was shot at noon.
Using a fixed value of say 5,500 K (which is what I have done lately): Well, #2 of my images is 6,000 K, and it's way off.

Maybe I can put it this way: the in-camera jpeg, despite all its shortcomings, still seems to be a tool that is my best bet in this situation. Better than AWB, and better than my memory after the shooting.

Good light - and true color
 
Bill, this was written before I read your post. I may return later.
« Last Edit: December 29, 2012, 03:35:26 PM by Hening Bettermann » Logged

Hening Bettermann
Sr. Member
****
Offline Offline

Posts: 570


WWW
« Reply #11 on: December 29, 2012, 03:50:26 PM »
ReplyReply

Bill,

I think the gray card method will give accurate color for the flower seen as an isolated object. But I think it will not solve the problem that "panopeeper" describes in the thread linked to in my above post #5: (page 1 in that thread as linked)

"When in the Lower Antelope Canyon, in Arizona, I shot a white card for later WB. When back home, developing the raw images, I tried to WB them based on that shot. The result was horrendeous. Although the resulting color matched the sand (we took a tiny amount in a small plastic bag), it did not resemble at all the scenery as I saw it, because the entire canyon inside was not lit by the "original" sunshine but by the reflections on the walls. Ultimately, I WBd based on my memory.

When shooting in a night club or bowling alley or other place, where the illumination is intentionally unnatural, like black light, should one WB so, that a white shirt become white? That would ruin the mood. I admit I have no idea how such a shot should be WBd if not purely subjectively."

I think that my "method" is a step in the direction to answer that last question. It views and compares the scene as a whole with all the reflections in place.

edit:
I think the gray card method will not show the color of the flower as you saw it at shooting time, because it will change that light into the light that makes the gray card neutral. (of course, if the light was near average daylight, the difference will be small).
« Last Edit: December 29, 2012, 05:31:28 PM by Hening Bettermann » Logged

digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #12 on: December 29, 2012, 04:58:26 PM »
ReplyReply

The magenta-green axis
I am aware of that that this is to be considered. But how much does the daylight change during the day along this axis?

Doesn't matter. It's just another rendering control to produce a desired image appearance. How much does daylight change that affects Clarity?

Quote
Let's turn it around: Which are my alternatives?
AWB - see above.
Gray card, WhiBal and the like: Will theoretically make every image look like it was shot at noon.
Using a fixed value of say 5,500 K (which is what I have done lately): Well, #2 of my images is 6,000 K, and it's way off.

Lots and lots of scenes do not warrant any WB (a model on the beach at sunset). WB that with a card, you get an awful appearing image. Again, whatever the value, it is simply a suggestion by any number of sources and in no way have to be honored even if honoring them would produce similar results in different products (it will not for all the reason's discussed).
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Schewe
Sr. Member
****
Offline Offline

Posts: 5498


WWW
« Reply #13 on: December 29, 2012, 05:18:17 PM »
ReplyReply

The magenta-green axis
I am aware of that that this is to be considered. But how much does the daylight change during the day along this axis? In another thread on this forum, which I fail to retrieve right now, I was told that it changes very little, and have since set the Tint in the raw converter to zero.
Bill, this was written before I read your post. I may return later.

While the magenta/green axis doesn't change a lot in normal blackbody radiators, the response of the magenta/green by a sensor can. Setting your Tint to zero is a mistake because ACR/LR looks at both the K & Tint to normalize White Balance. To have zero Tint is likely to be wrong in 90% + cases...

Really, your best bet for "accurate" vs. "pleasing" white balance is to shoot a known target and use the WB tool to set the starting point. Yes, there are problems if the light falling on the target is different in various areas of a scene...life's a bitch and then ya die, deal with it.

I'm reminded of a lecture Stephen Johnson gave on the Antarctic trip in 2009 (the 3rd of our trips to Antarctica). His whole lecture was about White Balance and he basically through up his hands and said it's impossible to have a technically accurate WB setting in Antarctic light. He shot a variety of "targets) and never got an "accurate" WB setting because, well, light it really, really tricky. His best advice (as I remember it) was to yank on the sliders until the image "looked" like you though it looked. If you know Stephen you would realize that what you though you saw and what you wanted to see and what you ended up capturing was all very different. At some point you need to get over what you though you saw and make the adjustments needed to make it look like what you want it to look like.

The bottom line? Sensors and human eyeballs are different...you can jump though as many hoops as you want, but what really matters is what you see on the display when you are done making your personal adjustments. There is zero chance of capturing what you thought you saw with your eyes...the best you can hope for is something that has a close approximation to what you think you saw. BTW, this all depends on the original illumination and it's luminance...any time you look at dimly lit scenes (such as the example image) your visual color accuracy goes right out the window.

The moment you think you have the keys to the universe, some SOB will change the locks...sorry bud, but you are barking up the wrong tree :~)
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2822



« Reply #14 on: December 30, 2012, 09:02:56 AM »
ReplyReply

Lots and lots of scenes do not warrant any WB (a model on the beach at sunset). WB that with a card, you get an awful appearing image. Again, whatever the value, it is simply a suggestion by any number of sources and in no way have to be honored even if honoring them would produce similar results in different products (it will not for all the reason's discussed).

That depends on what is meant by white balance. If you want the camera to reproduce the actual colors of the scene as accurately as possible, a white balance setting of 5500K is best, because the spectral power distribution for that color temperature is relatively flat, approaching an equal power distribution (Illuminant E). If no white balance at all is applied, the scene will appear quite green because most cameras are most sensitive to green. A white balance setting is needed to balance the output of the color channels.

Shown below is a sunset scene with daylight white balance of 5500K and without any white balance. Also shown are the SPDs of various natural illuminants. The spectrum of D55 is relatively flat over most of the visible spectrum.

Bill
Logged
bjanes
Sr. Member
****
Offline Offline

Posts: 2822



« Reply #15 on: December 30, 2012, 09:19:03 AM »
ReplyReply

Hi again

I have now read your (Andrew's) article on white point and re-read the ICC white paper.

I am aware of that scene-referred data have to be translated (rendered) to human vision, which is non-linear, unlike the camera sensor. This process of translation will necessarily imply individual judgement, in particular w.r.t. the tone curve. This is in my eyes NOT a carte blanche to whatever arbitrariness. We should try to apply a terminology that at least theoretically enables us to describe the difference between a naturalistic, if by necessity subjective, rendering intent and e.g. an intended change of hues. Using the term "pleasing" for both is confusing.

Hening,

It is a misunderstanding that our images are gamma encoded since the human vision is nonlinear (near logarithmic) and the gamma encoding is necessary to accommadate human vision. When the image is shown on the screen, the gamma encoding is reversed so that the final system gamma is close to 1.0. This is explained clearly by Sean McHugh on the Cambridge in Color web site. If the reproduction is to be viewed with a dark surround as in a movie theater, the gamma 1 reproduction will appear flat and an overall gamma of 1.5 is preferable as explained here.

Scene referred images do appear flat when tone mapping is done to fit the luminances to a low dynamic range output device. However, if one has a high dynamic range output device, a scene referred image will reproduce the actual scene. Modern displays do have a relatively large DR and a scene referred will look fine in most cases if one uses the proper profile such as LinearRimm. Otherwise, the scene referred image will appear dark.

Regards,

Bill
Logged
JRSmit
Sr. Member
****
Offline Offline

Posts: 390


WWW
« Reply #16 on: December 30, 2012, 10:20:05 AM »
ReplyReply

Bill, how do you set your camera to 5500k? The numbers on my nikon d700 and d800 do not give identical results.
Logged

Fine art photography: janrsmit.com
Fine Art Printing Specialist: www.fineartprintingspecialist.nl


Jan R. Smit
petermfiore
Sr. Member
****
Offline Offline

Posts: 572



WWW
« Reply #17 on: December 30, 2012, 11:52:25 AM »
ReplyReply

His best advice (as I remember it) was to yank on the sliders until the image "looked" like you though it looked. If you know Stephen you would realize that what you though you saw and what you wanted to see and what you ended up capturing was all very different. At some point you need to get over what you though you saw and make the adjustments needed to make it look like what you want it to look like.



BRAVO!  At some point we need to stop thinking "Perfect" color, and create with our minds eye. Be the Artist and interpret what we Feel and not what just what we see.

Peter
Logged

bjanes
Sr. Member
****
Offline Offline

Posts: 2822



« Reply #18 on: December 30, 2012, 12:17:55 PM »
ReplyReply

Bill, how do you set your camera to 5500k? The numbers on my nikon d700 and d800 do not give identical results.

Since 5500K is not a standard preset, you have to go to the shooting menu and navigate to the white balance tab and select choose color temperature and then set the color temperature as shown. As to the identical results on the D700 and D800, i'm not sure what you mean. If you are in Lightroom/ACR, ignore the color temperature and tint. For my D800e with 5500K and no tint, the ACR reading is 5300K and tint of +6.

Regards,

Bill

Logged
digitaldog
Sr. Member
****
Offline Offline

Posts: 9188



WWW
« Reply #19 on: December 30, 2012, 12:24:19 PM »
ReplyReply

If you want the camera to reproduce the actual colors of the scene as accurately as possible...

In the context of this tread, the use of the term "accurate color" isn't useful and isn't even an accurate term <g>. It only serves to confuse the OP and others. The OP is asking about setting a CCT value on the camera, which has no bearing on the raw data and will be interpreted by differing processes to an extent that the term "accurate" needs to be left out of the conversation. The OP should listen to the words of Steve Johnson who knows a few things about digital capture and image processing! The latest and most salient point the OP should be looking over, instead of digging a deeper rabbit hole in terms of "accurate" color and setting CCT values is this:

Quote
At some point we need to stop thinking "Perfect" color, and create with our minds eye. Be the Artist and interpret what we Feel and not what just what we see.

Peter

Maybe if someone on LuLa sets up a forum for scientific capture of digital images, the discussion would be better there. In this context, creating imagery with color in context, it's just a digression into theoretical verbal masturbation. Make the image appear on your calibrated display (or not, just on the print itself) as you desire. There's no accurate matrix one can assign to that process and for good reasons.
Logged

Andrew Rodney
Author “Color Management for Photographers”
http://digitaldog.net/
Pages: [1] 2 3 »   Top of Page
Print
Jump to:  

Ad
Ad
Ad