White Balance

DxO PhotoLab Feature requests
1

How is it possible? How can the White Balance be so different?
When I set in my camera Canon 5DS the WB to 5500 K. Then I see in PL 5292 K en tint - 9.
In Capture One it is 5374 K and tint 7,4.
When I set Daylight in PL I see 5200 K and tint 0. In CO it is 5102 K and tint 6,8.

When I set in my camera Canon 5DIII the WB to 5500 K. Then I see in PL 5126 K en tint -14.
In Capture One it is 4995 K and tint - 4,4.
When I set Daylight in PL I see 5200 K and tint 0. In CO it is 4768 K and tint - 4,7.

All the difference is camera and program related. Oké. But when I set in the program Daylight there is a difference in CO between Canon 5DS and Canon 5D III. But in PL it is the same 5200 K and tint 0. That I cannot understand.

2

Colour temperature values differs from software to software.
Can we do anything about it? Should we do anything about it? Should we even care?

From a technical point of view, answers would be 3 x yes, from a point of view of usability, art and getting the result we want, the answer is 3 x no (imo). While it can be interesting to know how manufacturers get the values they get, it’s of minor concern for daily use. I’ve stopped worrying about such things, it saves me a lot of headaches.

2 Likes
3

@Joanna posted a while ago an image of a gray card with a preset wb. When that image was opened in Pl the histogram did show 3 peaks next to each other and a different wb as was set. Changing that wb to the value of the preset did show a histogram where these 3 peaks did overlap, as it should.

George

4

Thank you, Platypus for your answer. You are right, headache, but I asked myself why is the White Balance ‘Daylight’ of Capture One, so it looks, camera related and of PL not. That is not rational.

1 Like
5

DxO team replied to this point.
I have reproduced it at the end of this article.

Pascal

1 Like
6

But that doesn’t explain what we saw with the image of @Joanna. Joanna, could you post that picture again? I can’t find it anymore.

George

7

Just read your interesting article.

You mention 5200 Kelvin as DxO’s Standard, but as if to me I’ve seen 5400 Kelvin as their starting point
– or is that something else?
(Only asking as normally I don’t care / have set my cam to Auto WB – except for sunset and such.)

8

5400K on bodies and other software.
Pascal

9

I’m not sure about an image but the only thread I can find is in a beta forum

10

@Pieloe
I am not sure I understand well what is said there.

So my question is :

Does DxO camera tests done when creating camera profiles take care of those kind of specificities (different white balance interpretation between brands and models) and tries to faithfully follow specific camera color interpretation nor not ?

An other way to ask this would be :
Is DxO trying to provide camera profiles that respects as much as possible camera native colors when cracking - doing reverse engeenering on - Raw code (those colors the constructor choosed for it’s camera - for instance what is called nikon specific colors by some nikon users).
Or do they apply some rules that fits they software and their way to see how colors have to behave without trying to respect the “character” of the cameras ?

11

You did shoot a grey card with a preset color temperature based on you color temperature meter. And was upset that pl didn’t take the values of your preset.

George

12

The interest of using an independent software is to unify the styles.
Pascal

13

This is your point of view.
Few people work with different brands (as you know, good lens are expensives !).
So few people need to unify styles.

Anyway this respond to my question. I probably have to give a new chance to nikon softwares.

14

Do you mean this message?

I just did a test shot of a grey card outside under a mixed cloudy and blue sky, with the light coming from the SSE:

  1. Measured the light on my Minolta ColorMeter III F at 5800°K
  2. Set the WB on my Nikon D810 to 5800°K
  3. Took the shot and opened it in PL4, with only the lens module corrections

On first opening, the “As Shot” WB was shown as


Capture d’écran 2021-09-21 Ă  12.15.47

And the histogram showed


Capture d’écran 2021-09-21 Ă  12.16.06

So, I adjusted the WB to 5800°K - ±0

The result of the WB pipette is that the RGB values are 120 : 120 : 120

Capture d’écran 2021-09-21 Ă  12.11.17

So, PL is definitely giving a false impression of the WB when relying on the “As Shot” setting

1 Like
15

This one, yes. Do you have the image self still? Or can you make a new one based on Minolta ColorMeter? And post it off course.

George

16

I don’t have the original and the problem is it would be a 45Mpx RAW file, which will not post here.

17

I’ll try to make one myself. But I’ve to search for my gray card. And use the colormeter of the camera.

George

18

I checked a picture of a grey card that I have. It seems that it is camera dependent, some profiles are not properly calibrated.

For example, I took a picture of a grey card with a Nikon D750, white balanced with the internal colormeter. When I load the picture into Photolab using the as shot white balance, the histogram looks fine, all color peaks align. Also manual white balance using the color picker delivers similar results (it would be nice if it was possible to choose a radius greater than 50 pixels to average a greater area):

nikon
nikon1907×353 326 KB

This is with the Photolab profile for the D750. When I use my own calibrated DCP profile, the results are similar.

However, when I pick a different profile of other cameras, there are shifts in the histogram peaks, which in my opinion should not happen. It only happened to a few profiles though, most prominent with the profile for the “Canon 1DMK3, 1DsMK3”:

canon.PNG
canon.PNG1877×390 30.4 KB

Nevertheless, I do not see such big issues as in your example.

1 Like
19

Could you share that image?

George

20

What about D850 ? Did you check with it ?