Color tint red and yellow

Get it calibrated? :wink:

Incredible photo and nice editing!

Hey Keith,

thank you for you posting in this thread! I am blown away by all the detailed answers I got in this short amount of time!

I like to add a few things to your answer and my post:

I got my monitor second hand from a Gamer friend of mine. He had calibrated this monitor to suit his gaming habit so he adjusted color to be more saturated, contrast and brightness to be higher. I started editing pictures and was disappointed by the results since something was off. I then calibrated my monitor and it got better. Now i noticed some slight color differences between my preview in PL3, exported JPEGs AND previwes of websites after uploading them. I then researched the problem to no avail and then I posted it in this blog. I think Greg found the problem: my monitor K°-value is too low.

Now I like tto share what i found out in my own research for this problem:

If you want to print out a large photograph/poster/etc. you should check the website of your print service provider. The colors you see on your display depend on the ICC file your monitor is calibrated on. So it is possible there is a difference between the colors you see on your screen and the colors that get printed out. Good print provider will have an ICC-file with which their printer is calibrated. If you want your foto to look exactly like on your screen you should export it with the ICC file you found on your print provider’s website.

BUT most printers will use sRGB so normally there shouldn’t be a problem anyway.

Hi Joanna,

I have been reading your posts but couldnt answer because my posts still need admin approval.

I think you were on the right track to solving my problem with White Balance but I didnt get it right away. :slight_smile:

As Greg pointed out I might have the wrong K° value on my monitor…calibrated or not :thinking:

G’day Greg

I think you might have cracked the code/found my problem!

My Monitor might be lacking red becaue my K-value is too low. So my preview looks good but in reality(sRGB) it is too red. As soon as i export it as JPEG with sRGB color profile reds are too saturated.

I will try to recalibrate it and/or raise the K-value or “the reds”

Thanks, Mate :grin:

Hi Florian. When exporting to JPG, do you have your ICC profile set to sRGB ?

If not then that may explain why you’re seeing a difference in colours.

John M

Hi John

I exported it always on sRGB but I started experimenting after I noticed this problem (tint) a few times.

Florian

ok did a [new test] and i think i was distracted by the slow changes because this test shows a steady RWB of 4532K/7 the old video did too but the 5400k/0 start everytime and the changes between confused the view.
there is still variations in what’s Raw’s WB numbers but that’s just inconvenient.

The 5400K is an initialization value of pl. Just a figure.

George

You may correct me. But something Johanna said made me thinking.
The color temperature is what the picture has. And it is used during the conversion to gain a picture without distracting color cast. Let’s assume that will be a picture with a temperature of 6500, much used here. So a raw with a color temperature of 3500 will have to move to 6500 some way during the conversion. That would explain some other issue. In converters a high temperature is red and a low temp is blue, just the opposite as it is. Well, if I give that image temperature of 2500, then the converter will correct that image to much and become blueish.

The color temperature slider and list do address the raw data, not the converted image. It just sais : tread this data as having a color temperature of … . Probably, just a guess, that 5400 is what pl uses as their goal to archive.

It doesn’t make sense to connect the color temperature in the converter to a color temperature of the monitor.

George

Congratulations, you’ve got it.

Edit: Actually, the color temperature of the monitor matters and the software needs to transfer the color space of the picture to the color space of the screen, so “white” in the picture looks “white” on screen and all other colors have to match as well. To make this reliable calibration and monitor profile come into the game.

I can not agree more today.
I had such a long and frustrating way trying to calibrate everything from camera to printer and not succeeding to be happy. Until last year, during a printer workshop. The “teacher” explained us: “in the end, you print what makes YOU happy”, “because (except in particular and very professional cases) you will never be able to have it exactly how it really is”.

@Joanna Thank you for all the tips you share with us.

Ok what you say is:
1 i set wb 5500K/0 in camera so the exposure is aiming to that lightwavelenght as exposure target. And tagging the rawfile as this is shot with 5500.
2 dxo reads: ok ,we use 5400K as central white point in our colorspace and change white point accoording to the tag if the rawfile.
3 then we, dxo, change that whitepoint result for the monitor colorspace in 6500K. Without changing the wb whitepoint of the jpeg file which would be exported on that moment.
Because my monitor has 6500K as default and this doesn’t change in WB setting not in sunlight or rainlydaylight or lightbulb, morning light. To view the carefull converted rawfile based on the tag 5500K.

Can i assume that the 1000K difference 6500-5500 , is fixed and when i change manual in dxo in wb in say 7800K my screen wil be 8800K?(simplified assumming thatvthe colorspaces of srgb are the same

In that case indeed a monitor of 5500K is better then the default 6500K.

@gregor and all the others,
Could you explain to me why a high color temperature is red and a low color temperature is blue?? It’s the opposite of the definition of color temperature.
In my thoughts the result of the demosaicing is predefined in the converter. It’s done towards that white point in the diagram. I don’t think you can change that. What you change is the interpretation of the raw data resulting in an over correction of the colors to the opposite.

From Johanna post 21.
The problem with auto white balance, like auto exposure, is that it always tries to “average” everything. So a “warm” sunset will be cooled down and a “cool” light will be warmed up.
It’s not only the nature of auto white balance, it’s the use of color temperature.

George

Low flame temp is red yellow high gasburnerflame is hot blue.
Maybe connected to that?

1 Like

Not sure why you reply to me because I don’t say any of this.

What I say:

  • Set WB in camera to your liking or try to be exact with a custom white balance for the lighting conditions at time of shooting by using a neutral grey/white target.

  • Calibrate and profile your monitor or, if it is a known good calibration from factory, use the factory profile for the monitor. This is not relevant for the image, but to make sure any kind of exported or printed image looks like you saw it.

  • In post processing, if you don’t need “correct” colors, use WB and tint sliders a creative tools to your liking.

  • When exporting use sRGB color space since it is the most compatible one.

I think there is a lot of misleading and confusing advice in this thread.

If you can read German (or browser translation can help) there is an excellent site explaining all of this in a correct way:

In English I can recommend Cambridge In Colour:
https://www.cambridgeincolour.com/tutorials/white-balance.htm
https://www.cambridgeincolour.com/color-management-printing.htm

If you want to go really deep in your understanding of color in photography I can recommend the lectures by Mark Levoy:

2 Likes

Color temperature is based, simplified, on the color of the glow of a hot object. Think of a piece of metal: If you heat it up it starts to glow dark red. The hotter it gets the glow changes from orange over yellow until it becomes a blue glow at really high temperatures. The definition is based on a so called black body, something that doesn’t exist in reality.

Unfortunately this goes against our intuition of reddish colors being “warm” and blueish colors being “cold”.

The WB value in the RAW file refers to the color temperature of the light source which was used to illuminate the scene. For a flash that is usually well known (unless you use color gels), otherwise it is guess work by the camera or a rough setting based on “shadow”, “daylight” etc. or somewhat exact measurement based on a neutral grey/white target (custom WB).

The WB slider (as well as the tint) in DxO PhotoLab allows you to change this assumed value after the image was shot. This is because the values recorded in the RAW file are not affected by the WB setting in the camera.

So you were spot on when you wrote:

Now, if you set the WB temperature lower than the actual illumination (regardless if in camera or in the software), the assumption is that you had a relatively reddish light source, e. g. Tungsten. The color transformation to the output compensates for that by making the picture more blue.

If, on the other hand, you set the WB to a higher temperature (e. g. shadow), then the assumption is that your light source was very blueish. To compensate for that, the picture will be displayed more reddish.

The goal of all of this is to make colors look the same regardless of the light source. @Joanna demonstrated that in an excellent way.

After all of this it is important to understand the the Kelvin value of the White Balance is not as exact science at it may seem: Nearly no light source has a spectrum close to a black body, especially not artificial light sources. Different materials react differently to the weird spectra of artificial light sources, because their reflection is not contiguous either. And then there are our brains constantly tricking us. We are just incapable to have an objective view on colour. So unless you can work in a completely controlled environment, from lighting to output, don’t worry too much.

If you want to learn more I can highly recommend the lectures from Marc Levoy, see my previous post.

I assume you mean the color temperature with wb number.
The color temperature is a tool to eliminate “dirty source light”, light that’s not pure white. You can use predefined values for that or let the camera/converter calculate one. The converter uses that value to correct the 3 color channels.
If you shoot a white card under different conditions, that white card should be white after conversion, despite they might have total different color temperatures. An eventual correction is done with the gray pipet in the “raw white balance”, which is called color picker in pl for some reason.
All that conversion is done in a certain color space. You can only select which color space, if that’s available.

George

Well, i think you misunderstood my writing,

i try to grasp why one say it’s important to set WB of camera fixed the same as monitor in my case 6500k non calibrated comercial cheapash monitor.
And an other say, just do natural daylight, dxo has 5400 some say 5600 because it’s no use.
The camera records it’s hole wavelengthspectrum that’s entring the lens so for that it doesn’t matter , the hole WB thing in a camera is for us to replicate what we see.
(i can imaging that wb in camera has some influence on the exposuretime.)
So i think if dxo chooses 5400 as first entry, default why not me too?

About the colorspaces of printing or viewing on a screen. As far as i know reflection and surface of the print influence the color more then the white point i choose in my camera of in dxo for that matter.
Screens has microsoft colorspace or there manufactorer’s one.
I assume dxo output try’s to convert there virtual jpeg i made in development to my screen accoording to the differences of what’s assumingbly white.
The point is why does it matter to not use AWB as all the rest is floating around in commercial cheap devices?

A camera has wb as estimated gues what kind of color temp the light is so if you like to get a out of camera jpeg procused i wil use that to replicate what you did see at that moment. So that would be a good startingpoint right?
Unless as @Johanna wrote you want that orange sundown light in the image and not be corrected to neutral. Then you set the camera to say 5600k or so. So it produces a more orange oocjpeg. Al this is a suggestion to the rawdeveloper which uses that estimated gues of the camera as startingpoint to give you a view of what the camera was thinking what was the image color temperature.
You can alter this to your liking. (so again, for the sake of argument, why should i bother not using awb most of the time?)

On screen and on paper or a wall or anything else, if you pick a wb point and print on different paper tints it look different. If you wife screw in a different lightbulb because the old one died, in the spot your image looks different on the wall.
If my tv is starting to age, my image’s start looking different. Same with my monitor.
So unless i really take care on calibrating my editing device on regular basis i am drifting in the wind and sea streams without any compas.

So what i try to get clear if you don’t have a calibrated screen what’s the best camera setting to help in getting a moderest steady output of jpegs out dxo in wb point of view.

I hope i get more clear about this now.

Exactly. That false light will be corrected to some neutral, I don’t know how to call it.That raw data is given a color temperature which is used to correct the colors. If awb gives white, then all you can say it has done a good job.

What I want to say in this thread, is that the color temperature is not referring to the image we see on the screen or anywhere else.

Did you solve the problem between the relation color temperature and cold/warm?

George