White Balance or Colour Balance
What is white balance or colour balance and how does it become out of balance? How do we know when it is correct?
The other day I was making a close-up of a nice apple that happened to have a yellow shadow impression of a leaf. My lighting was partially diffused daylight coming through a couple of windows. I looked at my subject and then at the screen and there was a very big difference in colour.
Above: This is how it looked to my eyes.
Above: This is how it looked on my screen. I thought, blah, that’s terrible.
What went wrong?
Auto White Balance (AWB) is to blame.
Yes, AWB wrecked the white balance of my picture and I had to circumvent it to get the proper colour. In this case I used a white balance reference card to read from. I made the reading, locked it, and made the picture. Perfect colour, no guessing, no editing, no fiddling with sliders.
I noticed something interesting while I was using my white balance card. The card view was filling my screen and when I took it away I saw the apple in its glorious natural colour, but just for a second. Then I saw the colour change as the AWB took over and the colour was leached out of my picture. The transformation took 2 or 3 seconds to go from right to very wrong. This was the adaptation at work. I had not realize it took a little time for the camera to study the image before settling on the colour it thought was right.
I returned the card to the apple position and this time I pressed the WB lock button. Now my screen image matched the actual subject.
RGB - the Additive Primaries.
Colour photos are made of 3 colour layers, Red, Green & Blue. This is where “RGB” comes from that describes the colour space our pictures live in. Why RGB and not some other colours? Why not the Red, Blue & Yellow that painters use? To understand that we need to look at the photographic colour wheel.
Imagine you have 3 projectors, each with a different colour filter, and you project the 3 light beams so they overlap somewhat. If you have the right three colours, they “add” up to white. The only time it adds up to white is when the colours are the additive primaries, Red, Green, & Blue. You can make any colour by adding proportions of these 3 colours.
I made the illustrations in Affinity Designer. I started with a black background, added 3 layers, each with a coloured circle. Yes, they really do add up to white. Likewise, the cyan, yellow and magenta are the natural result of this addition. You actually use “Add” in the layer mode selection.
Remember how we used to use a projector to show colour slides? The light passed through the RGB layers of the slide film and added them together on the screen. It isn’t that different from the way we look at pictures on a computer monitor or tablet or cell phone screen. In this case we are looking directly at the light and seeing the light coming through the 3 layers of our image. Sort of like looking at large slides on a light box.
Note that where red and green overlap we get yellow. I can feel the painters cringing now. Painters are mixing solid coloured pigments while photographers are mixing transparent coloured light. A painter’s primaries are those colours that cannot be made by mixing other colours together. Red, Blue & Yellow. When you add red and green light the result will be lighter than either of the two colours. Pigments, on the other hand, absorb light so when you mix two paint colours you always get something darker. When working with colour photos it really helps if you can keep the photographer’s colour wheel in mind and not the painter’s colour wheel.
When you add Blue and Green you get Cyan. Blue + Red = Magenta. Red + Green = Yellow.
Subtractive Primaries CMY
Now imagine you are looking towards a white light box and you have 3 coloured filters, cyan, magenta, and yellow. Each coloured filter passes its own colour and blocks the others. When the exact colours are overlapped you block all the light and you get black. Cyan, Magenta, & Yellow are the subtractive primaries and their best known application is in colour printing. The precise names are Process Cyan, Process Magenta, Process Yellow.
Above: You see that each pair of colours that overlap results in a darker colour.
Above: Now we have the 6 colours in the Photographic colour wheel. This is the same colour wheel used by physicists and scientists in the study of light, colour and the electromagnetic spectrum.
When we are thinking about colour balance or white balance it is easier to imagine this type of colour wheel with the colours faded towards white in the middle. In a correctly colour balanced photo you can imagine a dot in the very centre of the wheel representing neutrality.
When your photo has a colour shift the dot representing that shift on the wheel could be in any position, all around the 360º of the colours and farther from or closer to the centre representing the degree of imbalance. So we have the direction of the imbalance, plus the amount of imbalance represented by one dot.
You can think of this type of colour control as having a joystick in the centre. Let’s say our colour shift is towards red you are going to pull the joystick away from red and towards cyan until you get back to the centre.
Once you understand how Auto White Balance goes wrong you will have a good idea of which way it has gone and why, and how to correct it.
Auto White Balance looks at the whole picture and tries to strike a balance in the colours in the image. When it perceives an imbalance it pulls the colour away from the excess colour by adding the opposite colour.
In the case of my apple picture at the top. The Auto White Balance looks at that and sees mostly red, but also a big patch of yellow, and it says “where’s the cyan/blue?” so it piles on a bunch of cyan until it feels the scale come back into balance. From the AWB perspective you can imagine a dot placed on the colour wheel above probably 2/3 in the red and 1/3 towards the yellow. The opposite of that would be 2/3 cyan and 1/3 blue. The problem is in this case, the picture really is made up mostly of red (plus some yellow) so adding all that cyan/blue ruins the colour. In this case the AWB has caused the problem. It would have been better if the AWB just sat this one out. As soon as you look at a subject with so much red you can anticipate the response of the AWB will be to pile on the cyan.
From the AWB’s point of view it would call this “subject failure”. From our perspective we call it AWB failure. Essentially, we have presented it with a no-win situation.
More to come. This is my opening message. The how and why will follow.
What is white balance or colour balance and how does it become out of balance? How do we know when it is correct?
The other day I was making a close-up of a nice apple that happened to have a yellow shadow impression of a leaf. My lighting was partially diffused daylight coming through a couple of windows. I looked at my subject and then at the screen and there was a very big difference in colour.
Above: This is how it looked to my eyes.
Above: This is how it looked on my screen. I thought, blah, that’s terrible.
What went wrong?
Auto White Balance (AWB) is to blame.
Yes, AWB wrecked the white balance of my picture and I had to circumvent it to get the proper colour. In this case I used a white balance reference card to read from. I made the reading, locked it, and made the picture. Perfect colour, no guessing, no editing, no fiddling with sliders.
I noticed something interesting while I was using my white balance card. The card view was filling my screen and when I took it away I saw the apple in its glorious natural colour, but just for a second. Then I saw the colour change as the AWB took over and the colour was leached out of my picture. The transformation took 2 or 3 seconds to go from right to very wrong. This was the adaptation at work. I had not realize it took a little time for the camera to study the image before settling on the colour it thought was right.
I returned the card to the apple position and this time I pressed the WB lock button. Now my screen image matched the actual subject.
RGB - the Additive Primaries.
Colour photos are made of 3 colour layers, Red, Green & Blue. This is where “RGB” comes from that describes the colour space our pictures live in. Why RGB and not some other colours? Why not the Red, Blue & Yellow that painters use? To understand that we need to look at the photographic colour wheel.
Imagine you have 3 projectors, each with a different colour filter, and you project the 3 light beams so they overlap somewhat. If you have the right three colours, they “add” up to white. The only time it adds up to white is when the colours are the additive primaries, Red, Green, & Blue. You can make any colour by adding proportions of these 3 colours.
I made the illustrations in Affinity Designer. I started with a black background, added 3 layers, each with a coloured circle. Yes, they really do add up to white. Likewise, the cyan, yellow and magenta are the natural result of this addition. You actually use “Add” in the layer mode selection.
Remember how we used to use a projector to show colour slides? The light passed through the RGB layers of the slide film and added them together on the screen. It isn’t that different from the way we look at pictures on a computer monitor or tablet or cell phone screen. In this case we are looking directly at the light and seeing the light coming through the 3 layers of our image. Sort of like looking at large slides on a light box.
Note that where red and green overlap we get yellow. I can feel the painters cringing now. Painters are mixing solid coloured pigments while photographers are mixing transparent coloured light. A painter’s primaries are those colours that cannot be made by mixing other colours together. Red, Blue & Yellow. When you add red and green light the result will be lighter than either of the two colours. Pigments, on the other hand, absorb light so when you mix two paint colours you always get something darker. When working with colour photos it really helps if you can keep the photographer’s colour wheel in mind and not the painter’s colour wheel.
When you add Blue and Green you get Cyan. Blue + Red = Magenta. Red + Green = Yellow.
Subtractive Primaries CMY
Now imagine you are looking towards a white light box and you have 3 coloured filters, cyan, magenta, and yellow. Each coloured filter passes its own colour and blocks the others. When the exact colours are overlapped you block all the light and you get black. Cyan, Magenta, & Yellow are the subtractive primaries and their best known application is in colour printing. The precise names are Process Cyan, Process Magenta, Process Yellow.
Above: You see that each pair of colours that overlap results in a darker colour.
Above: Now we have the 6 colours in the Photographic colour wheel. This is the same colour wheel used by physicists and scientists in the study of light, colour and the electromagnetic spectrum.
When we are thinking about colour balance or white balance it is easier to imagine this type of colour wheel with the colours faded towards white in the middle. In a correctly colour balanced photo you can imagine a dot in the very centre of the wheel representing neutrality.
When your photo has a colour shift the dot representing that shift on the wheel could be in any position, all around the 360º of the colours and farther from or closer to the centre representing the degree of imbalance. So we have the direction of the imbalance, plus the amount of imbalance represented by one dot.
You can think of this type of colour control as having a joystick in the centre. Let’s say our colour shift is towards red you are going to pull the joystick away from red and towards cyan until you get back to the centre.
Once you understand how Auto White Balance goes wrong you will have a good idea of which way it has gone and why, and how to correct it.
Auto White Balance looks at the whole picture and tries to strike a balance in the colours in the image. When it perceives an imbalance it pulls the colour away from the excess colour by adding the opposite colour.
In the case of my apple picture at the top. The Auto White Balance looks at that and sees mostly red, but also a big patch of yellow, and it says “where’s the cyan/blue?” so it piles on a bunch of cyan until it feels the scale come back into balance. From the AWB perspective you can imagine a dot placed on the colour wheel above probably 2/3 in the red and 1/3 towards the yellow. The opposite of that would be 2/3 cyan and 1/3 blue. The problem is in this case, the picture really is made up mostly of red (plus some yellow) so adding all that cyan/blue ruins the colour. In this case the AWB has caused the problem. It would have been better if the AWB just sat this one out. As soon as you look at a subject with so much red you can anticipate the response of the AWB will be to pile on the cyan.
From the AWB’s point of view it would call this “subject failure”. From our perspective we call it AWB failure. Essentially, we have presented it with a no-win situation.
More to come. This is my opening message. The how and why will follow.
Last edited: