The use of curves is a natural extension from our MW2 about White Balance & Colour balance. Curves are often used when making corrections to tonality and colour in images.
Before we get into using curves for image correction it makes sense to introduce curves to make sure everyone understands what they represent and what they can do.
I realize this all looks like a bunch of graphs but before we start looking at images we need to explain what the curves are all about.
Introduction to Curves:
#1 Let’s start here. This is the standard format for curves to be displayed. The two gradients bars are not usually included to save space but their presence is implied and having them there to start with makes it easier to understand what the placement of points means on the graph. You can think of the lower right corner as the starting point.
Across the bottom we have tones becoming darker as we move to the left. Up the side the tones get lighter as we go up. If we wanted to put a point on the graph to represent white it would be in the upper right. And black at the lower left.
The background grid just helps us know the relative position of our curve as we move it.
#2 This “curve” is a straight line representing a linear relationship in the placement of tones. When you see this straight line it basically means “no change”. In terms of a response curve this indicates that what you have coming in is exactly the same going out.
Digital cameras have what is called Linear Gamma. Gamma is a description of the slope of the graph. It is not only images and cameras that use this curve but your monitor or screen also has a curve which tells the computer how to place tones on the screen to get a “normal” looking image.
We won’t be doing anything with Monitor gamma but it is probably worthwhile to mention it so you have a greater understanding of what is happening behind the scenes. Monitor Gamma can be quite different from device to device but it is very important to make our images look correct. Monitor calibration involves reading known test values off the screen and creating a new Monitor Gamma curve for your monitor so that the placement of tones and colours is exactly right.
Monitor calibration is necessary to compensate for variations in manufacturing, as well as decline in performance over time. In addition, a general purpose office monitor has no need to be precisely accurate, and they aren’t. It costs more money to make an accurate monitor. A graphics monitor is in the higher price range and what you get is much better colour and tonal accuracy. It is generally not possible to calibrate an office monitor up to graphics standards.
#3 This curve represents an increase in contrast in the image. Making the graph steeper increases the image contrast. Starting from the black point on the lower left, the assignment of “Black” has been dragged to the right somewhat towards grey. This means that all tones to the left of our new position have been rendered as black. Imagine that your image was made on a foggy day and doesn’t contain any blacks or pure whites. It is a low contrast image. If you wanted to increase the contrast a bit you would do this by pulling the black end inwards.
Looking at the white part of the graph at the top right: If we go back to the foggy day image, it has only weak light grey values with no strong white. Moving the top point of the graph left we are saying we want the light grey values to be brighter. Now anything to the right on the new white position will be pure white. These two adjustments might be good for a low contrast rainy day image but not for a sunny day image. A sunny day image already has a full range of tones from white to black to pulling the white end of the graph inwards will have the effect of bumping your highlight detain into pure white - not desireable at all. Likewise with your black values: pulling the black point inwards means your shadow detail will be rendered as solid black.
When you use the simple contrast slider in many editing apps what it is doing is changing the contrast slope of your “curve”, darkening blacks and lightening whites. However, without seeing some sort of histogram you might not realize that increasing the image contrast is fine for the mid tones but destructive if your image contains delicate highlight values and shadow tones. How can we get around this?
#4 Changing the curve to look like this lowers contrast. In the upper right, where the white are, the bright whites have been darkened to look light grey. Over at the black end, the blacks have been lightened to be only dark grey. In most images this would look very dull but it is important to understand how the graph works.
When you adjust a simple contrast slider towards lower contrast this is what is happening. Probably not what you wanted. Again, how can we get around this?
#5 Linear gamma (farther above) describes the output of digital cameras. This type of curve represents the typical tonal distribution of film. With digital is is entirely possible to have an linear relationship between the tones. With film we are talking about the threshold in sensitivity of the film not seeing any light to gradually detecting some light at the shadow end and gradually reaching the limit of what can be recorded at the highlight area without all turning white. Aha, you say. This is what makes film look different from digital.
If you think about the effect of this curve on the image it means the highlights (the shoulder area of the graph) are somewhat compressed, and becoming increasingly hard to differentiate. At the “toe” region, we have a stretched out area of shadow separation. The more or less straight midtone area has higher contrast.
Every film has it’s own distinctive response curve. Some were nowhere near linear in the mid tones. Some had a soft shoulder and others more sharp. Some films had a “long toe” while others were closer to linear. Here we are only looking at the overall tonality, as if it were a B&W film, but don’t forget that each RGB layer in a colour film also has it’s own response curve and they don’t necessarily lay on atop the other with perfect accuracy. This is where we get the differences in colour response between different types of films. Manufacturers struggled hard to get closer to linear gamma but is was impossible. So now we have linear gamma in digital cameras and for some reason some people want to go back to the nonlinear response curves.
So here it is: If you want to make your digital images look like film it is all in the curves. Maybe this will be some incentive for some people to pay attention to this part.
So I don’t leave you hanging until I post the next part,
#6 This curve represents brightening the mid tones without changing the whites or blacks in the image. This doesn’t mean there have not been slight changes in the upper and lower regions but we will look at the mid tone values first. You can see the curve has been pushed out about the width of one square in the background grid. The middle tones will be noticeably brighter. The upper light tones are also lightened so some highlight details may be at risk. (Pay attention).
Looking at the upper half of the curve you can see that the contrast slope is lowered for the upper tones. The upper half of the curve looks like the low contrast curve from diagram #4 above.
The contrast slope is increased in the lower tones. Like diagram #3 above. So it has become easier to differentiate shadow tones but harder in the bright tones. In the shadow area, even though black remains where it was the lower mid tone values have been lightened. There is a bigger increase in tone brightness as we leave black.
#7 This curve represents darkening the middle image values without changing the blacks or whites. Very useful. Also notice the contrast slope in the lower tones is decreased and the upper mid tone slope is increased. So we have greater separation in the upper middle tones but a bit less in the lower part.
Let’s suppose you wanted to increase the mid tone contrast without moving your whites and blacks. Look at diagram #5.
Continued next message. That may be tonight or tomorrow due to other commitments.
Before we get into using curves for image correction it makes sense to introduce curves to make sure everyone understands what they represent and what they can do.
I realize this all looks like a bunch of graphs but before we start looking at images we need to explain what the curves are all about.
Introduction to Curves:
#1 Let’s start here. This is the standard format for curves to be displayed. The two gradients bars are not usually included to save space but their presence is implied and having them there to start with makes it easier to understand what the placement of points means on the graph. You can think of the lower right corner as the starting point.
Across the bottom we have tones becoming darker as we move to the left. Up the side the tones get lighter as we go up. If we wanted to put a point on the graph to represent white it would be in the upper right. And black at the lower left.
The background grid just helps us know the relative position of our curve as we move it.
#2 This “curve” is a straight line representing a linear relationship in the placement of tones. When you see this straight line it basically means “no change”. In terms of a response curve this indicates that what you have coming in is exactly the same going out.
Digital cameras have what is called Linear Gamma. Gamma is a description of the slope of the graph. It is not only images and cameras that use this curve but your monitor or screen also has a curve which tells the computer how to place tones on the screen to get a “normal” looking image.
We won’t be doing anything with Monitor gamma but it is probably worthwhile to mention it so you have a greater understanding of what is happening behind the scenes. Monitor Gamma can be quite different from device to device but it is very important to make our images look correct. Monitor calibration involves reading known test values off the screen and creating a new Monitor Gamma curve for your monitor so that the placement of tones and colours is exactly right.
Monitor calibration is necessary to compensate for variations in manufacturing, as well as decline in performance over time. In addition, a general purpose office monitor has no need to be precisely accurate, and they aren’t. It costs more money to make an accurate monitor. A graphics monitor is in the higher price range and what you get is much better colour and tonal accuracy. It is generally not possible to calibrate an office monitor up to graphics standards.
#3 This curve represents an increase in contrast in the image. Making the graph steeper increases the image contrast. Starting from the black point on the lower left, the assignment of “Black” has been dragged to the right somewhat towards grey. This means that all tones to the left of our new position have been rendered as black. Imagine that your image was made on a foggy day and doesn’t contain any blacks or pure whites. It is a low contrast image. If you wanted to increase the contrast a bit you would do this by pulling the black end inwards.
Looking at the white part of the graph at the top right: If we go back to the foggy day image, it has only weak light grey values with no strong white. Moving the top point of the graph left we are saying we want the light grey values to be brighter. Now anything to the right on the new white position will be pure white. These two adjustments might be good for a low contrast rainy day image but not for a sunny day image. A sunny day image already has a full range of tones from white to black to pulling the white end of the graph inwards will have the effect of bumping your highlight detain into pure white - not desireable at all. Likewise with your black values: pulling the black point inwards means your shadow detail will be rendered as solid black.
When you use the simple contrast slider in many editing apps what it is doing is changing the contrast slope of your “curve”, darkening blacks and lightening whites. However, without seeing some sort of histogram you might not realize that increasing the image contrast is fine for the mid tones but destructive if your image contains delicate highlight values and shadow tones. How can we get around this?
#4 Changing the curve to look like this lowers contrast. In the upper right, where the white are, the bright whites have been darkened to look light grey. Over at the black end, the blacks have been lightened to be only dark grey. In most images this would look very dull but it is important to understand how the graph works.
When you adjust a simple contrast slider towards lower contrast this is what is happening. Probably not what you wanted. Again, how can we get around this?
#5 Linear gamma (farther above) describes the output of digital cameras. This type of curve represents the typical tonal distribution of film. With digital is is entirely possible to have an linear relationship between the tones. With film we are talking about the threshold in sensitivity of the film not seeing any light to gradually detecting some light at the shadow end and gradually reaching the limit of what can be recorded at the highlight area without all turning white. Aha, you say. This is what makes film look different from digital.
If you think about the effect of this curve on the image it means the highlights (the shoulder area of the graph) are somewhat compressed, and becoming increasingly hard to differentiate. At the “toe” region, we have a stretched out area of shadow separation. The more or less straight midtone area has higher contrast.
Every film has it’s own distinctive response curve. Some were nowhere near linear in the mid tones. Some had a soft shoulder and others more sharp. Some films had a “long toe” while others were closer to linear. Here we are only looking at the overall tonality, as if it were a B&W film, but don’t forget that each RGB layer in a colour film also has it’s own response curve and they don’t necessarily lay on atop the other with perfect accuracy. This is where we get the differences in colour response between different types of films. Manufacturers struggled hard to get closer to linear gamma but is was impossible. So now we have linear gamma in digital cameras and for some reason some people want to go back to the nonlinear response curves.
So here it is: If you want to make your digital images look like film it is all in the curves. Maybe this will be some incentive for some people to pay attention to this part.
So I don’t leave you hanging until I post the next part,
#6 This curve represents brightening the mid tones without changing the whites or blacks in the image. This doesn’t mean there have not been slight changes in the upper and lower regions but we will look at the mid tone values first. You can see the curve has been pushed out about the width of one square in the background grid. The middle tones will be noticeably brighter. The upper light tones are also lightened so some highlight details may be at risk. (Pay attention).
Looking at the upper half of the curve you can see that the contrast slope is lowered for the upper tones. The upper half of the curve looks like the low contrast curve from diagram #4 above.
The contrast slope is increased in the lower tones. Like diagram #3 above. So it has become easier to differentiate shadow tones but harder in the bright tones. In the shadow area, even though black remains where it was the lower mid tone values have been lightened. There is a bigger increase in tone brightness as we leave black.
#7 This curve represents darkening the middle image values without changing the blacks or whites. Very useful. Also notice the contrast slope in the lower tones is decreased and the upper mid tone slope is increased. So we have greater separation in the upper middle tones but a bit less in the lower part.
Let’s suppose you wanted to increase the mid tone contrast without moving your whites and blacks. Look at diagram #5.
Continued next message. That may be tonight or tomorrow due to other commitments.
Last edited: