This month's subject was "Fences". While fences aren't particularly interesting it gave us an excuse pre-processing and post-processing skills. Our pre-processing skills of framing, exposure, focus, and depth of field selection were exercised. Our post processing skiils using the myriad adjustments available to recreate the image we saw when we took the picture.
This was a productive meeting. We reviewed most of the pictures before we ran out o1100pxf time. The most interesting topic we explored was the subject of curves. That little box we have all seen and wondered what it was all about.
It all started with this picture. I looked at Lake Michican and said: "Is the lake smooth?" thinking of wind swept snow over a smooth sheet of ice. I was thinking speeding ice boats! Would those ever be fun pictures to take!
She said no. The wind had blown the ice up into big pressure ridges softened somewhat by snow. But the lake was very, very rough. So much for ice boat pictures.
That opened the curves discussion. Everyone knows about the "Shadows and Highlights" and the contrast adjustments. But sometimes these relatively course adjustments cannot bring out the detail we would like. Curves adjustments are the answer.
The curves shown to the right are for linear contrast as shown in the picture above and a "medium" contrast curve. The underlying histogram can be easily seen. The bump on the left are the trees and the distant mountain. The bump on the right is all the white snow. The "medium" contrast steepens the curve in the center of the histogram where there is nothing of interest.
The challenge is to steepen the curve on the far right to add contrast in the region of interest. We will compromise by flattening the curve in the center and add a little steepness on the left to retain interest in the trees and mountain.
That brings us to the following point curve:
What we accomplished here is radically steepened the curve in the area of the white snow, flattened the area of little information and no interest, and added a little steepness showing more detail in the pine trees.
That creates the next picture below.
This brings us to another challenge we may have when curves are radically steepened. We can see the sensor dust when present. This was a remarkably dirty sensor. Even more radical curves will reveal all dust, particularly if the underlying image is blurred sky or wall.
The question arises: when do we need to use curves? Generally we may want to add a little low to medium contrast to give the image some pop. Radical curves like this are useful to bring out the contrast in low contrast areas. Areas like bringing out detail in the snow fields to the right, the feathers in a great egret, and the sky in a sunset or sunrise come to mind.
Below the same image was reworked in Photoshop. First the trees and mountain were masked out. An masked exposure layer was added to brighten the snow then a masked curve layer was added to bring out the relief.
A complementary curve layer was added to bring out the detail in the mountains and trees. The details can be seen in the last image.
The example below shows a boardwalk on Lake Michigan.
The detail in the snow banks are all but lost. The sky looks drab. While this is an accurate representation of the scene it is not what the photographer saw. The snowbank was much deeper and the sky had more color.
The second picture is closer to what the photographer saw. Unfortunately, now there clearly is banding in the sky and the snow.
The problem is, by necessity, we are using an 8 bit jpeg file. Jpeg files have only 256 levels compared to the 65536 levels in a 16 bit tif file.
So the sky and the snow may have only 5 gradiations of color in the left pictures.
On the left, the blue channel goes from 241 to 247 for 6 levels. The right goes from 194 to 218 for 24 levels. But the levels are divided into only 6 colors or bands. The same colors at 16 bits are 30,918 to 31,714 for 996 levels. On the right levels go from 24,911 to 27,964 for 3,053 levels. It is a digital process so there is still banding. But the 996 levels spread across the entire sky is virtually impossible to detect. Six levels is easy to detect.
That is why we should always use raw images and do all of our post processing in 16 bits. The 8 bit jpeg files are for display only.