Finding hidden details in a photograph

Our senses are very inexact and there are huge variations amongst us. We often cannot see or hear what is there but at the same time we can see and hear things that aren’t there. Cast you mind back a long time when Dolby labs improved our listening pleasure. What they basically did was to suppress sound that was saturating the tapes and wasn’t really necessary, and enhancing other aspects of it to overcome some of the limitations of the media. Then came digital, both sound and vision and that changed a lot of things. There are still people who swear that vinyl sounds better than CDs, and they may be right. The analog world has an infinite number of levels between any two points, be they moments in time, or between two shades of grey. Digital reduces that to a fairly large number of discrete steps. Enough steps that most of the time we don’t notice what is missing in between. You must have seen, on programs like CSI, how they take a blurry picture of a license plate and with the press of a button, it instantly becomes clear and readable. Truth or fiction? Well, on the TV shows, somewhere in between, because to get that much clarity that must be using multiple shots, but you may be surprised at just how much can be done.

Now let’s think about what that means in terms of photographs. We hear a lot about how many megapixels a camera has and this has to do with how many individual points there are in an image. We know that a higher number is better, assuming that the quality of each of those pixels are identical. But they may not be, because each of those pixels has to record how much red, green and blue light is being received, and it also has to convert those into a fixed number of steps. For example, if it were 16-bits per pixel, it means that each color would have about 5 bits which means that for red there would be 64 different degrees of red, the same for green and blue, making a total of about 32,000 colors. Sounds like a lot, but that is not very good. Higher end cameras will have more like 24-bits per pixel making 16 million colors available.

But let’s go back to the number of pixels. Assume for a moment that we have a fairly low number of pixels. I have simulated that by zooming way into a picture. Notice in particular the lines that are not straight up and down or left to right. What we are seeing here is pixilation. Each pixel can only be on the line or off the line – it is not possible for it to be partially on. It has to basically be on it off it, and thus we see that stair step pattern. The only way that it can compensate is that if it were half on and the line was black and the background was white, it will register as being grey. If we were to view this picture at normal resolution you would never notice that as the eye corrects for many small errors. But image processing software can help, and almost as if by magic make it appear that there is more detail here than there actually is.

Here we see the same shot with a rather extreme amount of “sharpening”. This is a control that most image processing software will have and you will notice two things. First is that the edges now appear to be a lot better defined. While still sort of steppy, they are more distinct than they were before. This is because the software has detected an edge and has enhanced the edge. There is no need to go into how it does this, but it is a pretty good method of fooling the eye. However, there is something else that has happened when we did this, and that is the white background has become more noisy. That is because there is some light and shade here that it thinks is an edge and has tried to enhance that as well. Luckily there are probably other controls, such as radius, detail and masking in Lightroom that allow us to control things a little better and to tell it basically that the difference has to be large or significant before it should be enhanced. Very crudely, radius determines the size of “the brush” we are using to make the edges more distinct. The higher the pixel count of your camera, the lower you probably want this number to be. Detail and masking allow differences below a certain threshold to be ignored.

Now, if you apply too much sharpening, the photo can take on a strange appearance, so don’t overdo it and always make sure you are looking at the image in a 1:1 view when adjusting. If you zoom in too far, like I was for these examples you will not be seeing what the eye will see, the same if the image is too compressed. Remember, it is about fooling the eye.

Brought to you by Brian Bailey

[amazon_enhanced asin=”B003739DVY” /] [amazon_enhanced asin=”B003Q81EL2″ /] [amazon_enhanced asin=”0321680707″ /] [amazon_enhanced asin=”B0048EKIQK” /]

One thought on “Finding hidden details in a photograph

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.