Edges
December 31, 2019
December 28, 2019
December 25, 2019
December 22, 2019
December 19, 2019
December 16, 2019
December 13, 2019
Finding Edges
Finding Edges
In my well-used copy of Principles of Digital Image Processing: Fundamental Techniques, there's a chapter on Edges and Contours. I had looked through it several times, and had mostly understood the concepts, but I really had no use for the techniques it described.
Then one day I was reading a Lensrentals.com blog post titled "Stopping Down Some Bargain Primes and Zooms" that contained a short section with the heading "So How Do You Check Your Field Curvature". In it Roger Cicala describes how to use Photoshop's Find Edges filter to show the field of focus in a photograph. (And also the possible curvature or tilt of said field of focus.) Well, that got me going.
Yet Another Computer Program
So, of course, we now have a new computer app that takes a photo and produces an image of its edges. We could have just loaded up Photoshop and used it, but where's the fun in that? Here's kind of how things progressed. The new app implements two (apparently) old reliable edge detection filters: the Prewitt and the Sobel edge operators. Let's take a picture and run the app on it using the Sobel operator. Here's the picture:
Just looking at this picture, we see some edges in the tree bark, but no real edges in the out-of-focus background. Here's what the Sobel edge detection operator produces:
Not so interesting. The places in the picture where there are no edges are black; any edges are lighter in value. This image has no really prominent edges, so there are only the faintest lighter values. It would probably be prudent to scale the edge values so that the sharpest edges in the picture (whether they are actually very sharp or not) are pure white. When we do that, we get:
This is somewhat better. However, I prefer seeing the edges as black lines against a light background instead of light lines against a black background. So let's do that:
That's even better, but still a bit washed out. So we'll scale the edge values an arbitrary amount to get an image that suits us.
And finally, because we like this image just for itself, we'll adjust the color and contrast a bit to give it a final look.
Yes, it's likely you'll be seeing one or two of these "edge" images On the Cork Board.
Back to Checking Your Field of Focus.
OK, so what got us into this mess to start with was the desire to use an edge filter to show us a lens' field of focus. Here's a practical application. I was just out taking pictures in an overgrown pasture with a 70-200mm zoom at 70mm. Here's the shot I was trying to get:
A very real issue in this situation is "Where do I focus?" Knowing I had the find edges app, I took four exposures. In the first, I focused in the center of the picture. In each subsequent exposure, I moved the focus point closer to the bottom of the picture. When I got home, I ran all four exposures through the find edges app. Here's what I got.
You can see the field of focus getting closer to the camera. It shows up as darker than the out-of-focus areas. (Note also that my lens has a curved field of focus, which is not much of a surprise.) Based on the look I was after, I chose the upper right image as the one with the best field of focus.
Some of you may recognize that this "find edges" technique is what your camera is doing when you use "focus peaking". More fun with computation.
December 11, 2019
December 8, 2019
December 5, 2019
It looks like moving water blurred with a long exposure through a neutral density filter; but it's not. It's actually 25 normal exposures, shot at one second intervals and then averaged together. It's from one of my recent computer programs that lets you combine multiple exposures in various ways. Cedar Falls Park, Fountain Inn, SC
December 2, 2019
November 29, 2019
November 26, 2019
November 23, 2019
November 20, 2019
November 17, 2019
November 14, 2019
November 11, 2019
November 8, 2019
November 5, 2019
November 2, 2019
October 30, 2019
October 27, 2019
October 24, 2019
October 21, 2019
October 15, 2019
October 12, 2019
October 9, 2019
October 6, 2019
October 3, 2019
September 30, 2019
September 27, 2019
September 24, 2019
September 21, 2019
September 18, 2019
September 15, 2019
September 12, 2019
September 9, 2019
September 6, 2019
Subscribe to:
Posts (Atom)