 #...
·
1
like
|
|---|
 https://astrob.in/v95b8a/0/OTA: AP 175 Camera: FLI PL16070 Observatory: Deep Sky West Exposures: OIII 16 x1800” Ha 20 x 1800” SII 14 x1800” Total Exposure: 25 hrs Image width ~1 degree Processed by Alex Woronow (2019-20) North American Nebula in the region of the “Pelican Nebula” I hesitate to label this strictly an “artistic rendition” because what you see is really there, just not usually so obvious. Among my many experiments on how to bringmore detail to my images, I experimented with various implementations of a thing called a Sobel Operator. Although most implementations of that operator have been tuned to enhance edges, the operator actually delineates slopes or gradients in brightness and color. My coding steps back and simply stops at recognizing slopes In effect, this image portrays the first derivative of the parent image. You should think about viewing a mountain at the low illumination angle of sunset or sunrize--a prime time for photography or painting! This image is similar to that, except in this image the sun sets to the North, East, West, and South, all in the same “painting.” That is, gradients in all directions become pronounced; consequently, the structure and chaos of the nebula comes to the fore. Comments always welcomed! Alex P.S. I see that this group is becoming just another place to post every image ... Is there some way to discourage that? I posted as a "new topic" to avoid being lost in the array of irrelevant postings. But that's not ideal.  |
9.53
#...
·
1
like
|
|---|
| Excellent image Alex! It does look like a painting! |
12.30
#...
·
1
like
|
|---|
|
Beautiful image! Your description sent me down a rabbit hole looking up what a Sobel operator is and how it works, which led me to writing some code to try it out. It's very cool! But what I read was all about edge detection. If I may ask, how are you using it to show the slope? Thanks for sharing this! Tim |
#...
·
1
like
|
|---|
|
I implemented it in PI, which has the operator available in a somewhat different form than most out-of-the box implementations. The operator is just a running filter/convolution filter that approximates the local first derivative of the pixel values. In PI it is under the Convolution tool. To be honest, I never looked into how SO is manipulated to give edges. I run it in 4 directions for each of RGB separately. So that's 12 images to combine. Be happy to send the PI icons if you want to send me your email address (send it to Alex-at-Faintlightphotography-dot-com). (If anyone else wants the same, just write.) I largely abandoned the Sobel thing in favor of using a MatLab code for Laplacian pyramid sharpening provided by Paris, Hasinoff, and Kautz, "Local Laplacian Filters: % Edge-aware Image Processing with a Laplacian Pyramid", ACM % Transactions on Graphics (Proc. SIGGRAPH 2011), 30(4), 2011. It works well, but time-consuming to optimize the parameters for a given picture. Then abandoned that for Artificial Intelligence (commercial codes) for sharpenning, although I regress now and then. Thanks for asking....cheers, Alex |
