Milky Way vs. City Lights – My first attempt at computational astrophotography
From Mount Wilson on a moonless night, the Milky Way is faintly visible over Los Angeles. I wanted to see if I could extract enough information from multiple exposures to overcome the city lights and make the Milky Way become conspicuous.
This picture is the result of a fair amount of mathematical modeling and processing. All of the shots used to produce this image are taken from the same spot without moving the camera between exposures. I’ve included some further explanation and intermediate images in the first comment below.
[Nikon D90, Tokina 11-16mm @11mm, f/2.8, ISO 800, 15 minutes worth of exposures]
[Additional note -- May 19, 2013]
I've really appreciated all of the interest and nice remarks. And thanks especially to Phil Plait for the great article on his "Bad Astronomy" blog over at Slate. I have been an occasional reader of Phil's blog since before it moved to Slate, so it was a pleasant surprise when he emailed me with his questions.
I wanted to respond to the questions and interest I've received about sharing the software that I used to produce this image. (And I apologize for taking so long to reply.) Unfortunately, to say that the software is user-unfriendly would be an understatement. I wrote most of it in Mathematica, which saved me the trouble of having to do things like write routines to solve differential equations or perform a least-squares fit. But it means that you would need the Mathematica software and some experience in reading and writing code within it. Furthermore, I made little effort to do things like document which chunks of code worked well and which ones should be ignored because they include some mistake; it's not a small job for me to try to clean up the code that I wrote two years ago. Even for someone proficient with Mathematica, it would take a ton of work to try to do something useful with my code. Sorry.
(I've noticed some trends in my projects lately. They tend to be complicated and even more time consuming than I anticipate. They tend to provide a reasonable mixture of fun and frustration while I'm working on them, and sometimes a satisfying result. But if someone asked me to do it again, my first instinct is to run away screaming.)
But wait! I've been alerted to the existence of the free RegiStax software. I have no experience with it, but it looks pretty impressive. And considerably more user-friendly than what I've done. I hope some of you try it out and send me a note to let me know how it goes. I'd like to see your results!