Hello Everyone.
This past weekend, I came up with an idea for an experiment with time.  I was sitting in my apartment, watching storms pass by, thinking to myself “it would be really neat to make a timelapse of these passing storms”.  But the storms came and went over two days.  So I began to think about how you could actually capture something that lasts that long.
After thinking about it for a day, I came up with an idea – a full 24 hour timelapse.  Somehow, I would have a camera take photos for 24 hours.
My first thought was to have 24 hours in 24 seconds.  Since I live in the USA and we use 30 frames per second on our TVs, shrinking 60 minutes to 30 frames means 1 frame every 2 minutes.  After thinking about this longer, I decided this was a lot of work for under 30 seconds of video, so I decided to go with 40 frames per hour, or 1 every 90 seconds.  This works out to 960 photos in 24 hours.  I ended up going with 1000 photos over 25 hours as a ‘just in case’ precaution.
After more thoughts, I decided to go from midnight to midnight.
So, with a concept created, I got ready.  I cleared out my memory cards, set up my tripod in my windows, then set up the camera.  I decided to use my D750 because my D800 is testing out a new tripod, and my 24-70 f/2.8 lens.  My initial thoughts were that the D750 would be better because the low light capabilities are better (turned out this didn’t matter) and because the files are smaller (this DID make my life easier).
For the camera setup, I had to shoot through a window.  So, I added my C-PL to the front of the lens to eliminate as much of a reflection as possible, then used masking tape to hold a dark blue hand towel to the window to prevent as much reflections as possible.  The towel was taped to the window, camera, and tripod.  Because I would be asleep and unable to make changes, I selected f/9 and iso400 in Aperture priority mode.  The D750 has a min shutter speed of 1/4000th of a second and a max shutter speed of 30 seconds.  Back of the napkin math said this would probably be fine.  For the metering point, I picked the only white object in the scene — the Space Needle.  This would also allow the auto white balance to correctly handle the transition from night to day.
And so, on March 7th, 2016, I set up my camera with a timer set to start at midnight, and went to bed.
I woke up at around 7:15 and checked the camera.  Oops, battery is running low, glad I looked.  One more battery change at 6pm was required.  The last battery charge lasted from 6pm to 1am and still had about 40% remaining in the morning.

I am very happy with the results of this test.  Due to a limitation of iMovie, each image ended up lasting 0.13 seconds in the final result, and each hour is around 5.3 seconds.
The first mistake I realized I made was when I first looked at the images.  I always shoot in RAW format.  But now, I needed to convert the images to JPEG.  The conversion took 6 hours.  Next time, I’ll either shoot in RAW+JPEG or JPEG only.  The reason I would use RAW+JPEG is because there is a chance some of the images are good enough to become standalone images.
The next change I would make for the next attempt is to have more images.  I find the resulting video to be quite nice, but, I wish it was longer.  I would probably change to 1 image every 60 seconds and potentially try to go for 2 days.  But, this would require more battery changes, potentially 3 overnight, and probably an external power supply.  There is also less time to change a battery during the night, when the exposures can get to 20 or more seconds.  During the day, when the exposures were 1/400th of a second, there is a lot of extra time for a battery swap.
The final change would be in the framing of the image.  I made the mistake of handling the framing after dark on viewfinder that doesn’t cover 100% of the image.
I would also like to make the resulting video better and more professional.  Right now, this is a rough draft video that was produced quickly to see how well it turned out.  Now that I know how it turned out, I’ll spend some more time on the video, which is something I am not very good at.  Because I started with 24 MPix files, I could finalize the video at 4K, but likely will keep it at HD only.
While I certainly wish I could get lucky and predict when a fantastic sunset will happen (as I look out my window tonight, the sunset is nicer than when I ran the test), this is one of those ‘luck’ things, which is difficult to predict 1-2 days in advance.
-M