Saturday, March 10, 2012

Unequal-Interval Time-Lapse Proof of Concept

This is a proof of concept for a project I am working on. The intervals between the frames that make up the video below were determined using a sine function programmed into my TI-84 calculator/intervalometer. This video was prepared to prove to myself that the process works. It also got me looking into the 'non-linear' time-lapse work that others are doing. More after the break.

Unequal-interval Time-lapse: Clock from Dave C Griffith on Vimeo.

My Canmore Time-Lapse was one of the first I shot using a TI-84 graphing calculator as an intervalometer. Before I even had the calculator in my hands, I had been wondering about how I could do something other than a standard time-lapse. Besides being cheaper, a graphing calculator is more flexible than a run-of-the-mill intervalometer because of the programming feature. This feature is what is used to allow the calculator to act as an intervalometer. A program is written to the effect of: delay for x seconds, send signal, loop. The “send signal” part is (I think) typically used to send data between two linked calculators; however, it can also be used to trigger the shutter release on some cameras. The flexibility of the programming part it what I decided to play with.

In this case, a sine function was used to determine the time between frames based on the time which had elapsed from the start of the time lapse.

There are still some bugs to work out before I can create a full video made up of several clips. For really short intervals (where filling the camera buffer occurs) or really long intervals (where... I am actually not sure - I guess the camera goes to sleep?), things sort of fall apart. I think there is a ‘butter zone’ where this process will work nicely, although I have not yet determined the bounds for it. I also intend to try out other periodic functions (e.g., a sawtooth wave or triangle wave).

After shooting this video, I looked to see if anyone else has created what I was calling non-linear time-lapses. And the answer to that is yes… sort-of. There are some really interesting "non-linear" time-lapse projects out there. But it does not seem like anyone has posted anything quite the same as my clock (which I am now calling an ‘unequal-interval time-lapse’ – until I can think of something more pithy).

The coolest example of ‘non-linear’ time-lapse videos are the following videos from Watch them and see if you can work out how they did it. Scroll down for my attempt at an explanation (note that there website has some behind the scenes videos which are pretty interesting too).

As best as I can tell, the rig which moves the camera (either sliding along a track or turning for a pan) moves the camera quickly back and forth between frames to capture the scene at any given time from more than one location. So, for the truck example:
- a frame is captured at the front of the track (this appears at the start of the video);
- then a frame is captured at the rear of the track (this appears at the end of the video);
- then a frame is captured *almost* at the front of the track (this appears just after the start of the video);
- then a frame is captured *almost* at the rear of the track (this appears almost at the end of the video);
- this continues, with frames being captured alternatingly on the front and rear halves of the track with each one closer to the centre;
- the last frame captured during the shoot would be taken from the centre of the track and would appear be in the middle of the video.

Sorting those frames out and putting the video together must have been a nightmare!