Skip to main content

Time Lapse Techniques

· 3 min read

There's lots of high quality timelapse out there — one of the best is Morten Rustad (@morten.rustad). If you are starting from scratch, start there. The typical approach is capture relatively slow moving phenomena at 1-120 frames per hour and then running the frames at standard frame rates - 24-60 frames per second. Motion is fluid and properly blurred by keeping exposure coverage of at least 25%. So if you capture 1 frame every minute, the exposure should be 15 to 60 seconds. This is readily achieved using ND filters without compromising other exposure settings.

I've been trying something a little different to make use of high quality video cameras that capture 4-8k in log formats. In this time lapse technique, I use videos captured at typical frame rates and post process by variable averaging to get a designed motion and feel. Similar things are already done with frame rate ramp ups and ramp downs. But I wanted something extremely fluid and controllable. The main innovation here is the time scale and attention to periodicity. The ideal subjects are slow moving in real time with small changes at typical frame rates. With a high speed camera, more is possible such as changing the feel of everyday physics.

Some things to keep in mind:

  • Capture images at a relatively high frame rate. Adjacent frames should have small or unnoticeable changes on the order of a pixel. 24 fps is good for slow phenomena like clouds. 1000+ fps for faster motions is necessary.
  • Capture 50-100% of the motion using long shutter times/fill factor and manage dynamic range with ISO/f#/ND filters.
  • Stabilization and color correction on video.
  • Create frame rate curves.
  • Create new frames by averaging or other functions (perhaps weighted averaging).
  • Repeat stabilization on averaged frames.
  • Generate synchronized audio with velocity and acceleration dependence. You can also use optical flow to find types of motion in the scene and generate stereo audio that reflects that. This is a kind of sound from motion.
  • Apply camera movements like zooms and transforms with motion blur on the averaged results.