Making Movies of Trajectories

Todo

Explore having a trajectory movie hosted on this page.

Important

Cartopy 0.18.0 is required to successfully render maps and pass our internal tests.

To render a movie, we render short subsets of trajectories over and over. As such we can re-use all of the arguments and algorithms we already have for rendering trajectory maps with just a few additions for movie duration, frames per second, and trajectory length. In the examples below, TRACKTABLE_DATA refers to the tracktable-data directory of the Tracktable repository.

Terrestrial Movie

We’ll begin with a short movie (10 seconds long, 10 frames per second) where each moving object has a trail showing the last hour of its motion.

$ python -m "tracktable.examples.python_scripts.movie_from_points" \
     --trail-duration 3600 \
     --trajectory-linewidth 2 \
     --fps 10 \
     --duration 10 \
     TRACKTABLE_DATA/python_example_data/SampleTrajectories.csv \
     MovieExample1.mp4

Todo

Figure out why our videos won’t play in the browser, maybe host gifs instead.

This will encode a movie using vanilla MPEG4 that should be playable by anything less than ten years old. Quicktime Player, iTunes, and Windows Media Player can all handle this. If you don’t already have VLC installed we recommend that as well.

We have two more features to demonstrate here. First, instead of having the trajectory lines be of constant width along their length we can have them taper as they get older. We do this with --trajectory-width taper, --trajectory-initial-linewidth and --trajectory-final-linewidth. Second, we will also put a dot at the head of each trajectory with --decorate-trajectory-head and --trajectory-head-dot-size.

$ python -m "tracktable.examples.python_scripts.movie_from_points" \
   --trail-duration 3600 \
   --trajectory-linewidth taper \
   --trajectory-initial-linewidth 3 \
   --trajectory-final-linewidth 0 \
   --decorate-trajectory-head \
   --trajectory-head-dot-size 3 \
   --fps 10 \
   --duration 10 \
   TRACKTABLE_DATA/python_example_data/SampleTrajectories.csv \
   MovieExample2.mp4

Todo

Figure out why our videos won’t play in the browser, maybe host gifs instead.

Cartesian Movie

As with geographic data, we can also make movies from data in flat Cartesian space.

$ python -m "tracktable.examples.python_scripts.movie_from_points" \
  --domain cartesian2d \
  --map-bbox -100 -100 100 100 \
  --trajectory-linewidth taper \
  --trajectory-initial-linewidth 4 \
  --trajectory-final-linewidth 1 \
  TRACKTABLE_DATA/internal_test_data/SamplePointsCartesian.csv \
  MovieExample3.mp4

Todo

Figure out why our videos won’t play in the browser, maybe host gifs instead.

Note

Recall that trails are colored by their progress from start to finish and the default colormap (“heat”) is black at the beginning. If you would like to see them bright and vivid right from the start, add an argument like --trajectory-colormap prism (or any other Matplotib colormap you like).

Using Parallel Processing

Where available, we can make use of multiple processes simultaneously for the rendering of movies. In this example, the result will be the same as Cartesian example above.

$ python -m "tracktable.examples.python_scripts.parallel_movie_from_points" \
  --processors 8 \
  --domain cartesian2d \
  --object-id-column 0 \
  --timestamp-column 1 \
  --x-column 2 \
  --y-column 3 \
  --delimiter , \
  --map-bbox -100 -100 100 100 \
  --trajectory-linewidth taper \
  --trajectory-initial-linewidth 4 \
  --trajectory-final-linewidth 1 \
  TRACKTABLE_DATA/internal_test_data/SamplePointsCartesian.csv \
  MovieExample4.mp4

Todo

Figure out why our videos won’t play in the browser, maybe host gifs instead.

Note

The efficiency of this method is greatly dependent on the underlying operating system and the complexity of the movie being rendered. For example, on Windows, this method is likely to be slower than using the single threaded version.