A time lapse video assembled in ImageJ

Time-lapse video

A time lapse video shows action at a faster pace than it had taken place. It is the opposite of a slow-motion video. Cinema as well as video is implemented as a sequence of images shown fast enough to deceive our brain into perceiving movement as continuous. Thus, both in time lapse and slow motion, the trick is to display the frames at a different frequency than when acquired. The frequency, described as frames per second (fps), needs to be fast enough to deceive our visual system. A rate of 24 fps is the slowest usually used but 50 and even 60 fps are used to ensure that fast movements, especially on large screens, appear to viewers as smooth. Consequently, in practice what is altered to accelerate or slow down action, is the rate at which images are acquired.

Video frame rates

Many years ago, TV sets used the main-power line frequency as a time reference, thus 50 fps was used in Europe and 60 fps in the USA for NTSC and PAL standards. The frame rate used for films has also increased over the years. This is why in old films, such as those by Charlie Chaplin, action seems too fast when projected using newer equipment or at a frame rate matching that used for TV. This is not how they were intended to be viewed, but modifying the frame rate has been very challenging until recently.

Some modern cameras, and even smartphones, have features to facilitate both time-lapse and slow motion video capture. In some cases, the camera even assembles the video ready for viewing. In any case, maximum control can be had by acquiring individual photographs and assembling a video from them. This is the approach I used for the video shown below.

Assembling a video in ImageJ

1. Capture images at regular time intervals

   • Keep in mind that normal video frame rates very between 24 and 60 frames per second.
   • Aim at a minimum of \(15 \times 60 = 900\) images per minute of the final video.
   • Based on the desired time speed up compute the time step for image acquisition. E For a 2 min-long video for a whole day of action, we can compute the time speed up factor by dividing the number of minutes in the action by the length of the accelerated video also in minutes:

   \[\frac{24 \times 60}{2} = \frac{1440}{2} = 720\]

   If we aim at 24 frames per second a 2 min-long video requires in total 2880 images, considering the time acceleration factor above, 2 images per minute needs to be acquired during 1 day = 24 h = 1440 min. We should then acquire 1 image every 30 s.

   • Some modern cameras have a setting for sequential acquisition of images. If not, an external shutter release with this feature can be used.
   • The camera must remain switched on for the whole of this time, so normally the camera will require external power as the battery charge rarely will last this long. Some newer cameras can simply be plugged into an USB power supply or power bank. For some older camera “dummy batteries” can be used to supply power from a suitable charger.
   • During image acquisition set the camera on a tripod or other support.
   • Consider if you want the changes in illumination to be visible in the video. If yes use a manual exposure setting suitable for the most luminous condition expectes (or /or lock the initial auto-exposure setting, possibly manually adjusted). Otherwise set exposure to automatic with a fixed shutter speed.
   • A large enough memory card should be available, considering the size of the images acquired. In the example, assuming 1440 files each one 10 MB in size, about 15 GB of storage space would be needed.

2. Edit and convert the captured RAW images consistently. (If the images captured are already as JPEG out of the camera, this step can be skipped.)

3. Export the as JPEG files with original file names as these names will in most cases match the sequence of acquisition, storing then in folder with no other files.

4. Open the folder in ImageJ using the menu entry “File > Import > Image sequence” to open the folder.

5. Export the video from ImageJ using “File > Save as > AVI…”, choosing the desired number of frames per second (fps).

Steps I used to create the linked video

1. Use time-lapse setting in camera to take a series of 300 photographs, one every 10 seconds, using auto-exposure lock. Set camera on a tripod. As this took only 50 min, it was possible to rely on the camera battery for power.
2. Import the 300 images into Capture One (version 12.1). Edit the first image including correction for perspective and cropping. Select the 300 images, copy the edits from the first photograph to the remaining 299.
3. Export the 300 photographs JPEG, setting long edge maximum length at 1600 pix.
4. Read the 300 images into ImageJ (version 1.52p) using “File > Import > Image sequence”.
5. Export the video from ImageJ using “File > Save as > AVI…”, choosing the desired number of frames per second (fps).

Equipment

Camera Olympus E-M1 Mk II and M.Zuiko 12-40 mm f:2.8 objective. Camera on tripod. Zoom objective set at 12 mm, f:5.6, 1/400 s, ISO 200 (manual exposure setting),

Caveats

It is possible to create the video in camera, but I do prefer to convert from raw (ORF) and edit the images in Capture One before assembling the video.

Note

Although I used Capture One any RAW converter that can export JPEG files can be used instead. One could also use JPEG files from the camera.

Any camera, or a smart phone that can be internally or externally triggered to take a series of images with consistent exposure settings and equally spaced in time can substitute for the Olympus E-M1 Mk II camera I used. A camera that has built-in support for time lapse like the E-M1 Mk II only makes things a bit easier.

Other Examples at Flickr