Ever since travelling on the Trans-Siberian Railway some 10 years ago, I’ve wondered how one might capture such a journey: photos do not do the ever-changing view justice, and video remains difficult to consume. Even time-lapse videos do not allow the journey to be viewed as a whole, meaning macro changes are lost.
On a recent trip from London Euston to Chester, I tried something new. Using the very simple technique of joining fixed-width strips from each frame of a video, I was able to produce the following:
This represents about 2 minutes of video with each frame contributing 2 pixels to the overall image. Increasing the width of each frame to 5 pixels, objects on the horizon have roughly the correct proportions:
The vertical banding is the result of objects in the foreground, such as a lamp posts or telegraph poles, which fill a full frame of video. It should be possible to remove these with some basic image analysis, comparing neighbouring frames and simply dropping any frame that differs too much from previous and subsequent frames.