I’m interested in odd camera designs, such as slit photo finish cameras (historically used for race finishes lines when it’s a close call). It occurred to me that I could easily make slit photo finish images by taking a video (using my Leica D-LUX digital camera) and then writing a script to process the video (by taking a vertical line of pixels from each frame of the video, and stacking them end-to-end to make a single still image).
Then it occurred to me I could go one further - because I have a choice of which vertical line of pixels to use on the video, I could use every possible vertical line of pixels to make lots of photos finish images, and put them back into a video.
Then I realised another way of thinking about what I was doing: in effect, I was rotating (or transposing) the original video’s array of pixels (for a video is actually nought but a 3D array of pixels), swapping one space dimension and the time dimension.
So I made some videos as described. The following two pictures show a frame from an input video, and one from the corresponding output video:
To see the actual input and output videos, please visit the flickr video set.
For some actual analysis and thoughts on what is happening, and what sort of effects happen in the output videos, please visit this page on the Cookbook wiki.
(Coming soon: link to the python script for generating these videos.)