As an after thought to my last message, I will make these comments.
Those of us that have analyzed real experimental data may have a different
perspective as to what looks linear. Video analysis is an experimental tool,
not a computer game. As such, the data is real. Video files are just large
(visual) data base recordings of position versus time. They are not created
animations produced to fit physical laws. (It is possible, however, to
create animations with a spreadsheet and save them as avi files, if that is
the way you wish to use this tool, and in certain cases, this is an asset.)
When using video analysis as an experimental tool, experimental design is
just as important as with any other experimental method. And to obtain
perfect data, is not a realistic expectation. Done correctly, this
experimental approach gives comparable results to other timing methods
including spark timers, smart pulleys, and sonic rangers. Plus it has the
additional benefit of a visual recording.