Vorwort | Filmdaten bis 1920 | Filmdaten ab 1920 | Filmdaten noch nicht hier | Nicht-Filmdaten |
---|
Most of the researchers in time-psychology or time-physiology look at the reproduction of short time-intervals or the daily or annual periodicity. But all these theories of time are able to predict experiences only very scarcely. Otherwise this new one:
This disposition proofs the experience, how many time-intervals are processed. It is important that the time-measuring is unconscious.
The first step was the observation, that film-shots have a non-linear distribution. If the time scale is linear, one get for modern films a lognormal distribution, but for old films only some peeks. After a transformation of the time scale to 2 exp n (summation of all events between the units) one gets for all films a normal distribution (or nearby, as the events are so rare).
For evaluating if this time scale is typical only for films or not, there were tested eye-movements and sleeping-movements. The results were similar distribution, only the maximum are found at other positions.
So the hypothesis would be possible: Human beings order the experience of time intervals in a normal distribution by a logarithmic time-scale (like the octavo in acoustics; Fechner-law). This is very economical.
Some questions arise out of this result:
Here a short lesson how to use my method (I have a little program):
You have to count all shots between the following intervals:
1 2 3 4 5 6 7 8 9 10 11 seconds (or pictures or cm/inches)
and so on.
Then you get a curve of the movie "La passion de Jeanne d' Arc" (1928; Dreyer)
and the many lines of the movie "Der Kurier von Lyon" (1911). (Left part of the
picture).
Now you are doing the same but you are counting all shots between the
following intervalls:
1 2 4 8 16 32 64 seconds (or pictures or cm/inches):
Then you get curves (right part of the picture)
nearly like a Poisson distribution: the line = Passion, the points = Kurier.
The movie is longer, then you get more shots, then the curve is more similar to the Poisson distribution, and a very similar distribution you get for all films, only the maximum varies - falling down with the years: the "turning-point" is ca 1916. 1910 the maximum is about 20 sec, 1920 it is arriving the nearly modern point with 6 sec. With this method it is possible to find how many pictures per second ist the velocity.
If somebody wants a finer set of points he may look for interpolations (i.e. second-number * 1.414) => 1 - 1.4 - 2 - 2.8 - 4 - 5.6 - 8 - 11.3 a.s.o.
A very interesting result is: If you have only the length (silent movies!) you may try different velocities (16, 18, 20, 22, 24 frames per second) - and the smoothest curve will give the (nearly) correct projection velocity.