- From: Dr. Olaf Hoffmann <Dr.O.Hoffmann@gmx.de>
- Date: Thu, 25 Oct 2007 13:52:53 +0200
- To: www-smil@w3.org
..... > > > ------------- > > > > 12.3.2 > > > > 'It will produce a simple pendulum swing on the target > > (assume that the target is a pendulum shape with the transform > > origin at the top): > > <animateTransform type="rotate" from="20" to="-20" dur="1s" > > repeatCount="indefinite" > > accelerate=".5" decelerate=".5" autoReverse="true" ... /> > > > > The pendulum swings through an arc in one second, and then back again > > in a second. > > .... > > This produces a realistic looking animation of real-world pendulum > > motion.' > > -> Note that the motion of a (rotating, mathematical) pendulum is an > > anharmonic oscillation. It is technically not easy to build a > > pendulum > > with such a motion, therefore real-world pendulums behave >> different. > > -> The motion related to these attributes is that of a constant force > > (free fall close to the earth surface) as far as I understand this. > > With this example it should be possible to produce a quadratic spline > > approximation for a harmonic oscillation, because it includes > > autoReverse="true", however I did not check, if the given example > > really is the correct quadratic spline approximation for a sine > > related > > > to a harmomic oscillation and it is not the motion of a pendulum, > > especially not for large amplitudes. > > -> For (an)harmonic oscillations the autoReverse is still very useful, > > but the approximation of the motion requires itself a > > values-animation with calcMode spline and maybe keyTimes. > > Additionally the values list needs to be calculated symmetrically > > around '0' to make use of the autoReverse. > > Changed the wording to "... a rough approximation of a pendulum swing". > Note that it can hardly be expected that an implementation is accurate > enough to tell the difference. > '... an approximation of a pendulum swing' would be already sufficient, qualitative or even quantitative evaluation on the approximation quality for whatever purpose can be left to the reader/user in this case... .... > > ------------- > > 12.3.3 > > > > wording? > > > > 'r(t) is the speed modification due to acceleration and deceleration, > > at any time t within the simple duration.' > > > > -> as far as I understand this, r(t) is the run rate itself at any time t, > > not its modification, this would be dr/dt and would be called > > acceleration. > > better: > > -> > > 'r(t) is the run rate, time dependent due to acceleration and > > deceleration, > > at any time t within the simple duration.' > > Actually it *is* a modification, since there is a cascading effect if > time manipulations are nested. Well, if r(t) is something different than r, this requires another symbol/glyph, especially if it appears in the same section/chapter. If r is a constant, then r(t) = r, constant and nothing else. To describe the change or r one may write dr(t)/dt, especially if r is constant in time, dr(t)/dt=0. If the recommandation has another term/variable/entity in mind, another symbol has to be used, for example s(t) is the speed at each time t during the duration of the element. This is not much related to the meaning of the local 'time' t in the current position of the document. To get the relation to an 'ordinary time', lets call it 'o', one may write dr(o)/do = dr/dt * dt/do. > > The run rate r is defined as the maximum speed during playback. The > speed accelerates from 0 to the run rate during the acceleration phase, > runs at the run rate until the deceleration phase, and then decelerates > to 0. Using this definition the total time that is needed to play the > element doesn't change. > > r(t) is the speed at each time t during the duration of the element, > taking acceleration and deceleration into account. But because of > cascading effect, it isn't necessarily the speed, but just the local > modification of the speed. > > > ----
Received on Thursday, 25 October 2007 13:07:27 UTC