[gst-devel] GstController interpolation modes

[David Schleef]

On Tue, May 15, 2007 at 04:33:20PM +0200, Sebastian Dröge wrote:
> The only problem I see with the second derivative being zero at the
> endpoints is that we also have to handle values outside of the
> endpoints. With the slope being zero one could just continue with the
> value at the endpoints, with the second derivative being zero one could
> only continue with the same slope...

Well, you can't please everyone.  The user can just add more control
points if they don't like that behavior, as long as the behavior is
predictable (i.e., intuitive to someone playing around with a spline
editor).

> The linear interpolation currently uses the value at the endpoint for
> the values outside of them. Doing this for the cubic interpolation would
> yield to a non-continous curve at the endpoints unfortunately.

Non-continuous derivative, you mean?  Well, can't please everyone...



dave...