Thanks, guys, but that's not really what I'm getting at. I don't have any issues implementing the lerp function for camera movement. But when your camera is following your player as the player moves you are essentially changing the start and target positions of the lerp every frame. Which... is sort of incorrect usage for the lerp method. ...
It's not. I can give the lerp function whatever I want. It depends on the
context if I use it rightfully or not. And the context you are looking at it
is the wrong one leading to wrong conclusions because you think something
is interpolated.
... This is also why the lerp produces the laggy camera (assuming the time/percentage value is not too high). I think this explains what I'm getting at nicely:
http://www.blueraja.com/blog/404/how-to-use-unity-3ds-linear-interpolation-vector3-lerp-correctly
So... just wondering if anyone has taken this into account or if everyone just changes start and target value every frame and adjusts time/percentage value.
The first reply in the link you've posted is right. The lerp function was used
in another context, i.e. to damp something out, i.e.;
Code:
current = Lerp(current, target, T)
It looks like a linear interpolation, but lerp is used here for ease and not
to actually interpolate values even if it looks like it. And indeed, as the
first reply says, the lerp function, as used above, is an approximation to a
more complex function containing an exponential function. Approximating
exp(-x) to first order leads to 1-x, which will cast the complex function
into a lerp one. The approximation is good enough because x will be small
when x=dt (1/60s for example), which is the case above.
What lies behind
current = Lerp(current, target, T) is actually a low-pass
filter with a given damping coefficient determining the cut-off frequency of
the filter. If the damping is large the exponential in the filter can be
eliminated by approximating it with 1-x casting, after some algebraic
manipulation, said complex function into a lerp one.
But try for yourself! Here is a simplified low-pass filter where I removed
most of the filter terms for clarity;
Code:
float inertial_damp = (1.0f - expf(-dt));
current += (target - current)*inertial_damp;
Simplify exp as said above and you will arrive at your lerp function, which
gets used correctly, no? ;)
