I have the following "elastic out" curve:
$$f_b(t)=1+2^{-10t}\cdot \sin\left(\frac{(t-\frac{b}{4}) \cdot 2π}{b}\right)$$
I only need to use it in the interval $\left(0, 1\right)$ (values for $0$ and $1$ do not matter). It already describes the oscillation after "getting to" $y=1$ quite well, however, I am facing issues trying to design the dampening according to my needs and transforming this curve in general.
Problem
Above you can see what $b$ controls – I would say the dampening of the oscillation but also not really. What I want to have control over ideally is only the dampening about $1$ after the function comes back to $y=1$ for the first time, so I want to keep the initial bump past $y=1$ from say $b=0.4$, but I want more overshoot past $y=1$ afterwards.
This is a terrible sketch, but I hope that you see what I mean. The red graph would be what I have now with $b=0.4$ and the blue graph is what I am trying to achieve.
I tried a bunch of edits to the equation, but nothing I did got me close to what I want. The graph should keep oscillating about $y=1$ after some time.
Observations
Changing the factor before the $sin$ will obviously allow me to stretch and compress the whole graph, but that is not what I want.
Application
The point of this curve is to describe motion from one place to another. An object is supposed to move from say $A$ to $B$. $A$ is at $y=0$ and $B$ is at $y=1$. $y=1.1$ would mean that the object moved past the destination (past $B$), which is why this is "elastic". My goal is to keep the intial elasticity of the bounce, i.e. the overshoot past $B$, but I also want to have more bounciness when the object oscillates about point $B$ after going beyond it initially.
Best Answer
If I am understanding you correctly, you want an envelope function $g$ such that $g(0) = 0$, $g(x)\approx x$ for $x > 1$, but you want $g(x) < x$ in that range $(0, 1)$ so that $g(f_b(x))$ has lower troughs than you presently have.
If we were to use some sort of bell-shaped curve, say, a Gaussian $h(x) = e^{-x^2}$ then we could form such a function as $g(x) = x(1 - \alpha h(x/\beta)$ for two parameters $\alpha$ and $\beta$ that tell us about the strength and the spreadiness of the envelope. Some more shaping can be done by using $(h(x/\beta)/h(0))^n$ or so instead of just $h(x/\beta)$ in there. So for example with $g(x) = x(1 - 0.8 e^{-6 x^6})$ I can get a graph like
(blue is original, black is after the envelope transformation).
The only weird thing about this sort of approach is the way it messes with the slope near $x=0$.