If you have a SDE say:
$dX = \mu dt + \sigma d W$,
where $\mu$ and $\sigma$ are not necessarily constant. Say, this process corresponds to a probability $P$. If you apply a Girsanov transformation and obtain a process $d\hat{X}$, under a new probability $\hat{P}$, is this new probability density always that of a Brownian motion density? I.e,. something like:
$\frac{1}{\sqrt{t} \sqrt{2 \pi}} exp \left[-\frac{1}{2} \left(\frac{x}{\sqrt{t}}\right)^2\right] dx$,
where this is the density of $d\hat{W}(t)$ at $x$?