The theorem of BridgelandKingReid says that if $M$ is a smooth quasiprojective complex variety of dimension at most $3$ on which a finite group $G$ acts such that the canonical sheaf $\omega_M$ is $G$equivariantly locally trivial, then $M/G$ has a crepant resolution, $Y$, such that $D^b(Y)\cong D^b_G(M)$. In particular, $D^b_G(M)$ inherits from $D^b(Y)$ a new (?) tensor structure. Does this tensor product have a nice interpretation in terms of complexes of $G$equivariant sheaves?
