3d-3d correspondence revisited