Figure 7.

Controlled motion experiment. In (a), the extent of motion is shown by the difference image between the two acquisitions. With datasets mixed such that motion occurs between shots and intra-blade-pairs, both G_ref (b) and G_xc,4 (c) reconstructions lead to inferior weights, leading to shading and streaks despite that the motion correction itself is able to align the blades. On the other hand, if motion occurs between only inter-blade pairs, both methods reconstruct blades accurately and can be realigned prior to gridding like in d-e).