[Freesurfer] terrible movement; salvagable?

I have a pretty terrible T1 from a child. The functional data, acquired first, are fine. But the T1, acquired last, has pretty serious movement artifact (attached image). While I would not use this particular scan for morpohological analysis, it seems like a reasonable surface reconstruction could be used to project functional data.

My problem is, the original T1 seems to have a better grey-white boundary than the intensity-normalized image. Is there a way to adjust the parameters for normalization to try to improve this? Would this be adjustments to only the autorecon1 series? That is, if I can improve the contrast, should autorecon2 and 3 run normally? Are control points the only way to improve this?

Anthony