In a magnetic resonance method and apparatus, a series of magnetic resonance exposures of an examination subject is generated by implementing multiple first measurements (data acquisitions) with variation of a measurement parameter from acquisition-to-acquisition, which strongly influences a contrast of the first material type excited in the first measurements, implementing multiple second measurements in which a second material type is essentially selectively imaged that is less contrast-dependent with regard to this measurement parameter in a processor calculating spatial correction values for image data of the first measurements based on spatial differences between image data of different second measurements, and, also in the processor, spatially correcting image data of the first measurements (Di) and/or registration of image data of different first measurements to one another on the basis of the correction values.