Spatial demultiplexing of data (e.g. video data) is employed in a digital recorder comprising four heads (A, B, C, D) organized as 2 groups (A/C; B/D) each of 2 heads, whereby a field of data comprising a plurality of lines is spatially demultiplexed into i.times.4 data processing channels such that each data processing channel receives a different set of pixels including alternating pixels on every fourth line. The technique finds particular application to the recording of video data in compressed form as the distribution of the pixels to the data processing channels enables efficient compression to be achieved. The video data can be decorrelated before compression. The spatial demultiplexing is then performed on the decorrelated data.