'cover': The smaller axies of image (x axies) should match smaller axies (x axies) of container. So some part of image might be cut off.

 

'container': The large axies of image (x or y) should match larger axies (x or y) of contaner. So that the whole image should be visible.