Closed mwcraig closed 8 years ago
This is mostly a matter of making sure that images are not promoted to a higher precision type than they need to be to carry out the operation (generally float32 ought to be good enough for 16-bit images).
float32
This is mostly a matter of making sure that images are not promoted to a higher precision type than they need to be to carry out the operation (generally
float32
ought to be good enough for 16-bit images).