Might depend on the platform but my guess is that scaling the QImage is better.
I remember that at least at some point scaling a pixmap meant that it was transferred into a QImage internally, then scaled, then c
Create a QRect that has the size of the visible area.
Scale the image and get its rect(), take that rect's center() and move the first rectangle's center to that point.
Then use this modified rect on the scaled image's copy() and thus get the section you want to display.
Cheers,
_
Bookmarks