The QGraphicsView::viewportTransform() function has a wrong description in the documentation.

The current description is: "Returns a matrix that maps viewport coordinates to scene coordinates."

However, the function actually returns scene to viewport transformation.

This was found in Qt 4.8 documentation.

This issue was also highlighted in this post:
http://stackoverflow.com/questions/1719694/getting-the-bounding-box-of-a-constant-size-item-in-qt-graphics-views