Most of the skin graphics comes in extended bitmaps, meaning that we need to do a BitBlt always in fromSymbianCFbsBitmap in order to get a regular bitmap from which pixel data can be accessed directly. By using an NVG rasterizer to get a VGImage from the CFbsBitmap, an intermediate QImage->VGImage upload could be avoided. This would mean that fromSymbianCFbsBitmap handles extended bitmaps specially and works in a similar manner to fromSymbianRSgImage, resulting in a pixmap backed solely by a VGImage, without any pixel data in main memory (at least initially).