Flash is vector based for the most part, but a few versions back they added objects that deal with bitmaps the traditional way, it works pretty well. But on the other hand, why bother with the vector stage at all? The image will have to be rasterised before being displayed anyway, so what difference does it make whether you have a bitmap surface you can draw bezier curves on, or a vector environment that ends up as a bitmap?

An aside, bizzarely I find it's often faster to redraw some graphics "vector style" (calling the lineTo/bezierCurveTo etc methods again) on canvas than it is to use drawImage and use a cached version of the same graphics.