For async image decoding we have to decode for a specific size. Currently the request for async image decoding happens before the drawing starts. We can take the pageScaleFactor, the pageZoomFactor and the deviceScaleFactor into account easily. But the css and svg transformation both need more work to known before the drawing starts. Alternately the requests of async image decoding might be delayed till the drawing starts and all the transformations are applied to the graphics context.
<rdar://problem/30903403>