In tests where we apply a scale factor to a similar source, the recording surface will produce different results to image as the scaling of the source is performed on the geometry and not on a raster image and so avoids scaling artefacts.
