The integration of virtual renderings into video images is of great importance for many applications of computer graphics
such as TV and movie production and augmented reality. This submission proposes and examines methods for a smooth transition
from real to virtual image regions along the boundaries of graphical objects. This constitutes a new form of antialiasing,
which - unlike its counterpart in purely virtual images - has not been systematically investigated so far. We introduce
real-virtual supersampling based on the combination of a high-resolution rendering of the graphical elements and an upsampled
version of the video image. Several different methods for image upsampling are discussed. We compare results obtained through
these methods with simple image combination methods: image overlay without antialiasing and pixel averaging in the original
image resolution.