Experientially, I've figured that the h264 goddesses are pleased when video input is:
limited in color (using 2 colors, plus anti-aliasing)
limited in detail (vector graphics, simple primitives)
straight lines over curves (triangles and rectangles compress better than circles on average)
linear movement (falling downwards shapes is better than growing shapes)
@paul well I have a 20 yo mjpeg low-res video playing with the video cam noise and exagerated motion vector movements, and ... it's a hell to translate correctly in h264... I'll fiddle again next days with it but it's indeed really difficult to encode without loosing the point ...
it's really a codec tested against "some" sort of imagery ...