The increasing production and delivery of video contents in higher spatio-temporal resolutions is currently demanding for higher compression efficiency, notably beyond the state-of-art H.264/A VC video coding standard. Since the human viewers are always the last quality judge, the compression efficiency can be improved by exploiting the properties of the human visual system. This paper studies the integration of two perceptual coding tools, namely coefficient pruning and perceptual quantization step adjustment, in a H.264/A VC compliant video codec architecture. The obtained perceptually driven video codec allows reducing the coding rate up to 35% for the same visual quality in comparison with the reference H.264/A VC standard.