There is a fundamental problem with video compression systems for surveillance purposes. Video compression solutions are developed primarily for broadcast purposes. In the broadcast world the horsepower required to compress is of little relevance, as it only needs to be done once (to generate the program for broadcast). The decompression and rendering is done millions of times in millions of set-top boxes - and usually only one stream at a time.
Surveillance requirements are quite different. We require to compress hundreds, perhaps thousands of channels (i.e. however many cameras you have), so the horsepower to compress really matters - this is what makes IP cameras expensive. Most channels are simply recorded and typically 97% of that recorded video is never viewed and is therefore never decompressed.
Another difference in surveillance is that for a live-monitored system, we require to decompress many video channels at once (for a video wall for example) and as we all know, this requires a great deal of PC horsepower and does cause issues (such as image stutter and dropped frames) when systems are trying to display many mega-pixel camera streams at the same time.
What we really need is a video compression system designed from the ground up for surveillance purposes, taking into account specific requirements like high-noise low-light environments, frequent fixed image backgrounds, minimum motion blur and taking account of evidential integrity aspects too.
It would need an organisation like ONVIF or something similar to establish any new compression technology as a standard, but who would develop it in the first place ? It would have to be some sort of open source movement, but I'm not sure the market is big enough to generate the interest in doing such a project.
So it looks like we'll just have to accept whatever the mainstream broadcast industry develops for itself and adapt it as best we can to surveillance needs, as this is what we have always done.