Another factor to consider with H.264 is the computing power you need to decode it. The number of cameras a server can decode depends on the specific type of H.264 being used. Baseline and Constrained Baseline H.264 cameras use less server processing but, in turn, deliver lower quality images. Main Profile H.264 cameras deliver very good video quality with slightly higher processing requirements. Everything above Main Profile delivers very high video quality, however the processing required to decode those profiles can be prohibitive. Fortunately, the cost of processing continues to drop, thus making H.264 a good solution in more and more applications.
Finally, if you are designing a network for H.264 cameras make sure to account for "worst-case scenario" traffic. A perfect example to illustrate this is for school surveillance. During class there is no-one in the hallways and that is where H.264 really shines. During periods of little or no activity, it delivers excellent video using a fraction of the bandwidth of MJPEG cameras. However, when a class lets out, there will be a significant increase in the bandwidth needed. It should still be less than MJPEG bandwidth, so you are still getting the bandwidth benefit - but do not get caught designing your network for "best case scenario." If you design your network using the bandwidth for when the hallways are empty and that fire-alarm goes off and all the classes rush into the hallways at once, you could bring the network to a grinding halt and lose all that video.
MPEG: "I'm not dead, yet"
So, while H.264 is a strong choice and getting stronger, MJPEG is not dead yet. It is very predictable and works well in LAN environments where video is stored locally. While it does consume more storage, nowadays a terabyte is less than $150 - so, storage is much less of an issue than in the past and the increased storage cost may be less than the increase in servers required to process high-quality H.264 video.
H.264, on the other hand, can really save storage in applications where there is very little activity or motion - in some cases as much as 60 percent - while delivering video quality that is almost as good as MJPEG.
And the Winner is...
So, which compression is best for your application? It depends! It depends on a wide range of factors including amount of motion, scene lighting, weather conditions, etc., in the areas under surveillance. In LAN installations where there is constant activity and you need to view images live, MJPEG may be the better choice as the bandwidth, processing and storage requirements will be predictable. In WAN environments, where you are only recording - no live monitoring - and bandwidth is at a premium, H.264 is probably the better choice.
Unfortunately, there are no online calculators that objectively compare one compression against another to help you make this important decision. Therefore, my recommendation is to arrange with your integrator and/or manufacturer to test drive your preferred camera candidates before you buy and measure the performance for your unique application.
As for the eventual winner, the consumer market is driving huge investments in H.264-compatible hardware, so that each year the technology's quality and efficiency is improving and the costs are dropping. Today, I see MJPEG and H.264 as neck-and-neck, however as specifiers and integrators become more experienced in designing H.264 systems, it will increasingly be difficult for MJPEG to keep up.
Paul Bodell is vice president of Global Business Development for IQinVision.