There is almost no difference between MJPEG and JPEG2000 from an image quality standpoint (JPEG2000 is perhaps ~10% more efficient in terms of compression but that is not crucial).
Depending on the i-frame interval and bitrate settings, H264 may pose problems if these two settings are not correctly tuned.
If you are worried about this - you can set your ONVIF camera to MJPEG mode.
To answer the question there is no reason why Avigilon is the only solution that will work.
Avigilon's JPEG2000 has one advantage which is that it can scale the image down without extra CPU usage (known as a scalable codec). Avigilon calls this HDSM and it optimizes bandwidth when transmitting large numbers of streams over narrow band networks. H264 and MJPEG cameras deal with this problem by using the dual streaming technique.
In terms of whether the client needs a 5MP camera - do you want to cover all multiple tellers with a single camera? If so, and the ceiling is not too low - a 5MP camera can reduce the number of cameras needs to cover the scene. In general a 5MP camera can replace 2 x 1.3MP cameras if the scene has no obstructions. But in terms of searching for footage more cameras may prove more user friendly.
Another note about 5MP cameras - the max framerate is typically 15fps or less - this may not meet you needs (fast cash counting).
I think the one key advantage Avigilon has is not JPEG2000 but their automatic image adjustment / enhancement feature (which is a VMS feature rather than a camera one).
The other key issue is compression level. Regardless of the CODEC used, one can configure that amount of compression that is applied. In our Avigilon JPEG2000 camera test, we found their bandwidth levels where far higher than equivalent resolution / fps MJPEG cameras. This is likely because Avigilon defaults the compression level lower than their competitors (which can capture more fine details at the expense of more bandwidth).
As Bohan mentions, if you use H.264, keep the i frame interval (1 second or less) and the compression levels low, to maximize details. What level depends on how wide the FoV is.
One other important note. Avigilon has discontinued all 5MP and under JPEG2000 cameras. So beware about availability.
I heard a single 29MP camera can cover 95 tellers.
I would actually be careful with Avigilon's VMS and their image adjustment features if you're looking for superfine details like currency denominations. I've seen it introduce some artifacts where there were none, resulting in text being less clear, like in this image:
IPVMU Certified | 07/11/13 09:01am
I find it interesting that the subject matter always surrounds quality and bandwidth/storage. Given what is available on the market anyone can get quality video with decent frame rates and good storage and it doesn't have to be Avigilon.
For banking finding footage, exporting and handling video is crucial for any bank that wants to reduce fraud and losses. This is rarely if ever reviewed in any depth or detail. I've looked at Avigilon multiple times and from this perspective they are not very good. There are several companies that out perform Avigilon in this department.
One must pay attention to many issues when trying to achive a quality image, especially in this case. Lighting plays a huge part as the scene must be illuminated properly to suit the camera to use as high a shutter speed as possible to prevent blurring. I agree with John on all his comments, in particlular in regards to how slowly a teller counts cash with a customer coming from being properly trained. Other fraudulent activity can also be prevented with proper training and a teller adhereing to the banks policies when handling cash so a single 3 megapixel camera per teller properly set up with MJPEG recording will work in most applications. Storage is getting cheaper these days so cost should not be the main issue to obtaining good records and again as John said, more emphasis should be on the search and archiving features of the VMS.
Interesting dialog - I am a former Avigilon Rd and still a huge Avigilon fan. I agree with most of what you have pointed out and Avigilon is not the only game in town. The real significant difference between Avigilon and everyone else is flexibility. Avigilon can compete head to head with anyone and do well or most often win based on traditional design. Avigilon however, can show H.264 cameras and place them where they are best served and at the same time incorporate JPEG2000 cameras in strategic places. Even if a customer doesn't need a huge camera right now - they love to know that they have that option. The Avigilon VMS blows the competition away - that shouldn't be up for discussion as evidenced by Avigilon's rapid growth. It's always nice to see multiple spins on this.
IPVMU Certified | 07/11/13 05:13pm
We have compared Avigilon with lots of other VMSs, in pixel searching (smart search in other softwares) when you select specific area for the software to be searched the result in Avigilon is incomparable. Just after you click the search button the result is ready.
We also tested Axxon and Digifort, searching an specific area in a camera in period of a week took 2 and 3 hours.
I think the main reason is that Avigilon is using metadata when storing videos and they refer to metadata instead of source images.
Luxriot, HD Witness and Milestone also use motion metadata for instant smart/pixel searches.
I am in the process of aiding a 300-location national chain store with the upgrade of their month-old Avigilon system to my equipment. Cost, real-world functionality, and practicality are all reasons they're returning the Avigilon gear, and they claim the product was "grossly misrepresented by the Avigilon reps with whom they were working", threatening to sue should the manufacturer choose not to take the gear back- 12 sites were deployed. Apparently their complaints held water, because the aforementioned is taking all the equipment back.
I think the key here, as the customer has found, is that Avigilon is pretty much just like everything else, with marketing dollars being the key exception.
Matt, please share some specific real world functionality that Avigilon lacked for this customer. What specifically are you claiming Avigilon overpromised and could not deliver?
One was they were told Avigilon was PCI-compliant when they are not- only two manufacturers can make that claim, but the primary point of contention was the real scalability of Avigilon's streaming technique, which when challenged with a handful of locations running ISDNs (yes, they still exist, and in many parts of the country are still some folks' only option), which Avigilon claimed wouldn't have any issues, yet when taxed with the challenge, wouldn't even DISPLAY a video stream.
Needless to say, our equipment not only worked, but worked WELL in the very same conditions with 2MP cameras, identical conditions.
" The Avigilon VMS blows the competition away - that shouldn't be up for discussion as evidenced by Avigilon's rapid growth."
Your if/then statement is flawed.
1. Their VMS existed (in primarily its current iteration) for many years before they experienced 'rapid growth'.
2. 'Rapid growth' of an existing product is almost never an indicator of the product being the 'best' - as your comment purports to 'prove'.
3. Increases in marketing spend generally translate to increases in sales (unless you really suck at either marketing or sales).
Stick to facts when making subjective statements if you want to claim your comments 'shouldn't be up for discussion'.
IPVMU Certified | 07/15/13 12:41pm
For the record, motion pixel search is almost useless for banks investigating robbery and fraud.
If someone robs the bank what good will motion search do? People are at the teller line all day long, motion search is going to yield more video to plow through needlessly.
John, good point. Any responses to the contrary?
I can see motion based searches helpful for scenarios like "who stole the laptop or car overnight?" but not typical banking ones.