Testing Arecont Vision's AV3105 H.264 Camera

Published Jul 08, 2009 00:00 AM

Arecont Vision's H.264 megapixel product line has drawn an immense amount of interest. At the beginning of 2008, Arecont was the first surveillance company to announce H.264 multi-megapixel and is currently the only manufacturer offering H.264 megapixel up to 5MP.

At the same time, a lot of questions have been raised about image quality, CPU utilization and actual bandwidth savings.

In this test of the AV3105DN with the MPL4-10 lens, we explore the performance and benefits of Arecont's H.264 offering.

Key findings include:

  • Image quality was sharp with basic manual focusing of the lens
  • Compared to MJPEG, H.264 CPU usage was consistently lower
  • Bandwidth utilization for H.264 streams was up to 90% lower than MJPEG
  • High motion or low lux levels caused H.264 bandwidth utilization to rise significantly, reducing efficiency gains
  • Numerous problems existed using the Arecont Vision VMS software though 3rd party VMS software worked well

Product Overview / Key Findings

ArecontVision's 3105 is a 3MP MJPEG/H.264 camera rated for up to 15 fps. The MSRP of the AV3105DN is $1,190 USD. The MSRP of the MPL4-10 lens, sold separately, is $150 USD. On-line prices are likely to be about 30% less than MSRP. While Arecont Vision's cameras are supported widely (over 40 total VMS vendors), according to their recently released NVR support matrix [link no longer available] about half of their partners currently support the H.264 streams.

Test setup: The camera tested was using S/W version 64829 which is the most to up date version at the time of the test. The camera was tested on a Dell Studio laptop with Intel Core 2 Duo 2.00 GHz with 4.0 GB of RAM running Windows Vista Home Premium. The computer was restarted prior to each test and a wired connection was used to a TrendNet 10/100 Mb/s switch.

Image Quality: View the sample test videos to judge for yourself. Image detail was good displaying details of the indoor test area used. In some conditions, colors did not look true to life and this varied by light levels, etc.

Frame Rates: Frame rates were measured using the built-in display of the AV100 software (Cntrl-S to display) While the camera is rated at 15 fps, sometimes the frame rate was significantly lower. H.264 streams generally close to 15fps. However, MJPEG streams were often significantly lower (closer to 6fps).

CPU Utilization: Using the Windows Task Manager CPU monitor on multiple PCs and with the web client, AV's software and ExacqVision, the CPU utilization for H.264 was almost always lower (often significantly) than the MJPEG stream. Since this is a hotly debated item and since our test was not very sophisticated, we would limit the strength of our claim here. However, we have not seen why for multi-megabit video, MJPEG would be less CPU intensive than H.264.

Bandwidth Utilization: [Note: screencast later demonstrates these findings] 3MP H.264 bandwidth utilization was as low as 1Mb/s for well lit (200lux+) areas with minimal to no motion. By contrast, such scenes required about 12Mb/s for MJPEG streams capturing equivalent views. When motion increased (such as the intersection of a road), H.265 bandwidth utilization increased to 5 Mb/s to 10 Mb/s (MJPEG increased slightly to about 15 Mb/s). Additionally, when lux levels dropped, 3MP H.264 streams increased to 10Mb/s even for scenes with no motion (contrast to about 12 Mb/s for equivalent MJPEG streams). 

Bandwidth Utilization / Video Demonstration

Bandwidth/storage utilization is a key element in H.264's value proposition. In the screencast below, we demonstrate how bandwidth utilization varied significantly depending on motion and light levels.

The sample of the traffic intersection with swaying trees and cars is used as an example of high motion. By contrast, the indoor test is an example of low motion and more common in traditional indoor security uses. It is important to identify which scenario is most similar to an intended deployment to project the potential bandwidth impacts.

Physical/Mechanical Overview

The following video demonstrates the camera physically overviewing its size and physical features provided.

Setup and Configuration

The following screencast explains how to configure the camera. Most of the setup is common to all IP cameras.

We had challenges using the AV NVR software. The biggest problem was issues with installing and using it with firewalls. We tried it on 3 PCs. It worked fine on 1 PC where we shut off our firewall. We had to uninstall it on another PC because it created problems with the stability of our system. On our 3rd system, it forced our PC to continuously reboot. I do not know how common such issues are with others use but this is the first time I have such significant issues in VMS software during my recent test of 10 different applications. By contrast, use with the Exacq software worked fine with no problems.

Also, if you need to do a firmware upgrade, no connection to the camera can be running (web interface, VMS, etc.). Otherwise, the upgrade will time out. When all connections are disconnected, the firmware upgrade took 25 minutes.  Arecont reports that this upgrade was an especially large and out of the ordinary size.

Competitive Comparison and Recommendations

There are a few main dimensions specifiers usually consider when comparing megapixel cameras. Here are a few and where I believe the AV3105 stands:

  • Frame Rate: While the frame rate is rated at 15fps, if you really need 10 or more, you should carefully test that this camera can deliver. Sometimes it did for our tests, other times it did not. Most applications do not need more than 10 fps (especially for multi-megapixel) but if this is a requirement be careful.
  • CPU Utilization: It does not seem CPU utilization is any more of an issue than it would be using an MJPEG camera.
  • Bandwidth/Storage Savings: This is the big 'theoretical' driver of using H.264. While in some scenarios, the savings were 90%, in others they were only 10%. It definitely depends on the environment and lighting conditions being used (more motion, less light, less bandwidth savings). You should be careful in determining how much savings your conditions produced. While they could be very high, it may turn out to be not much at all.
  • Bells and Whistles: If you need them (analog or HDMI out, SD card, etc., zoom lens), the AV3105 is not a good choice. It simply does not have them. Compare to Axis Q1755.
  • Price: Arecont's product pricing is generally lower than their direct competitors such as Axis, IQinVision, often by a few hundred dollars. The AV3105 street price is approximately $900 USD (with lens). An Axis 223M, a 2MP camera is about $150-$250 USD more.
  • Image Quality: This is an important factor, one that requires very carefully controlled comparative testing. As I have not done this, I cannot offer recommendations.