Testing Axis's Q1755 HD Surveillance Camera v. 2By John Honovich, Published on Jan 04, 2010
This report shares extensive test results, video clips and training screencasts on the Axis Q1755 HD camera. The goal is to educate on how the camera works and what issues to consider when selecting or deploying this camera.
Updated Testing - January 2010
This version provides 2 major updates from the original April 2009 test:
- Testing at multiple exposure settings (the default 1/30s and 1/8s) for analyzing low light performance. While we were originally critical of the Q1755's low light performance, further tests of competitors revealed that many used much lower maximum exposure settings than Axis providing an unrealistic and unfair comparison. Q1755 tests with 1/8 second maximum exposure show dramatic improvements for noise reduction and visible detail. For background, see our critique and analysis of how exposure setting 'tricks' can impact low light performance.
- Providing test sample videos consistent with the 9 other megapixel camera tests we have done (review in our surveillance test directory).
Key Observations from the Test
In the report, I will explain a number of details, both in the written review and the series of screencasts below. Here are the observations that I found most interesting or surprising:
- The setup was easy - no focusing, no tuning of image settings - delivered high quality immediately
- H.264 bit rate varied between 1 and 6 Mbps and between 50% to 95% lower than MJPEG
- H.264 consumed less CPU than MJPEG on my system
- The visible image quality/resolution seemed to be about the same for 1080i and 720p
Below is our screencast commenting and showing the Q1755's image quality. More importantly, is a ZIP package of video clips for you to review yourself.
Download the package of Q1755 test video clips and image snapshots [link no longer available] (41 MB total). Note: the videos were generated from Axis Camera Station VMS and may require installation of the included CODEC.
Overview of Using the Axis Q1755 Camera
The video below explains the key characteristics of the Q1755 and its general use.
The other basic element to ensure is that your VMS system supports (1) megapixel cameras, (2) H.264 codec and (3) the 16 x 9 aspect ratio of this camera. The second and third points are especially important as VMS manufacturers are still in the process of adding support. A few providers have already announced support but many have not. If in doubt, check on this before deploying.
Mechanical and Physical Overview of the Camera
As you watch the video below, you may want to reference the Q1755 datasheet [link no longer available], the user manual [link no longer available].
Configuration and Software Features of the Q1755
In this video, we focus on specific features and options that are either unique or especially useful to the Q1755 camera.
Comparing 720p and 1080i image quality
For megapixel resolution, the Q1755 offers two options: 1280 x 720 and 1920 x 1080. What's interesting to me is that the image quality / details displayed were not much better at 1920 x 1080 then they were at 1280 x 720. This was generally true for day time viewing and certainly true for night time viewing.
Of course, this raises a question: Is this really a 1080i camera or is it a 720p camera?
On the other, what's most important is the effective image quality. Future tests will compare different manufacturers megapixel cameras to better understand and contrast the effective image quality provided.
Night Time Viewing
As you can see from the downloaded video clips, night time viewing is especially challenging. The camera is rated at 2.0 lux for color mode and 0.2 lux for B&W. In either mode, the image quality was very noisy and it was difficult to see many details.
If you are going to use the Q1755 in dark conditions, you should be careful.
Comparing H.264 and MJPEG - Storage/Bandwidth/Quality
I found the image quality to be similar for both H.264 and MJPEG. However, it did appear that when cars were moving, the cars moved smoothly in the MJPEG video but not in the H.264. If smooth motion and small changes for moving objects are important for you, you should consider this point.
While marketing material suggests H.264 reduces bit-rate by 80%-90% compared to MJPEG, my tests displayed greater variance. Sometimes, the savings were as great as 95% but others they were only 60%. The complexity of the scene and the amount of motion were common scenarios where the bit-rate reduction was lower. Also, at night, H.264 consistently consumed more bandwidth than the day (often 5-6 Mb/s at night whereas the same scene in the day only consumed 1 - 1.5 Mb/s). This is especially concerning given that the night time view provides less image detail.
The video below shows the measurements in action for both day and night:
Examining CPU Use for H.264 Codec
Both Axis and many VMS manufacturers acknolwedge that H.264 can require increased CPU resources. The metric often cited is 50% to 100% increase in CPU load.
I tested CPU use 5 times and for 4 of those 5 times, CPU utilization was lower with H.264 than with MJPEG. I am not sure why and I could certainly be doing something wrong. I also cannot run a screencast because it would significantly throw off the CPU measurements.
One theory could be that my graphics card is off-loading the load and therefore my CPU is not being effected. Also, since I used Exacq's software and they do not run any motion detection in their server, there was no CPU differential between MJPEG and H.264.