The video below provides a quick overview of the models and manufacturers included:
Cameras were ranked across 10 criteria:
Field of view options: Repositionable multi-imager models varied greatly in how closely imagers may be positioned, whether they can be panned or rotated to corridor mode, and how easily they may be moved.
Zoom range: Some tested models include fixed lenses while others include motorized varifocal options with big differences in focal length range.
Size / weight differences: Cameras ranged from under four pounds to over 15, with sizes ranging from low profile and small diameter to over 12" wide.
Daytime details: Cameras were tested at ~300 lux at distances ranging from 20' to 80' to see how well lit scene performance differed.
Night details: Cameras were tested in a very dark <0.1 lux scene to show differences between IR cameras and non-IR models.
Low light detail: Low light color performance was tested at multiple ranges at ~2 lux to see how details differ.
WDR performance: All models tested included true WDR performance, tested to see how each handles strong backlight and darker areas of the scene.
Bandwidth efficiency: Smart codec bandwidth was tested in day and night scenes using H.264 and H.265, included in most cameras, to see how bitrates compare.
Video analytics / VMD performance: Cameras' on board analytics and VMD performance varied drastically, from highly accurate to easily triggered by everyday nuisances like foliage, shadows, or animals.
Pricing comparison: Pricing ranged from a low of ~$900 USD to almost $3,000, with higher price not necessarily equaling higher performance/more features.
Plus we examine additional factors which did not impact rankings but are key information for decision-makers, including:
VMS H.265/H.264 streaming and configuration support
VMS video analytic/VMD event support
Number of VMS licenses required
Country of origin
And we declare winners for best value, worst value, best low light, best positioning, and best analytics.
None of them stitch views. Stitching is typically only used in fixed multi-imager models (180° or 360°). Because the imagers of this and other repositionable models may be moved, and there is likely to be gaps and overlap in where cameras are positioned, they don't stitch.
Budget, scene requirements, budget, target distance, and did I mention budget? Ten cameras = $5K difference. I have both camera types in service, and I do like the 32MP, but when the intended viewing area only requires a 12MP, why spend the extra $500 that could be put towards another camera installation somewhere else?
Finally talked to our Sony rep about this. According to our rep, Vivotek isn't supplying anything to Sony for this camera. Rather, Sony and Vivotek happen to be using the same third-party manufacturer for the housing and mount. Sony and Vivotek are merely dropping their parts into the housing. Sony is not using any internal parts made by Vivotek.
I guess that explains why the weight really isn't identical ;) The Sony is approximately 1660g and the Vivotek is 1572g. If the housing is the heaviest part it makes sense for them to have similar weights, but you can see there's a difference of 88g.
Rather, Sony and Vivotek happen to be using the same third-party manufacturer for the housing and mount. Sony and Vivotek are merely dropping their parts into the housing.
Perhaps they’re dropping the same lenses (as pointed out by Ethan) in as well:
Now it would appear they are using a different sensor, with a slightly larger one for the Vivotek 1/2.7” vs the Sony 1/2.8”.
But one question, since angle of view is determined by both sensor size and focal length, wouldn’t the angle of views likely be slightly different if projected onto to differing sensors by the same lens?
Great test IPVM! Multi-imagers are currently becoming our most sold units. This test aligns closely with my much less scientific observations between Axis, Avigilon, and Panasonic. It is great to see some real numbers and detailed testing.
Interestingly, there is a correlation between power consumption and night time image quality, a contrast to:
Correlation does not ensure causality, of course, but Axis dim low-light images came with lower power consumption while Avigilon and Hikvision, who did not much better, had much higher power consumption.
In this context, it seems to make sense to me that the greater power consumption allowed for stronger IR lighting, agree/disagree?
Avigilon also has their analytics running and probably sucking up a fair amount of power. Did anyone notice the flat metal heat sink on the back of the housing? When my techs saw that the first thing they said was "this must run hot".
Easy, AVO Mike, it's not a criticism. Avigilon's IR ring is optional but a strength relative to competitors as the test shows. Better to consume 20 watts more power than have a poor quality image at night.
Appearance Search does not run on the camera. Appearance Search requires Avigilon cameras with analytics (or AI appliance) to feed the servers for Apperence Search to work so yes the H4 MH camera will work with Apperence Search.
Can you advise on what version of ExacqVision your testing was conducted on? In addition what was the firmware version of the Vivotek camera? I have an install that we took over and looking to replace some old Arecont 180 cameras with multi-sensor cameras and these are the only multi-sensor cameras that I have see that take a 24V AC power input. This is beneficial based on the setup in place and cabling installed.
Imho, bitrate comparison must only come together with: 1) static image quality and(!) 2) dynamic/moving object image quality.
It's easy to make such a codec that will produce very low bitrate and very nice static picture quality but would be useless on dynamic scenes.
To compare apples to apples, while you do bitrate comparison/quality comparison you always must provide: bitrate, static object image, moving object image. Otherwise, it might and likely will be misleading.
Why were the sample shots taken with different backgrounds and sometimes with the object at different angles? (There are a few with these differences above. In particular, I'm referencing the WDR Performance Comparison.) Can't both of these variables can account for different affects of light on the scene?
It has to do with how the VMS chooses to handle multi-channel devices. Some will consider anything with one MAC address to use one license. Most notably, Milestone has done this for some time. Others require a license per video channel by default, but will charge a single license after the camera is officially integrated.