LOOKING at new product is one of life’s great pleasures – particularly when the new gear offers enhanced features and improved performance. New product is never boring. It either impresses the reviewer as breaking new ground or confirms the elevated status of the former best performer.
Of course, one of the challenges of reviewing surveillance cameras is that there’s an element of the subjective introduced when assessing performance using intuition. This comes down in part to the expectations and values of the reviewer and in part to the supposed application into which the camera may be installed.
Every reviewer is subject to cognitive bias. A subjective assessment, made with the eye based on experience – that’s either direct or indirect comparison – is not the same as an entirely objective assessment made by measuring a predetermined range of relevant camera output signals. Nevertheless, the appearance of a given camera in a range of conditions offers the observer significant ability to say things about its performance parameters.
Additionally, there are particular specifications that confirm performance capabilities not related to image quality. As we know, once upon a time, security cameras were judged, perhaps wrongly, on the basis of 2 numbers – horizontal resolution and minimum scene illumination. But these days there’s a heck of a lot more to think about. There’s bandwidth consumption and associated compression choices and maximum resolutions and the ability to network with third party management solutions. Then there are considerations like warranty and local support.
On the performance side, you are looking at image quality and the purity of colour reproduction across variable conditions. You also want clarity and definition in the image – this is what gives you facial recognition and LPR at marginal depths of field. Low light performance is important, too, if the camera will not be supported by artificial lighting systems. But a camera that simultaneously offers vision deep into the background of a partly-lit scene can be considered superior.
Then there’s the ability to handle backlight. A camera that cannot handle backlight will be limited in its ability to offer meaningful images outside the foreground. If the foreground is your target area – the counter of a store for instance – then that’s great. But if you want to push the camera view outside – let’s say a perpetrator runs from a store and gets into a car – then poor wide dynamic range will mean the crook will vanish from view into a sea of white bloom as they go through the glass door into sunlight.
In my experience what is intriguing about modern IP cameras in all their guises is the enormous differences in their individual capabilities. Pick 10 cameras with 720HD resolution and all of them have quite different strengths to the naked eye. One may be better in long range low light, one better close up with low light, one will be better at long range with strong backlight and one better at closer face recognition with strong backlight.
It seems that product developers interpret what it is their main body of clients wants and in building their chipsets and software algorithms to focus on a primary capability – say low light performance – they will sacrifice processing ability elsewhere to retain what they believe the market sees as the blue ribbon virtue. I’ve never yet seen a camera that performs brilliantly at everything and this is something that should be taken into account by installers and end users. You need to select cameras on the basis of their strengths and your requirements.
Features and specifications
The camera we’re looking at is Basler’s BIP2-1300c Day/Night, which is available from Geutebruck Australia, henceforth this camera will be known as the Basler. For the purposes of illustrating some of the points I’ve made above, I’m going to compare the Basler with a couple of other cameras, as well as making observations about its performance.
For a start this camera is a compact wee thing. I think it’s perhaps half the volume of cameras from other makers and these are modest in size, too, when compared to the products of yore. The Basler is simply far more compact that most the full body cameras available on the market today at 109.7mm long x 44mm wide x 29mm high. That’s tiny and weight is just over 200 grams. It’s a nice-looking unit. Being wider that it is tall gives the camera a different look to my eye, fresher and more streamlined.
The core numbers are strong. There’s 720p HD at 30fps and the 1280 x 960 pixels are coming via a Sony CCD with retractable IR cut filter to enhance low light performance. Compression options are MJPEG, MPEG-4, and H.264 (AVC high profile levels up to 5). Pixel Size horizontal/vertical is 3.75 µm x 3.75 µm.
Minimum illumination in color is 0.34 lux (F1.2/33ms), and with the cut filter activated 0.09 lux (F1.2/33ms) in low light in black and white. My test did not include a light meter so I can’t confirm the veracity Basler’s figures but they seem close to the mark to me, if you take into account the fact the low light images are vague and grainy – as all low light image streams tend to be.
When it comes to video streaming there’s multi-encoding and multi-streaming for MJPEG, H.264, and MPEG-4; VBR and CBR for MJPEG and MPEG-4, VBR, CBR, and CVBR for H.264, multicast and unicast. Providing the grunt under the bonnet is a multimedia video processor with RAM of 256MB and Memory of 64MB. Those numbers are good for a video processor.
In the area of image setting selections the Basler is particularly strong. There’s automatic gain, exposure area, backlight compensation, white balance, electronic shutter, 180-degree image rotation, anti-flicker, electronic PTZ via AOI (API), text overlay, privacy masks, motion detection and multiple Areas of Interest (AOIs). All this is good stuff from Basler and increases the camera’s flexibility considerably.
When it comes to alarm management there’s a ring buffer configuration supported for pre and post alarm images as well as for events triggered by motion detection, or external input (real-time trigger). Alarm event images can be upload over FTP, e-mail, or HTTP. The baseline profile protocols list is very complete containing TCP/IP, HTTP, FTP, UDP, ICMP, ARP, DHCP, NTP, RTP, RTSP,* RTCP,* SMTP,* IGMP,* ZEROCONF,* QoS Layer 3 and SNMP.
The cameras operates between 0 °C – 50 °C (32 °F – 122 °F), <90% relative humidity (non-condensing), has a CS-mount for lenses and 3 configurable digital input/outputs. Physical connectivity is RJ-45 connector for 10/100 BASE-T Ethernet, full or half duplex, along with 8-pin terminal for DC power, the digital I/O and RS-485. The camera is PoE IEEE 802.3af Class 2 or 12 to 24 VDC with a consumption of around 3.5 W.
Browsers supported included Internet Explorer (recommended), Mozilla Firefox (plug-in required), Google Chrome and Opera, Safari and Conqueror for mobile operations. The camera is not ONVIF compliant according to Basler’s literature though it’s certain to have a list of supporting management solutions.
Performance and comparison
Now all the numbers are out of the way, what is the performance of the camera like in a real world environment? My test environment is harsh – it’s a room lit by halogen downlights to between 90 and 150 lux leading out through wall to wall glass to full sun. A scene like this is something that is common in real world retail applications.
The room is about 6 metres deep looking out onto a covered balcony a bit over 2 metres deep then out over a sunlit scene of about 500 metres, with plenty of reflections, mid-field elements and pedestrians at far distances. If there’s a weakness to the scene from the point of view of this test it’s that there’s not a lot of movement.
I look at the Basler in association with 2 other cameras, which serves to give a broader feel for the performance of this camera in relation to its peers. I’m viewing on a Geutebruck GSCView video management platform. With all the cameras also recording in real time at high resolution to a local server I can also compare night time footage in adjacent tiles on the GSCView screen, so it’s a great test for my general purpose.
The cameras I’ve included for comparison include what I consider to be the 2 market leaders that I am designating Camera 1 and Camera 2. All the cameras I’m comparing were set at default by their manufacturers or distributors.
The primary comparison is between Camera 1, Camera 2 and the Basler and the cameras are set up in a row with virtually the same field of view.
I consider room views, long views through the room, the windows and open sliding door and then check my face in the doorway with strong backlight.
I look at the camera views as they look out across the room into bright light.
Camera 1 is the pick in terms of its ability to see out of the space into the bright area and across to the far side of the scene – 500 metres-plus. Camera 1 is giving me a perfectly seamless transition from foreground to background. It’s a very complete front-to-back image given the tough view it’s taking on.
Camera 2 is doing well in the room itself but not handling backlight quite as well as the class-leading Camera 1. Camera 2 is doing well with the farther distance but this seems to come at the expense of some quality inside the room. Overall it’s good performance from Camera 2, while Camera 1 simply impresses more under backlit conditions. It’s the winner here.
Meanwhile, the Basler is actually doing even better than Camera 1 in terms of the clarity of the image close up – it seems Camera 1 is letting some of the detail go in order to handle processing demanded by the deeper parts of the image. The Basler is offering brilliant colours in the room itself, though Camera 1 is perhaps more honest in terms of colour rendition.
But at the same time the Basler is losing all the background in a bloom of backlight. You can’t see through the doorway to the outside as you can with both Camera 1 and Camera 2. However, when I test face recognition by standing in the doorway, the Basler is far superior in its ability to offer court admissible face recognition at 6 metres than the others – there’s no question about this. The Basler sacrifices something to do so – in this case, mid and long range scenes which backlight blooms away.
In any case, the Basler rules in the area of close range face recognition with backlight. It’s the best of all these cameras in that regard – no question about that at all.
Next we look at night time shots and performance is turned completely on its head. Be aware that there is no light in the foreground of these scenes. The scenes I view were taken a 1.30am – the room is in complete darkness with only city bloom lighting up the background. There are street lights, office lights, neons – the usual mix – but they are far off across a dark expanse of water.
Camera 1 is not doing well in the office with no light. You can’t really see much at all in the foreground and that’s all there is to say about it. Camera 1 needs some help with artificial light in extreme low light conditions. You can’t identify shapes in the middle distance though it is doing better but not brilliantly at long range low light.
In comparison, Camera 2 flexes considerable muscle in low light, which is where IP cameras traditionally struggle. Low light is obviously the challenge Camera 2 was built for. With no light at all in the foreground Camera 2 is still contriving to show colour – I can see the lounge is yellow, make out the pink and blue colour of folders on the coffee table. What this means is that at less than 1 lux Camera 2 has still not gone to black and white.
Camera 2 is also handling the long range low light by far the best of this group – that’s distances of 300-500 metres. I can make out amazing mid and long range detail while still seeing all the detail as well as some stronger colour in the room. So, Camera 2 wins in low light.
I turn to the Basler. It has dropped back to black and white and I judge you could see if there was a person in the foreground with this camera. While it’s a world of shadows and grain, you can still make out the lounge, chairs, window surround, desks – you can see everything in the room but the scene is simply vague. There’d be no way you’d recognise a person but you’d know someone was there. Conversely, you’re getting a good image outside at longer range. Not as good as Camera 2 but better than Camera 1.
So what does it all mean in terms of this review? In my opinion, the Basler’s performance is between the current class leaders, Camera 1 and Camera 2 yet with areas of strength all its own. This is interesting, I tell my tape recorder. There are profound gradations in the performance of the latest IP surveillance cameras. It’s not as if one camera is by far the best, nor are all cameras now the same allowing us to buy a single camera for all applications. There is a considerable difference in performance that is not evidenced by the paper specifications.
In closing, if you need a compact and affordable all-round camera that produces good images inside during the day without needing to view past sunlit windows, as well as useful images sub 1-lux at night, all while giving class-leading face recognition at close range with very strong backlight, then the Basler BIP2-1300c Day/Night is the camera for you.
