Choosing HD Cameras: Part One
Posted by Security Electronics and Networks | @Articles CCTV | March 21, 2013, 7:00am AEDT
IT would be nice to think that HD cameras were all identical – that they conformed to a cast iron specification and we could depend on their performance to be true in all applications. But that’s simply not the case.
There’s plenty to think about when it comes to choosing the best HD cameras and in this issue we spoke to a number of different manufacturers and commentators about the sorts of things integrators, installers and end users should be looking for in their HD camera solutions.
First cab off the rank is Vlado Damjanovski of C.R. Kennedy and ViDi Labs. Damjanovski’s credentials go pretty much without saying and he has no doubt the area of importance when it comes to HD camera selection – it’s image quality.
“Image quality is the key and this depends firstly and foremost on the sensor used,” Damjanovski says. “I refer here to resolution, dynamic range, noise and speed of read-out. There are only a handful HD sensor manufacturers in the world, and as such there wouldn’t be a big difference when the raw image coming from an HD image sensor is concerned.
“Instead the big difference may come from the intelligent processing of the raw video and its interface to the external world. And, if the signal is compressed, how good the compression is, what level of compression is used and what default settings are used.
“Another very important feature is how many concurrent encoders can be used (if more than one) for the purposes of dual or triple streaming, one for recording, display and another perhaps for remote streaming with lower bandwidth. Also, for larger projects how multicasting is handled might be the next most important thing.
“Of course, the standard installer demands are still relevant,” Damjanovski says. “These include ease of installation (preview), PoE, mounting, etc. This depends on the camera manufacturer, not the sensor manufacturer. Additional factors of course are the lens quality. We all know that if a lens is of an inferior quality (compared to the sensor resolution), it is impossible to make the HD video any better.”
Given the multiplicity of qualities we’ve touched on here, what specific features should installers look for when choosing a quality HD camera?
“In CCTV, most often it is the low light performance that might be decisive for a lot of projects,” explains Damjanovski. “But, as many would appreciate, the pixel size in most HD sensors is even smaller than the analog resolution sensor, hence it will be a tough ask to get an HD camera to outperform a good SD analog in low light.
“So my advice to installers is to be aware of where the HD camera needs to be installed first. HD cameras will produce better picture, there is no doubt, but if low light performance is the most important requirement, installers should consider additional illumination (normal or IR) to produce better signal of HD at low light.
“Next thing to consider is having power-over-Ethernet (PoE) to simplify cabling, but also the HD camera sensor capability. Some manufacturers, like Dallmeier, offers cameras with multi-formats (SD, 720, 1080, 3 MP, 4MP or 5 MP) switchable through the software.
“Certainly, the most important of all might be the image quality, so before you decide to invest in HD cameras, get a sample footage and check the video quality for yourself. Even better, get various cameras and test them on your site at the light levels typical for that site.”
So 720p or 1080p, which is best? Does improved compression mean there’s no longer a need for 720p?
“Theoretically, more resolution is always better, but, some will argue that a 720p sensor may have larger pixels, hence better low light performance, and makes smaller stream, as it has half the amount of pixels (1280 x 720) then the 1080 (1920 x 1080),” Damjanovksi says.
“But be aware that some cameras, like the Dallmeier DF4910 range, offer a choice of resolution in the same camera, because the sensor they use in this camera is a 5MP sensor and it allows you to choose whatever suits the application.
“Even better, such a camera uses pixel-binning when in 720p mode to produce better low light performance, so it all depends on the camera design. The most obvious benefit of such multi-format sensor cameras is the angle of view switching without changing the lens – when a sensor mode is changed the angle of view changes.”
There are lingering questions over the weaknesses of HD cameras – for instance, how are they in low light compared to the best analogue cameras? How are they in very strong backlight? What about white noise in low light scenes impacting on storage volumes?
“The low light performance is very much dependant on the sensor type and pixel size,” Damjanovski says. “A typical HD sensor pixel size would be around 2 microns (4 square microns), and on some sensors even smaller. As a comparison, analog sensors typical pixel size is 6 microns (36 square microns).
“In surface area this is 9 times bigger in the SD cameras compared to the HD. This is the most important factor defining low light performance under the same conditions (temperature and camera processing) and applies to majority of cameras in our industry today. But, the technology development doesn’t stop, and we will soon see some new inventions, like back-illuminated CMOS sensor.
“These are used on the iPhones, and other smart phones and they will increase low light performance without increasing the pixel size. The dynamic range, back-light processing is directly dependant on the pixels size too, but here too, there are some developments where double exposure is used to simulate high dynamic range.
“As for the storage of noisy signals, yes, it is true that noise affects the storage requirements. But, a good compression means not only good encoder in the HD camera, but also noise reduction before compressing, thus achieving reasonably low streams with very high picture quality.”
According to Damjanovski, an example from Dallmeier again, is the DF4910 that he says can produce an excellent true 1080p HD signal with as low as 4 Mb/s using H.264 (at 25 fps).
“With this camera model, we can switch to a 5MP signal using the same 4 Mb/s H.264 encoding (with slightly reduced frame rate of 10 fps). As I said earlier, if for a particular job the very low light performance is the most important thing, then, perhaps good quality analogue should still be considered.”
How important is lens selection with HD cameras? Is digital or optical zoom preferable, or both?
“The lens is most important thing after sensor quality,” says Damjanovski. “Whether I prefer digital or optical zoom, it all depends in what context. With the Dallmeier’s Panomera design for example, digital zooming looks quite different. So, even though I would usually argue that optical zoom offers better magnification of details.
“In the case of Panomera it is different because the Panomera digital zooming far outperforms optical zooming simply because of the different approach in designing this multi-sensor multi-focal camera. With Panomera’s digital zooming you can see the whole picture without losing the details where you have not zoomed in, and this is done for both live and playback video.
“With a normal optical zoom once you point the camera to a specific area that is it – you can’t see anything else,” Damjanovski says.
“And of course, if you have recorded such a zoomed in video signal and recorded it, it is not possible to retrieve what the zoomed camera has not seen. Certainly if you know where and when the incident will occur, and zoom in that direction, the optical zooming will be more beneficial.”
“In the past, HD cameras would not be recommended for low light applications. However, with the continuous improvements in technology, HD cameras are now available for these conditions”
What about the new H.265 standard? Is it likely to reduce the bandwidth demands of MP video streams and/or lead to higher resolution cameras being adopted – bigger 5MP cameras? Or do you argue for many installations, particularly indoors, 1080p is going to be more than enough?
“The upcoming H.265 will further improve network capacity and design, but I don’t think it will not encourage (nor stop) any development on resolution,” Damjanovski says. “It is only going to be as useful as the H.265 encoder chips get introduced into our CCTV products. This might be very soon, or it may take another 5 years perhaps.
“Yes, it is said that it will offer at least 50 per cent bandwidth saving over H.264 for the same visual quality, and this would certainly be welcomed in CCTV. But this bandwidth saving will be paid for by the higher processing power that will be required from the PC machines decoding such compression in the software. So, once again, with H.265, we will come to yet another demand for new and faster computers, better network switchers and better network designs.
“The fact is, the “good old” MPEG-2 is still available and in many cases like your cable television) still used, without stopping the HD technology develop. In my opinion the biggest benefit our industry could have is the development of new sensor technologies with better low light performances (one of which is back-illuminated sensors) higher dynamic range and smaller sensor size, with the very important – lens quality improvements.”
Meanwhile, Pacific Communications Mark Shannon says that from the end-user’s perspective, the key features of a quality HD camera, as a minimum, are brilliant images, the ability to show near real life images (picture and colour rendition accuracy), sharp focus, low light capabilities including Day/Night operation, good wide dynamic range, and low bandwidth requirement.
Shannon says that installers should also look for cameras that provide ease of set up, configuration, and installation. These features are provided by cameras with attributes including a local monitor output for lens setup, Auto Back Focus, convenient location of Ethernet connection at the back of the camera and a variety of mounting options and accessories.
And when it comes to 720p or 1080p Shannon says it’s all about the application and the budget of the customer.
“Camera selection is still as balanced a consideration today as it has ever been with analogue,” he explains. “However, at present the network bandwidth and storage are additional considerations. System design and camera application is a skill and not a one size fits all and this will never change.”
Shannon says that in recent years, HD cameras have come a long way in overcoming their inadequacies such as performance degradation in low light and strong backlight.
“In saying this, all cameras need lighting to provide good images and accurate rendition, whether it be overt (white light) or covert (IR light) depends on the applications,” he explains. “In the past, HD cameras would not be recommended for low light applications. However, with the continuous improvements in technology, HD cameras are now available for these conditions.
“Lens selection is also critical with HD cameras. In most applications, optical zoom is generally used because it gives an accurate magnification of the image (does not introduce artefacts).”
According to Shannon, H.265 should be seen in the light of the ongoing improvements taking place in the video surveillance market.
“Ten years ago, no one imagined that HD will be in the IP video surveillance market,” he explains. “Today, HD cameras are the norm in the IP video surveillance market. In the future, the advancement of technology will go hand in hand with the improvement of IP surveillance cameras; therefore, there will be a migration to higher resolution with better compression and storage on offer.”
Over at Axis Communications’ Chris Tangsilsat says key features of a quality HD camera for end users include things such as pixel count, frame rate and colour representation but he points out they are only the tip of the iceberg when it comes to what end users look for in a quality HDTV IP camera.
“We also need to consider versatility as the objective of having surveillance for most end users is to obtain ‘usable’ video at all times regardless of lighting and other environmental conditions,” he explains.
“For example a HDTV camera located in a building lobby may provide usable quality images for the majority of the day but is subjected to intense backlight in the late afternoon and inadequate lighting after hours.”
When it comes to installers, Tangsilsat says that apart from quality images he thinks the majority of systems integrators out there would place an equally high value on the reliability and ease of installation of these units.
“Whether the installer spends 20 minutes or 1 hour installing and configuring a camera will make a sizable impact on labour cost, especially when you consider that some installations could potentially consist of hundreds or thousands of cameras either at a single site or spread across multiple locations,” Tangsilsat says.
“Post installation, the reliability of the product will determine how much time the installer keeps investing resources in returning to the same site for maintenance and repair, or is able to move on and tackle the next project.”
Interestingly, Tangsilsat says he doesn’t believe there is what could be considered to be a better resolution format – either 720p or 1080p when it comes to HD.
“It’s more important to identify the customer’s requirement in terms of level of detail vs. coverage area,” he says. “For example identifying customers entering a store may require a 720p HDTV camera focused on an area that’s 2 meters wide, while keeping track of whether a meeting room is occupied may only require an SVGA (800×600) camera covering an area 5 meters wide.
“Alternatively, capturing license plates of trucks moving into a loading dock could warrant the use of a 1080p HDTV camera. There will always be a need for a large variation in resolutions. Additionally, implementing an effective surveillance solution could be as simple as using a pair of 720p HDTV cameras in a small retail environment.
“In such a case, a camera is placed at the entry point using a high focal length lens to capture of the faces of anyone entering in detail, while another camera of the same type would use a wide angle lens to provide a general overview of the store. It’s all about strategic placement and identifying where the customer requires that extra level of detail.”
And what are the weaknesses of HD cameras in low light and in backlight?
“If you were to ask me this same question 5 years ago I would say that in extreme low light conditions there are some quality analogue cameras which will outperform the IP offerings,” Tangsilasat says.
“However, due to the continued development of IP surveillance technology, I can say that the latest generation of HDTV IP cameras offers superior image sensors and encoding chipsets and will have better low light performance compared to their current model analogue counterparts.
“Low light performance through features such as Axis’ Light Finder, however, have not been the only development as in order to increase the level of versatility, features such as Wide Dynamic Capture are also available in order to cope with strong backlight scenarios. All of this leads to a usable quality image containing less artefacts/white noise resulting in lower storage requirements.”
When it comes to optical or digital lens selection, Tangsilsat says he thinks lens selection is highly critical.
“A good quality lens will truly demonstrate what the camera is capable or, while a poor quality or incompatible lens will degrade and misrepresent the quality of the camera,” he explains.
“Installers will notice that each Axis IP camera will always include a lens as part of the same product code. This is because all lenses which are included with the cameras along with optional types listed on our website have undergone testing within our R&D to confirm they are compatible with the intended model and provide the optimum image quality.”
According to Tangsilsat, optical vs. digital, this has been a topic of discussion ever since megapixel cameras were first introduced and he says it will continue to be so as higher megapixel cameras evolve alongside higher optical zoom types.
“There really isn’t a right or wrong here – in most cases the choice comes down to the end user’s requirements and resources in place,” he explains.
“Digital zoom will allow the user to retain a wide general overview of a scene with limited ability to zoom into certain areas, while optical zoom sacrifices the general overview in place of superior detail on areas of interest.
“Again, this comes down to the end user requirement as one customer may just want a static high resolution, wide angle camera to record with the ability to digitally zoom after an incident. Another customer may have a control room with operators actively manning the PTZ cameras with optical zoom.”
Tangsilsat says the new H.265 standard is definitely interesting news and he believes that once adopted it will ensure a reduction in both bandwidth and storage requirements.
“However, whether we will see a push within the market for higher resolution cameras as a direct result remains to be seen as there are other factors to consider when it comes to usable quality images in surveillance.
“These factors include low light performance. Due to the way current image sensors convert light into images, a 720p HDTV camera will outperform a 5 Megapixel or higher resolution model in low light scenarios regardless of post capture compression method.
“Also important is frame rate. IP cameras up to 1080p HDTV resolution will easily achieve a full 25/30 frames per second. At 3 Megapixel this typically drops to around 21 fps, and at 5 Megapixel 12 fps. Will moving into higher resolution ranges regardless of compression method start to result in unusable frame rates? That’s something to think about.”