DVR features and Input/Outputs: This is a big section –
How to choose a DVR? It is the same process as choosing any other piece of equipment, a combination of feature tradeoff with cost. So, features:

Number of Inputs:

Pretty straightforward, how many camera do you want around the house and how much effort are you willing to go to wiring them all up. You should consider buying a DVR that will support two more cameras than you currently have plans for so you have room for future expansion. Consider front door, back of house, one overall inside as a minimum. A four channel DVR will meet most consumer needs and keep your costs down. Keep in mind it is significantly cheaper to buy one 8 channel than two 4 channels. Choose one with BNC connectors for good reliable long term connections and DVRs with one channel of audio is very useful for that one internal camera.

One of the more popular the Aposonic R22 has 8 inputs – see it here –

Video compression type:
This could actually be a big topic, but there are some basics you need to know.

Mjpg – or motion jpeg is an older format that is cheaper but provides lower quality and uses more hard drive space. Most of the “dirt cheap” DVRs on the market use mjpg. You should completely research any mjpg product before ordering and see if you can afford the extra $50 for a better quality compression engine. Mjpg works great for monitoring via custom applications and through the web, but it is generally incompatible with cell phone monitoring. Those codecs are generally not included and not loadable on small consumer electronics like cell phones.

H.264 – this is one of the modern video compression standards. It provides better video quality, higher frame rates, and better compression (less hard drive space and better video over the internet). H.264 is also generally an included codec in cellphones and with the right tool can work on just about any computer (mac or Win) as well as ipads and other connected devices. Note that for many DVRs, you will need to support java or active-X controls to be able to view and control the video as well. How it works is irrelevant, but look for it as a feature.

One other note is that H.264 is a variable bit rate standard. This means that if you are encoding the same boring scene with no movement, the storage required is hugely minimized. If there is a lot of movement, it automatically bumps up the bitrate to capture high quality and moving images. Perfect for security.

How does compression actually work? Yet again, another college course, but the most common algorithms include getting rid of whitespace and blackspace, setting a threshold and getting rid of things that do not change by taking a virtual picture and then only recording change from that picture and not a full new picture every frame, and a lot of math.

This is complicated stuff and requires hardware compression – meaning a dedicated chip to do this compression and NOT software running on a microcontroller to do it. The problem is that you get the best performance from a dedicated “ASIC” solution but then the compression technology is not upgradeable by changing upgrading firmware. On the other hand, a software codec that can be upgraded to newer technologies will not perform as quickly or efficiently as the hardware solution.

So, for home use, this might not be as ciritcal – buy something that is low priced, runs cool, and has good quality video. For enterprise, you might want to invest in a more expensive solution that is firmware upgradeable if not to follow compression trends, at least to be able to upgrade to current security standards.

Note that many DVRs have the capability to store the recorded video in an avi file format (still H.264 inside this “wrapper”). This is very useful in that most PCs will play this file type without any further conversion.

Frame Rate:
Generally, this refers to the total framerate of the video compression chip(s). The specs can be confusing in that many are rated at 120fps (frames per second) for one resolution or 60fps for a higher resolution. Keep in mind that this is the total frame rate and is divided between the camera inputs that an individual compression chip “services”. So, if it is a 120fps at a 320px size picture and there are 4 cameras feeding it, then each is recorded at 30 fps maximum. Pretty good. If you switch it to high quality record mode, it is now 720px size picture at 60fps total or 15fps per camera.

Caution – there are three common terms here and some manufacturers intentionally mislead you. Real time video (like TV) is 30 frames per second, but in interlaced NTSC video, one frame is actually two “fields” or images per frame. So what you really have is 60 images per second. That is slightly misleading as each picture in NTSC is interlaced and basically half of a frame. In any case, the end test is to fully load up your DVR with cameras, put them all on record, and make sure you get video that “looks real time” can be paused or put in slow motion without distortion. In the end, measurements aside, what will matter is if the video is good enough for your purposes.

This is where it gets tricky. 30fps is like TV video – real time video. The other extreme would be 1fps (or less) and is like taking a still picture once a second. Obviously you will not have any action and with someone walking by the camera you might not even catch their face. A “good” capture would be 12-15fps per camera at the highest resolution possible. Adequate would be 5fps. Obviously the picture quality is a combination of frame rate, resolution of the encoder, resolution of the camera, aperature of the camera, light sources, and a lot of other factors – we will go into some of that, but start with the highest frame rates and resolutions available and back them off to trade off hard drive space as you see what you can live with quality wise.

Since this is security, let’s take an extreme case – you want to save hard drive space so you go with a seriously low frame rate, say 1 picture every 5 second. A thief comes in, and in one frame he is standing in front of the jewelry case alone and in the next frame 5 seconds later the big gold ring is missing. Obviously he took it and you have the proof. Uh, not really, the defense lawyer will get up and spin a story about how someone ran up from behind and grabbed the ring – disappearing into the night leaving his poor client dazed. What happened in between frames? Who knows. What happened in between the thugs standing around the guy on the ground and his bruise – wouldn’t it be better to get a picture of one of them actually hitting him? Want to catch cheating in vegas – go for really fast frame rates…

If you are recording at least 5 frames per second with good video quality, you have enough information to get what you need and normally to prosecute as well. Keep in mind that it is usually just one frame that makes the difference in a case – and that frame will look the same in a 5 frames per second video as it does in a 30 frames per second – you just have to have caught it.

Video Quality:
Again, another topic that could (and is) a college course, but broken down to its simplest terms is normally measured in resolution. Unfortunately, the numbers given my manufacturers are frequently inadequate to truly represent quality because there are so many variables involved.

For instance, what the resolution is (640X480, 720X, CIF, QCIF) is merely an upper bounds of what the quality could be if every pixel were in fact perfect. In most systems, the combination of the video processing channel and the compression engine tend to smear recorded pixels around and not provide a perfect pixel by pixel image. Confusing? Yes – again your eye in worst case conditions is normally a pretty good judge.

Video Sizes:
So, PAL folks, you are on your own, I’m only going to run through NTSC here…

NTSC (North America, Taiwan, and other pockets of addicted TV viewers) is the original broadcast standard. As a side note, PAL – used in Europe, South America and most of the rest of the world is actually better at representing video and especially colors more precisely. Proponents actually make fun of NTSC by proscribing the acronym with a definition of “Never the same Color”. It turns out that while PAL might be technically superior, it suffers from the far more ruthless enemy of the government bureaucrat as there are now many many different PAL standards which are not compatible with each other. Theoretically all this goes away with the worldwide advent of ATSC – perhaps the first semblance of a one-word government but I digress. ATSC is not ready for the low end security market as encoders and decoders are not readily available at the price points necessary to compete with the low end of the market.

Back to NTSC – the standard calls for 30 frames per second at 525 lines, all of which are not displayed on screen.

But, video for security normally is specified in different terms.

CIF is 352X240 pixels or roughly one quarter of a standard NTSC screen resolution

QCIF is one quarter of that or 1/16th of a standard NTSC screen.

4CIF is full screen video or 704X480 – roughly the same as D1 in practical purposes

D1 is 720X480 and considered the first level of high definition.

The problem is that this speaks nothing of the quality of video – having CIF simply means that 352 pixels across the screen will be filled with something not that each pixel is distinct. So resolution only defines the upper-bounds of quality. A CIF image filled with more “bits” can actually produce a better image than a 4CIF image.

Bitrate is a better measure of quality – how many bits can be processed per unit of time and stored. On top of that, the compression rate determines quality as well. All compression is lossy, meaning you give up quality to be able to store video with less space. So the reproduced image will never be as good as the original. This is because compression is simply a mathematical way to throw away some data either because it is repetitive or can be guessed at when reproducing the video. Lots of math involved.

So, here are some rules of thumb –

To view a full screen 4CIF video, typically you need 1.5Mb/s of bitrate using H.264, 3Mb/s using Mpeg4, and 4 or 5Mb/s using MPEG2

Since this relates directly to recording space, it is useful to use the same rules of thumb to determine hard drive space. For H.264, at these acceptable bitrates, it will require approximately 1GB/hour. Keep in mind that with motion detection this can be stretched out indefinitely.

One quick note is that with modern encoders, image resolution, bit rate, display frame rate, encoded (recorded) frame rate, and motion detection can all be adjusted. The better DVRs allow you to adjust each of these independently by camera channel to perfectly adjust your system for each camera use.

Video Output:
Generally, there are two video outputs, a VGA connector and a BNC/composite connector. This allows you to connect a cheap VGA monitor (you probably have an old one in the garage), as well as plug the video output composite into a whole-home distribution system – this is what you would plug into a BOCS channel.

The composite output is frequently adjustable to standard NTSC or a 720 line output. The higher resolution is good for a locally connected monitor but not much else. I would not recommend paying more for a DVR with this feature.

Web Connections:
Its great to have a way to record events and play them back later for identification of intruders and the like, but making that entire system available both home-wide and world-wide makes the system infinitely more valuable. You can easily monitor all home activities form any computer and just about any cell phone. Its not worth going through the tradeoffs of DVRs with and without this capability as it is almost impossible to find one without it. But there are differences in features and functions.

By the way, this stuff is all related – If you are wanting to watch a CIF video stream of 30fps NTSC video remotely, that would take over 35Mb/s if it were not compressed. So, the better the compression you have the better quality video you can transmit over a smaller “pipe”.

Remote Viewing – proprietary software:
Most devices come with some kind of proprietary software to connect a PC to the DVR remotely (via Ethernet) – why? If we go back a few years, a proprietary piece of software could be said to provide more security (harder for others to tap in), more stability (ie can get viruses and have connectin issues based on other settings like proxy servers), and a known platform (every browser interpreted html a little differently and handled add-ons differently. Plus, keep in mind that these DVRs are running off microcontrollers and not full up windows PCs … A proprietary piece of software can have specific local buttons that command your local PC to do something or grab something from the DVR, a web browser merely asks the remote device to do something, further bogging down the little processor. Plus, this gives a solid thing to fall back on when you call in and complain that you can’t get your browser to connect to your DVR…

See a real time demo of remote viewing – log in, control and view live security video to get a good idea of what it looks like over the web –

Remote Viewing – web:
Generally, DVRs ask you to install some kind of active control to allow fast refresh and remote control of the device. The better ones use javascript to do this without the extra add-ons. IT is most definitely worth looking for reviews of the different devices available to see how much luck different folks have had with web viewing. This is perhaps the one thing that causes a lot of DVR manufacturer’s trouble, yet perhaps one of the most valuable things.

Email/FTP capability:
This is becoming more rare since real time access and long storage times are now the norm, but this feature, if present, would send you an email with a picture of an event and/or upload it to an ftp site. This is actually one of the first DVR/real time monitoring options. I have a stardot camera/webserver that uses a connected PC as the DVR (theoretically saving money by double using the connected PC although it reduces system reliability since now recording and events require a good Ethernet connection and a windows PC that has not yet crashed that day)… On that particular device, I can set up an FTP server and it will record all events or even just a time lapse string of pictures to that FTP site. Kind of a poor man’s DVR. The one thing I really liked is being able to set a specific motion event at my front door to email me a picture of who is standing there. I then have a visual record of everyone that stepped on my welcome mat. Today? I would definitely not pay more for the option.

Alarm Inputs/Outputs:
Nearly every DVR has the ability to interface with the outside world directly. Considering that it is only piece of an overall security system, this is only logical. So, let’s consider two scenarios to make this interaction more clear.

Consider that you also have a perimeter home alarm system. If you are away, and it activates then you know something momentous occurred. Perimeter systems tend to be very reliable (if installed correctly), and give a good indication of an intrusion event. Once that goes off, what you really want is every camera in every part of your home to instantly begin full resolution recording. Even if a camera has motion detection activated to record when something moves in the shot, having every camera on full record during an intrusion event is key. Who knows what one of them might catch: a license plate number of a very slow moving car, the face of an intruder as they run by a camera (if you run fast enough, the camera might not start recording until you are almost out of range) – note that some systems have an automatic backup so it actually senses motion and backs up 5 seconds to start recording but this is normally a higher end option.
You have a camera inside your locked “valuables closet” – gun safe, or whatever. There is NEVER any movement of any kind in there unless someone enters. You can also use your camera system as a motion detector to actually set off your perimeter alarm, or turn on a light, or even call you any time it sees movement in a particular area.
The DVR interacts with the world through the alarm in/out connector. Generally, each channel has its own alarm input – this can normally be set to begin continuous recording on just that channel or on all channels. For instance, you might want to cover your driveway entry gate that is surrounded by a lot of trees. If you rely on motion detection alone, the camera might be recording nearly all the time due to wind blowing the trees around. So, you can hook the alarm input for that channel up to a gate open sensor and record only when the gate actually opens and lets someone through.

Some devices also have a serial or rs485 connector allowing better integration with a home automation system.

Storage Capacity/Upgradeability:
This is simple, but do not overlook how the system firmware is set up. What you want is a system that stores ONLY recorded audio/video on the drive and not its own operating software. That way, upgrading means simply plugging in a bigger drive. Also, if your hard drive crashes (which it will eventually – they are the weakest link in the chain), you just go out and put a new one in. If the device stores its own operating software on the drive and the drive crashes it is dead and not recoverable. It is a temptation for companies to do this, and some of the lower end models are guilty. Just watch out.

Generally, drives accepted are SATA (look for this since older IDE drives are getting harder and harder to find) and will accept up to 1TB. Some even have two HDD connectors.

I definitely recommend allowing the DVR to format the drive itself – keep in mind that this is not a full windows PC and sometimes the format it uses is slightly different that a real windows PC. Let it handle the drive.

The R22 – again, the main example here, comes without a hard drive – buy one yourself to save money. Note that you can choose to add one on this site if you so choose.

Motion Detection:
Flexibility is the name of the game here. The more you have, the more happy you will be and the less frustrated with false activations. Keep in mind that you have to apply some intelligence here to make sure it records when you need it to but gives as few false events as possible. So, figure out what your DVR has to offer in the way of features, set up all your cameras and pre-plan how and when you want your system to respond before you dive in. As a side note, it is critical that you lock your cameras down well as any movement will change the whole motion detection setup.

There are a variety of different systems available and each has a different way of setting up motion detection, but lets go through a couple of examples:

Sensitivity – you definitely need to be able to set the level of motion that the device will respond to. Keep in mind that this is directly related to resolution so the better the picture quality the more flexibility you will have in setting sensitivity. Ideally, you want the ability to set different sensitivities on different parts of the screen, but so far I’ve not seen one that will do that. A range of 1 to 10 would be nice, although most of the lower end units are a range of 1-4. Note that there is much more experimentation than science here in that you should start with the default and adjust this to minimize false alarms but ALWAYS catch movements you are looking for. If you cannot get it adjusted correctly then turn off motion detection and use the alarm input for that channel and some kind of physical sensor (light beam, door contact, IR motion detector) to trigger recording on that channel.
Motion area – basically there are two categories – and two selections. “Block” area or “grid” and “include” or “Exclude” – depending on what brand/model you purchase.
Block – basically, the system lets you draw a box on the screen and it will only pay attention to motion within that box – simple but life is rarely that simple as you might want to draw a box around your entertainment center but exclude your TV so the camera does not record all the time when it is on
Grid – by far the best depending on the size of the grid. This normally involves clicking on the various small squares on screen where you want the device to pay attention to motion. For instance you might highlight a door on one side of the screen and the refrigerator on the other.
Include – this is the default and what 90% of DVRs offer – include this area for motion detection either by designating a block or various squares in a grid
Exclude – this is more rare but exceptional when you find it and you get it for free in a grid approach – pay attention to the whole screen except this area where the wind chime is (for instance).
Note that motion detection normally has a variety of other options including record length on a motion event, how far to back up to start recording prior to each motion event etc.

Be sure to test all your settings in day, night, and low light conditions to make sure it all works according to your plan. If you have trouble, better cameras or more IR lighting at night might be the solution to your problem rather than changing motion detection settings.

Event recording and search:
Huge huge huge. If you ever do need to go back and find an event, this combination of features will save you hours. The lower end models tend to have a very simplistic approach in that they treat the system more like a videotape. It records when motion is detected on a camera (or you set a record schedule, or all the time) and finding an event involves setting a start time and watching the video using normal play/ff/rew buttons until you see the event you want. Most devices also have an event list, but in the lower end models these two things are not connected. You might have to go to the event list, write down a few times of interest when it says something happened and then go over to the playback menu and find those times to see what it was.

Better DVRs integrate all this into one function giving you the ability to browse and event list (some even have thumbnail previews of the event) and click on them to play the video from that event. Personally, I do like the option of integrating all events from all cameras into one single timeline so you can track specific activities around your property.

Backup and Save:
Look for devices with the ability to save quickly to either a CD/DVD, onto a USB stick, or to a connected PC. If you have an event you will be wanting to hand the physical evidence of that event to the police as quickly as possible. Ideally, saved video has a time/date overlay to further corroborate your evidence. Make sure you set your clock on the DVR or the evidence might not be admissible later.

Personally, I would not pay for the integrated cd writer as USB sticks and email suffice for this kind of backup.

For this feature it is all about simplicity – make sure you can quickly and accurately grab a section of video and send it to someone. Every device and piece of software is different – just make sure it does not confuse you.

Health Monitoring:

Some devices have an ability to send out a regular ping that they are alive so you can instantly know if your DVR system is down and needs service. Lots of variations exist, but a common ability is to interface with an external system that send you an alert if it looses DVR connectivity.

Whole-Home Connectivity:
I want to see my security cameras anywhere in my home

The normal answer to being able to see a security “feed” is either to go watch it on your computer OR to put in a modulator that creates a new TV channel in your home with the security feed. So you might go to channel 70 to see the four cameras in your home. Ah – but there is a problem. Unfortunately, some relatively new technical hurdles have rendered the process of installing filters and modulators to distribute security channels almost impossible in a Cable-TV home or office. There simply is no longer available bandwidth to add new channels. If you set up a system to broadcast security channels on channel 70, you are likely to kill MANY valuable stations like ESPN, CNN, or others since cable companies have gone digital and use almost every available space.

Everyone in the business is familiar with using video modulators: pick a channel, run a new wire to the demarcation point, insert a filter and a backwards splitter (combiner) and you have just added a new channel to the home for the security system. Even if you could still do that, say you found an unused channel that only kills the home shopping network – how many return calls do you expect to get in the next couple years as cable companies shift their lineups around? And what if you need cable TV in the location where you put the Security DVR? Yes, that means you have to run a new home-run back to the demarc point since you can’t run signals both up and down the same coax wire… or can you?

A whole new generation of distribution is now available. The BOCS system does not interfere with cable-tv programs, allows you to receive cable-tv in the same location you are sending security video from, and only takes minutes to install.

Two things make this possible:

1) BOCS actually uses the FM band, which cable companies avoid like the plague since it tends to leak into their lines and cause interference. This is only possible because of patented technology that allows the filter and modulator to work together to clean up the bad and properly insert the new signal. In this case, BOCS uses channels 14, 98, and 96 for redistribution (see the picture below).

2) A new cable tv splitter called a “Supercombiner” allows cable signals to be used at the same location where you are creating the new security channels. The Supercombiner also takes care of filtering, signal conditioning, level balancing, and redistributing the new security channels to every other TV on the system.

Each BOCS system comes complete with three distribution channels – allowing you to dedicate one for the baby, one for the front door, and one for cycling or a grid – all on top of the cable TV already present. If it is a home installation, any of the inputs could also be populated with a standard DVR or DVD player to integrate entertainment with security.

So, you want your security system on all your TVs so you can see who is at the front door? BOCS is the only way to do it.

Things to check before you buy (or very quickly thereafter in case you decide to return the system):

Video quality out the composite out (and/or VGA out) is goo in dark, light, and low light conditions
You can play back video in slow motion without distortion
Remote video is high quality on both the proprietary software and via the web
Recording and display speeds (frame rates etc) are as advertised
Do a test recording and make sure the storage space used is as advertised
Do a full 24 hour test and make sure there are no reboots or glitches
Blow one camera up to full screen and make sure the video is still clear
Freeze a frame and make sure the video is still clear (not double or blurry)
After running for 24 hours it is not overly hot to the touch or overheating
After running for 24 hours it is not unduly noisy
Test burning a disc and/or saving to a USB stick and/or to a local PC some portion of video
Make sure the motion detection setting meet your needs, (block vs grid), that it functions reliably, and that it does not give undue false alarms
Verify the DVR can remain connected to a remote PC for long periods of time
Verify that you can view the video from outside your home (through your firewall)
Be sure that remote playback is real time and that the network connection does not become bogged down when a lot of motion takes place.
Connect a camera to every input and make sure it all still functions as expected with a full load – inferior products are usually discovered under full load.
And give the tech support line a call to make sure there is someone you can quickly get hold of that speaks English and has good technical skills