Oh yeah.
That’s a very common feature.
I’m going to say that almost all digital video systems nowadays support that, some better than others.
Actually, you can accomplish the same thing with software on the DVR. Even some of the fancier VCRs can do that, all without having to wire 14 different motion sensors into the I/O boards on the DVRs or VCRs.
I don’t know about this - one of the other uses I believe for a CCTV system is to have a record in case of a fire. If the motion sensor will pick up wisps of smoke from an electrical fire starting behind the drywall, that seems reasonable. I’d think that there is some benefit for having the fire reconstruction available, in case that happens.
(Of course, this more likely to be useful for a system that features off-site image/record storage.)
The imagery often IS good enough.
Frequently, it is not used to identify the criminal. The police have already caught him, and he still had in his possession the stolen property. All the video imagery has to do is make it easy to convince a jury that they are seeing him commit the crime. Or often, convince the accused (or his lawyer) that it isn’t worth fighting this case, since the prosecution has this video, and they should try to plead guilty to a deal.
So the video imagery is useful as prevention, and later in prosecution of a criminal. It doesn’t have to be very good to be effective at that.
In the UK, it often isn’t. I have a relative who works in the courts as a porter, they were surprised at how easily evidence could be poo-pooed by solicitors. CCTV footage seems to be a complete waste of time, they’ve never heard of anyone being able to use it to help prosecute anyone.
Not even the cameras the police have mounted on the station in the town itself would suffice. I found that surprising given that I live in Northern Ireland and the bigger stations really are fortresses.
So I take it that, despite what I’ve seen on television, it’s not possible to zoom in on security camera footage and read, for example, the ID tag being worn by someone in the room?
“M’lord, the amorphous fuzzy blob on the camera could be any amorphous fuzzy blob and not the amorphous fuzzy blob I am defending here in court today.”
That’s correct, with your average CCTV footage… frequently using more than about 4X zoom on the image results not in you acquiring more insight as to what was going on in the shot, it just gives you… bigger blocky pixels.
I just downloaded some footage from a customer with rather good surveillance coverage, and here’s what I’ve got:
A JPEG, 320x240 resolution, and 22.3 KB.
My computer is running in 1280x1024 mode.
I open up the image in my vintage copy of MS Photo Editor.
The zoom is set at 100%, so the picture is literally using 320x240 of my pixels.
Looks pretty good. Looks better than TV, or so it seems.
Zoom now at 200%. Still looks like a “real” scene, but it’s got a little bit of pixelization, especially where the overhead lights meet with, for instance, the corner of a doorway in the distance.
At 400% now. Still a ‘decent’ shot, but the jaggies are showing up in serious numbers now.
At 800%, I’ve got compression artifacts showing up all over the place.
At 1600%, the scene is PAINFULLY over-zoomed, and the whole thing is seeming silly.
Okay, meant to do more of this post, but Lady Slant needs some PC help…
Hi. This is weird. 
The other thing not to underestimate, is the placebo effect. Not only is it a deterrent to theives (or whomever you’re trying to catch) - security cameras make people they are “protecting” feel good (safer, more in control, etc), regardless of their actual quality. Mr Slant makes a very good point about reviewing the footage from the new cam before considering the job done, but i wonder how often that really happens?
It’s kinda like buying insurance, without reading the fine print to know that you’re actually not covered if x happens, even tho you assume you are. Actually, it is exactly like that.
Wishful thinking is alive and well - we all do it, I think…
Mr. Slant is correct about this not being available on average systems, but the technology is available and is used in high security applications like airports and police stations.
The key point is that to be able to zoom in on a recording, you have to have recorded & saved it at high resolution. That implies high resolution cameras and no- or low-loss recording. Given that modern video compression can reduce recording size by 10x or more, a decision to preserve the ability to zoom way in is a decision to spend $10x as much on storage, or even more.
This is where adding some intelligence to what you store really pays off. If you’re watching a corridor, don’t store anything when it’s empty. Watch continuously, but don’t store what you see. But when you detect a change from the empty scene, store everything from 5 seconds ago to 10 seconds after it goes back to normal at max resolution.
Naturally, this particular tactic works best on relatively static scenes. But I have seen demos of intelligent surveillance systems that watch intersections and store only the weird traffic, not the routine comings & goings. At setup, you let it watch the intersection for a day to train it to what normal looks like. After that, you end up with a high-def recording for the next 24-hour day which only contains a few minutes of illegal U-turns, a fender-bender, and that pedestrain who jay-walked diagonally.
I’m not real sure all this is a good idea. But it is cool.
LSLGuy more or less issued a good response, but here’s the equation:
Storage Drive / FPS [frames per second] / Image Size / 86400 = Days of storage
It works out to 1 GB per day if the system is running 12 KB images at 1 FPS.
Older (late 90s-early ‘00s) systems might have run at that resolution, depending on customer base.
Nowadays even the systems that emphasize long-term retention at the expense of image quality are running closer to 18 KB shots, depending on scene.
So you’re looking at 1 to 1.5 GB per day just to get 1 FPS and above-average but still mediocre shots.
Now that’s 1 FPS… total. Not 1 FPS per camera, so if I’ve got 32 cameras, that 1.5 GB per day if giving a given cam 2 shots per minute, which sucks.
Let’s do 4 FPS per camera, which is the absolute least you want on shots with action in them[1]. Okay, that makes it 128 frames per second to worry about.
Our disk usage is now 192 GB and we just recorded… 24 hours.
90% of the DVRs on the market use internal IDE or SATA drives, and have space for no more than 4 of those mounted internally.
We’ve now got 4 terabytes at our disposal for images; I’m pretending our DVR is booting from ROM and has no OS partition, and stores all settings in a ROM, like a PC BIOS gets stored.
4 terabytes divided by 192 GB gives me 20.83 days of storage.
How long does your facility need storage for, to defend lawsuits, false credit card chargebacks, and fraud that gets discovered later?
Typically places that are concerned about employee theft, robbery and shoplifting choose to record about 30 days worth of shots. [This means most retail facilities.]
Banks, due to various federal regulations, will want to have 90 days as a dead minimum [2] and usually shoot for upwards of 180 days.
Let’s thin these images by deleting all of the non-motion shots… generally 80% of shots recorded are totally still.
Using current (rather mediocre) image resolution and 4 FPS, just to get to 104 days, you’ll need a system with gobs of storage.
If you want to be able to do that CSI crap where you zoom in on a subject’s leather shoe to look up the skirt of the lady next to him and identify the brand of her panties or the license plate of a car 200 feet away in the dark, you’re ramping up your image size.
I figure you’re looking at 1/2 MB to 1 MB shots for that.
Going with 500 KB apiece for those shots, I’m looking at 5.5 terabytes per day before motion thinning, and 1.1 terabytes per day once we’ve cut out still images.
Thus, the answer as to why you don’t see movie-style surveillance in most real-life situations is that 1.1 terabytes per day is hella’ expensive, and 15-30KB shots will usually cut the mustard.
[1] If you’re trying to bag shoplifters, 4 FPS will generally get the money shot, but it gets confusing if you’re watching an altercation.
[2] The Feds give customers 30 days from bank statement day to complain about fraud, while a statement cycle is 30 days. That gets you to 60. We have to alot a few days for the bank to print statements and the post office to deliver it (just got to 75), after which we’ll need to give the bank a few days to investigate the situation and pull shots that may be dispositive (90). Some bankers are overworked, disorganized or lazy, or require police assistance to figure out what shot is what (120).