MaxIcon

If you want to go with a mix of analog and IP, you should look at hybrid systems from the start, so you don't have 2 different systems to monitor. I'm using an NV5000 on a dedicated box with good results, though you might would want to go with one of their higher-end PCIe cards to get to 16 channels while keeping a good frame rate. The software is flexible and stable for me, though the cards can be finicky about what type of hardware they work with. As far as D1 vs CIF, I'd never even consider recording CIF. There's not much point in it, IMO, as the resolution is terrible, but I'm also switching over to IP due to the much better resolution. If you want to be able to ID people and events, more resolution is always better than less, and less compression is always better than more. I have a couple of Vivotek IP8332 cams that are in your price range, and the daytime performance is miles above analog, but they do suffer at night, even with the built-in IR. They're still better than the analog cams I replaced with them, and I'm looking at improving the ambient lighting instead of spending on more sensitive cameras.

It occurred to me I left out one other important step - setting up the cameras themselves, from the IP address. Some cameras have an app for it, others use the web address, and some require you to use IE while others work OK with Chrome or Firefox. If there's not a vendor utility for setup, once you have the camera IP address, plug it into your browser, and it will usually come up with a web page with the camera view and settings. Every camera's different, though, and some are better than others. Finally, some software gives better quality video if you match the frame rate settings on both the camera and the recorder. Others override the camera setting, but some give jerky video if the camera's set for 10 fps and the recorder's set for 7.5 fps (as an example). Definitely more troublesome than analog, but once you get the quirks out of the way, it should be stable. I keep plenty of notes!

I don't know the Nano or Acti, but there are 2 steps for connecting IP cams to most recorders: - Find out the IP address of the camera after connecting it to the network. As Seanhawg says, some use DHCP and get an IP address automatically from your router, while others come with a fixed IP address and need to be changed to a unique address that fits your network, usually with a utility program from the vendor. If they use DHCP, you'll need to find out what IP address the system automatically assigned them. The vendor program may search them out and show it to you, your router may show it to you, or Windows 7 is fairly good at spotting some IP cams. This can be a tricky area, and it totally depends on the camera's programming. People with more Acti experience can tell you how it will work. - Once you know the IP address for each camera, you generally need to tell the recorder what it is, and assign it to a channel. Some NVRs will auto-detect cameras, but that can be unreliable, depending on the NVR and camera. You could have 20 IP cams on a network and only want to record 5 of them on the Nano, so you'd have to tell it which 5. Once the DVR/NVR has connected to the cams, you set it up much like any DVR for frame rate, recording parameters, and whatever else it supports. Once you've done it a few times, it'll be much clearer, but it's good to start learning basic networking if you'll be working with IP cams in the future. They're a good bit more complicated than analog cams that way.

Most of these have board cam lenses (usually M12 thread), but there are always exceptions in the low-end cam market, and you never know until you try. In general, the cheapest way to lens a camera like this is to use an inexpensive board cam lens, so that's what they do. The good news is you can buy a single lens, or sets of 3 or more lenses, cheaply from ebay, amazon, or dealextreme, if you want to test some different ranges. I'm guessing the quality isn't great, but that comes with the price range.

With a tight budget for IP cams, be sure to look at both initial license costs and whether licenses have to be renewed yearly.

Are you saying that Camera 1 used to see IR OK and sometimes doesn't, or has it never seen IR the way that Camera 2 does? If it used to see IR, it's likely the IR cut filter is stuck, as Draken says, assuming it has one. If it never did, it may have a fixed IR filter and will never be able to see IR unless the filter is removed.

You should avoid GVI, as they've gone out of business.

Your AVC798BD supports D1 recording, but only at 7.5 fps. I always choose a lower frame rate and higher resolution if quality matters at all. You also want to set it for the best quality to see how it does. If your camera live view looks much better than your recordings, then yes, D1 recording will give you better quality. If the live view doesn't look good, the camera may be the problem. Better quality takes up more space and reduces the number of hours you can record, as does higher frame rates. If this is an h.264 only recorder, you have to trade off frame rate, resolution, and compression quality against the number of hours.

Both of those look like they have preamps, based on the specs and the pics. They just don't call it out on one of them. AFAIK, all CCTV mics have preamps, as DVRs have line-level inputs. The round bit at the end is the actual mic, and the lumpy things in the heat-shrink are the preamp components.

The US Supreme Court recently heard a case regarding police use of GPS trackers for surveillance, and determined that it amounted to a search, which required a warrant: http://online.wsj.com/article/SB10001424052970203806504577178811800873358.html This case would complicate someone's position if they were to place a tracking device on a vehicle without the owner's knowledge. Anyway, they tracked this guy for 4 weeks. I can't imagine they had someone changing the battery once or twice a week, but you never know. It would be easy enough to attach to part of the car's electrical system if you did a little homework first. Even just connecting to a brake light wire would allow it to charge whenever the brakes were applied, which could easily keep a low power GPS running indefinitely. The display uses a lot of power on commercial GPS', and these devices wouldn't have displays. You'd also need to look at the broadcast capabilities of the unit. If it constantly broadcast its location to the trackers, that would use a lot more power, but if it were designed to wake up and transmit on a certain signal or send bursts hourly/whatever, you'd save a lot of power.

Since IR works up close, you don't have an IR filter. The problem is that IR illumination drops off rapidly with distance, just like visible illumination. The drop-off follows the inverse square law - boiled down, doubling the distance requires 4x the IR emitted power (or tighter focus of the IR to get more of the IR on the subject by spreading the beam over a smaller area). A faster f-stop would definitely help, and auto-iris would ensure maximum exposure in low light (as would setting a manual lens wide open at the same f-stop), but at the cost of lower depth of field and more fiddly focus. The other options are to get the IR closer to the subject, provide a tighter focused beam for the IR you have, or provide more IR emitted power. If the distance you want to illuminate is 4x as far away as a good distance, you'd need 16x more IR power on the subject.

An increasing trend in California is not to respond to burglar alarms without independent verification that it's not a false alarm. San Jose recently stated they would implement this policy due to budget cuts. Some neighborhoods, response is pretty quick (maybe 5 minutes if the're not far away), especially for issues involving violence, but for other areas, or non-critical problems, it can be much longer.

Remember the old saying: When seconds count, the police are only minutes away.

If you have spare channels, you can split a cam's output and use one channel for motion, one for full-time. I used to do this for a critical cam where I didn't want to miss anything but wanted motion to help me find the good stuff. If the input impedance on the DVR isn't high enough or the cam's output is a bit weak, splitting it may cause both images to degrade, but this is dependent on both DVR and camera. The only way to be sure is to watch it as you plug in the second input.

Intermittent fan failures seem like a good shot. These MBs should have temperature alarms for both CPU and MB that can pop up alerts, though you'd need to run monitoring software, either Asus or 3rd party. You probably have spare boxes with that many installs, so another option would be to swap the whole thing out and bench test it, or start replacing key parts like RAM, PS, etc. The most common intermittent failures I've seen in recent years have been caused by fans and RAM.

The "armed household" signs have the same issues - they alert bad guys that there are small, valuable, easily sellable items in the house. It's fine while you're home and can defend the castle, but when nobody's home, it's no help. This is the same reason I don't put NRA stickers on my car - I just don't like to advertise such stuff locally. I believe alarm signs are the most effective while not inviting trouble - burglars tend to stay away from houses with alarms for all the obvious reasons, and alarm systems aren't very valuable as fenced items (though they do imply that valuables are inside). Video surveillance signs would alert smarter crooks that there's gear to be found, and that they should be masked to avoid recognition. Of course, many crooks aren't very smart, don't read the signs, and don't even look around for cams and such, as the various screen shots have shown. Burglar alarms are reactive (or preventive with signs), guns are reactive, cams are forensic.

I bet I can guess where you're from, Fiona! I'm a Maxwell myself, but many generations removed from Scotland.

I've been thinking on using an Arduino for flexible external IR control, and the idea of a transmitter/receiver to determine how it's controlled is an interesting one. The main problem would be where to put the receiver that would allow it to see the external conditions and still be easy to connect to the system. Along the wall would be the easiest, but wouldn't necessarily show the conditions along the camera line of sight. Another option would be side by side transmitter/receiver pair that receives reflected IR from the ambient (similar to a proximity sensor), and once it gets too high, would shut off the IR and trigger the floodlight. You'd have to figure out when to turn it back on again; possibly a photocell to detect when the floodlight turns off, assuming it's on a timer. A simple, non-intelligent way would be to ramp the IR up and down constantly, using a duty cycle that ensured multiple frames captured during the ramp at a given frame rate, taking into account the speed that particular camera changes exposure settings. Your effective frame rate would drop, depending on how much of the IR ramp gave you a useful exposure; outside of this range you'd either be too dark or too bright. You'd probably need a slow duty cycle at night - a few seconds - due to the lower frame rates and the camera's exposure adjustments. This should be easy enough to breadboard and test. I've been considering this for license plate reading using external IR as a way of dealing with the IR washout, but you could tie it in to the onboard IR with a little hardware hacking.

You might try a different rev of NV5000 software; they usually have several versions available for download. This board can also be finicky about the hardware it's installed on. Is the PC dedicated, or general purpose, and is it overclocked any? This can change the PCI bus timing, which can cause trouble with some hardware. Is it a clean install of the OS, and is it 32 bit or 64 bit? I've heard the older NV5000 cards are finicky about Win7-64, while the newer ones (NV5000T) are better, but I've only run mine on XP. I never could find out the actual difference between the NV5000 and the NV5000T. When I bought my NV5000 used, I installed it on one of my utility boxes to make sure it worked, and it wouldn't install the software at all. This was with an Asus P4P800-E Deluxe MB (quite popular back in the day) with a 2.8GHz P4 and 2GB RAM running XP, and the OS install is at least 5 years old. I switched it to an ancient IBM Netvista P4 2.4GHz box with 512M RAM and a clean OS, and it installed and ran fine, but maxed out the CPU quickly. It's currently installed in a Dell Dimension 8300 P4 3GHz with 3GB RAM and a clean XP install, and is running very stable, only maxing out the CPU if I turn up the framerate on the IP cams too much. Another check you could try is to pull the NV5000 and see if it continues to crash. If it does, that could indicate marginal memory (tricky to troubleshoot unless you can swap it or pull some sticks to test) or other basic hardware problems. ETA: After re-reading your post, it seems unlikely it's bad hardware, since it's happening on 2 systems. That would make me think more along the lines of a basic incompatability with Win7 or with the MB. Do you have other similar boxes that run OK?

8 more got posted recently, and one just closed at $243, if anyone's interested.

Time to repurpose the old gunsmith joke: Q: How do you make a small fortune in the security cam biz? A: Start with a large fortune!

LPR with IR can be tricky. I have a 1MP cam that reads license plates clearly in the daytime, but when the IR comes on, the high reflectivity of the license plates makes them much brighter than the ambient, so they wash out completely. What's super clear in the daytime is just a block of white at night. If the license plates are going to be in a repeatable location, you might could zoom on that spot and use the license plate location as the exposure control, but I haven't actually tried that, and your camera would have to support selectable exposure control zones. Depending on the distance and local lighting, I believe you'd have better luck with no IR and a sensitive cam with a fast lens zoomed in on the area in concern, since most license plates are lit at night anyway and you've got local illumination, but I'm no expert on this.

The problem with IR specs, as I understand it, is that it's difficult to measure IR output without special gear, and also difficult to specify without testing. Most CCTV IR illuminators are spec'd in terms of consumed power, effective distance, or number of LEDs, none of which tell us anything useful (though distance could be slightly useful if held to specific measurement standards). IR output should be measured in IR power output, and I'm not aware of any affordable consumer gear that measures IR power. Consumed power is related to output power by the efficiency of the LED, and the actual irradiance (watts/square meter) and effective distance are dependent on LED drive, number of LEDs, beam angle (and a definition of effective illumination). You can't use lumens/lux or a lightmeter, as they're designed for visible light. Some meters will measure into the non-visible spectrum (though many block IR), but that's dependent on the response curve of the specific meter. Stanislav's example of extrapolating IR illuminance is to use his software to match the intensity of a specific pixel illuminated with a reference visible light source and with the IR source. This is probably the easiest method (though not that easy), but depends on his very nice but pricey software and a test bench setup. For visible light, I can use a relatively inexpensive lux meter and post a screen grab, while specifying camera, lens, settings, and lux, and that allows me to change one thing at a time for comparison, and allows others to duplicate the setup. It's no good at very low light levels, as Stanislav points out, because measuring light levels under 1 lux is difficult with inexpensive gear. So, the real question for IR: Is there a relatively simple way to measure actual IR output/illumination with affordable consumer gear?

Same here - upgrading from good analog gear to mid-range megapixel IP (with a few VGA in the mix). I miss the long-term stability of a good analog DVR, that's for sure. I'll echo the recommendations to stick with good, well-supported name brands (check the web site to see that the software is regularly updated and multiple versions of firmware are available), and keep to the same brand, if possible. I've got software that works with some cams and not others, and vice-versa, so it's been tricky to find something that supports my whole zoo of cams. My Vivotek 1MP IP8332s are great in the daytime - clear and sharp, with good facial recognition if they're close enough to get enough pixels. At night, the sharpness drops dramatically and noise increases, and it's back to looking like old analog views, even with IR (which washes out their features if they get too close). Best bet is to have sufficient security lighting if nighttime recognition is important. I like the IR capability, but performance is best with good area lighting. Most of these mid-range cams have 1/4" or 1/3" sensors with high pixel density, and low-light performance is mediocre. More MP are better, but can drop the frame rate, require lots more bandwidth and storage, and tax older PCs to their limits. I've got 2 Vivotek IP8332 and one FD8134 1MP, all with IR, and I like them a lot, but would love to have the 2MP versions. I also have several 5MP Areconts without IR and a Messoa 2MP NCR870 outdoor cam with IR, and while higher resolution is great, none of them are as stable and user-friendly as the Vivoteks, which have been close to bulletproof, with mature, solid, flexible, friendly firmware. They're the only ones of the lot that I can count on to run for weeks at a time without hiccups, and have worked with a wide variety of software, either through native support or as ONVIF cams. All 3 of my Vivoteks (bought used on ebay) arrived with the IR not working, but reseating the board fixed it up on all 3. I also like the extras - power supply (though I use POE), decent mount (if not industrial grade), and spare cable seal so I could switch to a direct connect net cable. The Messoa setup (also bought used) isn't nearly as nice, with a less flexible mount, no way to replace the bundle of cables hanging out the back, and it runs a good bit warmer than the Vivoteks (though I haven't measured the power draw). The Messoa software's not as mature, and it has a hard time with all sorts of software. For $300, it's hard to beat the Vivoteks. If your budget's bigger, there are a lot more choices. ETA: I disabled the wireless on my 2 Y-cams, as it was very unreliable. I'd recommend trying for a wired solution if possible. Also, with SD cards, you can record on-board and not need an NVR, but if someone steals your camera, the recording's gone too! Blue Iris provides a good software NVR for $50, if the one that comes with your cams isn't up to snuff.

It's definitely Windows Embedded 7. It looks to me like Win7e is a lot less crippled than XPe was. It shows as Windows Embedded in the Properties screen, and when it downloads updates, they're listed as Embedded. For those still interested in the performance of these, I installed Milestone XProtect Go (the free version) on it last night. I had to install .net 4 before XProtect would install, and put on Adobe Reader at the same time for the Milestone manual, as well as UltraVNC and a few other utilities. I thought about installing True Image for backups, but the Win7 backup software works on this, so I'll put that off for a bit. XProtect Go ran fine, though it had trouble with a few of my cams. The same ones ran fine in IE and BI3, and XProtect Go isn't a good solution for me due to the single client and 5 day recording limit, so I probably won't spend a lot of time debugging it. On the plus side, XProtect used very low CPU for the 3 cams I had online (Vivotek IP8332 1MP, Arecont AV5105DN 5MP, and Messoa 870 2MP that dropped out intermittently). The box has IIS running on port 80, and I haven't disabled that yet, so I set XProtect's server for port 81 with no problems. I'll probably go ahead and shut IIS down to keep the confusion down. This was all on a fresh box with the autoIP software still running in the background. Generally, you want to disable this, since it auto-detects cameras and starts recording them, using small amounts of CPU and larger amounts of HD space. I'm thinking I'll keep using one of the spare boxes for testing different NVR software while the primary records with BI3.