dnieweg

Sorry for the delay. I am getting the answers to your questions and will respond as soon as I have the info. Thanks.

4CIF is pretty close to full D1. 16 channels at 4CIF @ 30fps/channel PLUS 16 IP cameras including megapixel at full resolution PLUS 16 or 32 channels of audio PLUS full POS or connection to any open database PLUS full logging of incoming and out going Caller ID data PLUS ... well the list goes on. 3xlogic

Most people use a VCR lockbox which is available from many video security equipment resellers. It can be bolted to a surface at least slowing someone down. At 3xLogic we also have a wall-mount model that of course mounts flat on the wall which is good in some of these businesses that have very limited room. It also provides security and pretty muck looks like some type of equipment box.

Colin, is $1,200+ expensive for a camera? HA HA In certain situations though, like over cash registers this can sometimes replace several cameras so that helps I think. But yeah, the cost can put it out of reach .. but those pictures!

I don't know Thomas. Add more light? Wait for megapixel technology to get better? That is a good point and a real issue though since the CMOS chips used in IP cameras are far from being as efficiant as the CCD chips in most analog cameras. With the rich detail of megapixel though, there are applications where they a definite advantage, wouldn't you agree? What do you think of the new IQeye750 Day/Night Series camera? I have not had a chance to see it in action yet, but I am told that it is sensitive down to .09 lux in night mode and can be used with IR.

It seems that ANPR (Plate Recognition) is much more in demand in the UK than it is here in the US and I wonder why that is. I've talked to many station owners, parking lot's and even airports and while they appear to see the value, in the end they don't pursue it. Anyone have thoughts or comments on this? Anyone in the US have a client using ANPR successfully to help manage their business? Thanks

CameraGimp, That a great idea. So your IR pass filter is actually a cut filter for visible light? Have you tried this? Can you provide a link for this type of filter? I can think of several situations in which this could prove useful. Thanks! Dave

Well .. I have to admit I just don't know. It could be so many things. Did you happen to take a regular monitor out to the site and see what kind of picture the camera itself produces. You really have to isolate the issue to the camera, the cabling or the DVR/display. From what I see I am almost wondering if you are having more of a contrast issue than a dark image issue. The contrast looks very poor. This also could be caused by many things .. poor cameras, poor connections on the cabling, cabling distance. If you get a chance isolate it down to the camera, cabling or DVR/monitor let me know. Dave

DISCLAIMER - I am starting with a disclaimer right off the bat because this post really has no point .. lol .. other than to lament on the state of video quality in the security industry I suppose. I know we have come a long way, but with the exception of megapixel (which I will talk about at the end) most everything else is still substandard to television from the 1960's. With that said you can read my ramblings .. or run now. Getting good (or even great) pictures requires much more than just the capture card. As they say, it will only be as good as the weakest point. Let's start with the capture card though since you asked. First thing is to get the highest resolution capture, and I assume you are using analog cameras since you want a capture card. I will talk about IP cameras in a moment. Almost all cards that capture analog video (your standard composite camera) will do so at either CIF, 2CIF or 4CIF which relates to the image size in pixels. 4CIF is the highest resolution typically offered which is 704 pixels wide and 576 pixels tall. This is a pretty good size picture but if you relate it to your computers resolution it's probably a little smaller if you run at say 1024 X 768 which these days is considered lower resolution for a computer monitor. If you relate 4CIF in terms of megapixel, that equates to 704 X 576 = 405k pixels which is less than half a megapixel. But if your looking for high-res with an analog camera, then you are looking for 4CIF. The next part of the chain, and the one that is going to degrade your picture, is compression. You cannot get away without compression because if you followed the math earlier every single image when uncompressed will be around a half megabyte. Meaning that at 30 fps you would be storing 15 megabytes per second of video. I would assume that whatever size drive you have in mind wouldn't last long. So all pictures need to be compressed. Here is where it gets tricky .. all of the capture cards you have probably looked at don't compress the video. Nope. They pass the entire uncompressed image which was digitized by the card to the DVR software. This means that the whole 400k image is passed over the PCI bus to the processor. It is actually the DVR software that does the compression. (I will touch on HW compression in a moment) All DVR's use some flavor of one codec (the compression part) or another. MJPEG used to be popular, but now MPEG4 has taken pretty firm hold. The thing to keep in mind is that every codec used in security is a lossy codec meaning that the restored image will never look as good as the original. In order to compress down to the small sizes required it has to dump a ton of quality data. There are other wonderful non-lossy codecs out there, but they are not used in the security industry because of the processing power needed and the fact that it takes longer to compress than capture. So the question you are really asking is not "what is the best capture card", it is, "what is the best compression codec". My answer to that is none of them. The two newest codecs in popular use are MPEG4 and H.264. Both are wonderful compared to the codecs we all used earlier, but when compared to the world of codec's they really sort of suck. A big part of the problem is that both codecs are optimized for real time video. At this point you will have to excuse me because a discussion on codecs that reduce temporal redundancy (such as mpeg and h.264) quickly degrades into techno mumbo jumbo so I am going to simplify and leave a lot out .. but ... H.264 and MPEG4 are both designed for 30 fps, but most security is recorded at lower frame-rates. What happens is that if you lower the frame rate to say 1fps, you have now completely eliminated these newer codecs ability to effectively reduce file size because they must capture a certain amount of keyframes regardless of the framerate. Another codec that is somewhat passe at this time is MPEG2. MPEG2 was/is the standard used for broadcast television (i.e. satellite/DVD's) .. so it is broadcast quality. While this used to be good, you may have taken notice that we are now buying HDTV's .. because we have now decided that NTSC broadcast quality also sucks. LOL The real problem with MPEG2 though is that it produced huge video files, so it was rough on storage. MPEG4 and H.264 are roughly as good but with much smaller file sizes. So, what was the point of all that? Simply that if you really want good pictures/video it really does not have much to do with the capture card. In the end the codec will kill your chances of great video, and even if you pick the best codec, it's really all in the implementation by the DVR software manufacturer because that is where the quality issue is really decided. There is another flavor of analog camera compression which takes place on the capture card called hardware compression. It's really the same codec but it's done on the card instead of the processor. Companies like HikVision and others offer these for a reasonable price and they typically use H.264. And while the pictures look pretty good, it's still limited to that 4CIF resolution and marginally good codecs such as MPEG4 or H.264. Finally I want to talk about "REAL" quality video in the security industry and that is megapixel IP cameras. Right out of the gate we are talking image resolutions of around 1.3 megapixels and heading up to 5 megapixels in routine use in the security industry. These are still compressed with MPEG4 or H.264 but you have so many pixels that the image still looks wonderful. And, it does not even require a capture card because the camera does the compressing. But of course they are expensive. So what do I recommend? Look for a hybrid solution that can handle megapixel IP cameras. For cameras where routine quality is ok, use the cheap analog cameras. For the few places where you really need that intense quality, use a megapixel IP camera. Confused yet?

Looks like the white balance could be off or whacked. If the camera has automatic white balance (AWB or WB) try turning it off and then on. Before you turn the auto white balance on try pointing the camera at a white wall. Typically the white balance sensor looks at the center of the image. The real place to start with troubleshooting though would be a process like this so we have all the data. 1) Reset the camera to factory default 2) Get a different monitor and a different cable and take the camera into a different room. If the picture is still bad it's likely the camera and one of several things could be wrong depending on the adjustments available on the camera. Some things to look for as I said earlier is white balance, and perhaps trying some of the other adjustments provided on the camera. If none of the adjustments make any difference at all, you might need to replace the camera. But if the picture is good it's not the camera and we go to the next step. 3) Bring the camera back in to the room with the same different monitor and cable you were using and see if the picture changes due to the possiblity of strange lighting interference. If the picture is still good ... 4) Use the same different monitor you've been using and see how the picture looks over the installed cable. If it's bad check things like terminations, damage etc and replace the cable if needed. If it still looks good over the installed cable then ... 5) At this point it would seem whatever is capturing the image is at fault and you would need to tell us what that is.

Nope, it probably won't work and here is why: Regardless of the resolution of the camera, the effective resolution of your captured image will be the result of your recorder and the compression applied to the image. Let's start with a standard 320 X 240 captured image. In order to get even minimal license plate recognition you are going to need around 60 pixels wide covering the plate. With a captured resolution of 320 wide this means that 320/60 = 5.3 which means that the license plate will need to fill at least 1/5th of the width of the picture. A license plate is about a foot wide in the US so that means your maximum field-of-view needs to be about 5-6 feet wide. At 50 feet your 1/3" camera has the ability to zoom (set to 9mm) to a width of 35 feet. So you are looking at too wide an area. So next lets assume your recorder can record at 4CIF so we will use the maximum horizontal resolution which is 704 pixels. 704/60 = 11.73 and again since a plate is about a foot wide this means your maximum horizontal width is about 11-12 feet. Again, I say your camera can only zoom to 35 feet wide at that distance. But you have other problems such as lighting. Your camera is not really a low light camera .. it says zero lux because they have included those nifty little IR led's that they claim will illuminate up to 100 feet. Well, first thing is that they won't. In the dark you will see movement and objects but not detail. You did say that the area was well lit, but I will guess that it is overhead lighting and only a small percentage will be reflected towards your camera. Next is the real deal killer, and that is that your target (the car) will probably have it's headlights on at night. Those two little glaring blips of light will flood the scene with light, the electronic shutter will speed up and your plate will be very very dark. Electronic shutters are a cheap mans auto-iris which provides a more limited dynamic light range than you would with an auto-iris. Next problem, your target is moving, and assuming we solve the headlight problem we will now be dealing with low light levels.. The electronic shutter will slow down so the CCD can capture enough light for an image. The camera is interlaced so it will capture one frame when the car is at one spot and capture the next frame a few feet further down the road. This will make the picture look like it's made up of two pictures that don't exactly match up .. becuase that is exactly what it is. I should point out that there is software you can use to clean this up after the fact .. but hey .. So what do you do? Can you add more zoom to reduce the field of view? Yes, but the more you zoom the less light you can capture. To improve the image to where you need you would need this: • 10 foot max field of view at that distance. • Higher light levels • Light directed directly onto the plate so that it will reflect back to your camera. • A slower moving car • and finally a recorder that could capture the required resolution We also did not touch on compression. Compression will destroy the quality of your image so you would need to back it way way off and of course this will increase your file size. So then you might decrease your frame rate to compensate, but you can't do that because you have a moving target and you need as many images as you can get in the hopes that you will capture one at the correct lighting and angle. Is all lost? lol Perhaps, but the real deal is that people all want to capture license plates and it really is more of an engineering challenge than most people anticipate. One very real option you have today are the newer megapixel cameras. Consider that the camera you specified at 4CIF recording would be the equivalent of 400k pixels .. less than half a megapixel. Now contrast that to one of the newer 5 megapixel IP cameras. That is about 10 times the resolution which would be fully capable of the resolution you need at that distance. But, you need a recorder that will record the entire 5 megapixels per image. Check out 3xlogic .. of course .. lol You still need to deal with the fact that your car is moving and that you have light issues and more specifically headlight issues. With more light your car can move faster, with less light your car will need to be going slower. You can always do a little trial and error and see what your real issues are. Dave

I looking at your images one more time I do see a lot of artifacts (especially the lower right image) which would be indicative of the image being heavily compressed. Heavy compression makes a smaller file size, but the trade off is poor quality.

It is very hard from a picture to troubleshoot the exact problem for the poor video, but you can quickly figure out if it is a camera issue, the cabling or a DVR issue. Take a good quality analog monitor out to the job site and a short coax cable that you know is good. Go to each camera location and connect directly to the camera and see what the picture looks like. If it is good or at least acceptable then it is not the camera. If it is the camera, try setting the camera to the default settings, check to ensure the auto-iris is working, properly set the back focus (so the images are still in focus when the auto-iris opens up), and all that .. this is all camera dependent of course. Next, reconnect the camera to the installed coax and go to the head-end. Disconnect the coax from the DVR and connect it to your monitor. IF the image is ok, then it is not the coax. It if IS the coax, try reinstalling new BNC's on both side or walk down the cable. If the picture is still bad, then that does leave the DVR but it could be multiple things. I believe the Integral Express Series DOES NOT have live overlay so the pictures probably look exactly the same live as the stored ones and all images are processed before you see them. I see that you are mentioning frame rate, but low frame rate does not cause poor images. It's both the image size and the amount of compression applied. If anything I would lower the frame rate to 1 fps for testing and then I would select the largest image size you can. Then be sure to reduce compression to the lowest allowed. Sorry, it's been a long time since I've configured an Integral box so I can't remember the actual settings. Now take a look at your images and see if they are better. If it still seems a bit funky, open the setting for the individual camera and set the color/contrast and all that to default. See if that makes a difference. Then tweak the settings to see if you can get a better picture .. i.e. color/contrast/brightness, etc... If none of this works, well then .. I wouldn't know what to say. As for the dark images at night ... there is one recommendation that always works and is fairly inexpensive and that is lighting. Dave

I know our product has been used is quite a few casinos. The product is sold and installed through integrators though .. not sure who they are at the moment. Anyone here perhaps? We wrote an app (Vigil Gatekeeper) that integrates the DVR with the Pelcoâ„¢ CM9740/60/70/80 series matrix switch. It uses a touch screen monitor with maps of the building to control analog and IP based cameras through the switch. We also have made the switch from MPEG 2 to H.264. We are using hardware encoding with 16 channel / box at 4CIF/30fps. It's a hybrid so they are using 5mp (megapixel) cams in key areas. Dave

3xlogic (my company), makes a hybrid and NVR recorder that records megapixel IP cameras and we also offer recompression of the already compressed image which brings the file size down to about 1/10th of the original file size. This is a pretty new feature after years of work to make sure we maintain the quality of the original image (video) and while we think it's pretty cool stuff we are still wondering if the market will find it useful. I think I'll try to get some sample images available and maybe some of you experts out there can let me know what you think of the image quality and if you'd find this feature (recompression) useful. By the way, 3xLogic is our new name. Our product is Vigil and you may have known us by our old name CSI-Tech and CamAcc. We've been OEM'ing for the last 10 years and now are offering under our own label ... 3xLogic. I hope this didn't sound like an advertisement. I'll pull this post if you think it does. Dave 3xLogic

I have to disagree with you because yes, using linux as an OS does make something NOT an embedded at least traditionally. But, before you think I am offending you, please understand that I was an embedded system programmer many years ago. Before all these readily available OS's that offer services like UI, drivers and communications. We used microprocessors like the 8051 and others for which there was no operating system, and we programmed applications that would go inside of appliances like VCR's, telephone systems, cars, stereos, answering machines, modems and hundreds of other computerized gadgets. These were embedded systems and we had no OS because our User Interface (UI) was typically a set of buttons, networks were non-existent, and we didn't interface with a bunch of other stuff. The reason embedded systems were so secure is that it was all proprietary code and it was unto itself. "Embedded", meaning internalized. I do understand that common usage can change the meaning of a term, and we in the security industry have done just that meaning that embedded is now often used to define a system in which the "shell" or the interface to the Operating System itself (like explorer in windows) has been for the most part successfully repressed meaning the user cannot directly access other programs that might be in the system. So I can accept that, but I personally cringe every time I hear someone say they want the security of an embedded system, because "embedded" systems today almost always have an accessible OS running in the background and this offers no more security other than the user doesn't have a direct option to access the OS in the primary User Interface. My personal opinion, is that today's "embedded" systems can present additional challenges to security simply because they are standard OS with some of the functionality stripped out of them making it nearly impossible to apply any type of patch or update that may have been release to deal with a discovered security flaw. Dave

I agree, they are all DVR's because they are all Digital Video Recorders. However, the security market has come to define an analog only recorder as a DVR and IP only recorder (or software) as an NVR short for Network (Camera) Video Recorder. However, both connect to the network/internet for remote viewing so they are really both "Networkable" Video Recorders as well! ha But let's talk about the middle ground and that is the Hybrid Recorder. No one has come up with a nifty acronym yet for that one that I know of. How about the HVR? he he Now we can confuse the market even more. By the way, an HVR (<--- new official term according to me) can record both analog and IP cameras and therefore can provide flexibility for those who have or want to use analog cameras along side IP cameras. I guess I should point out that most IP recording and management solutions allow you to convert an analog camera to an IP camera by using what is called a video server. However, I would also like to point out that if you want to use a lot of analog cameras (which is sometimes a good choice) this is way too expensive and all those little boxes are quite a clutter in the back room. A hybrid recorder basically has the Analog to IP converter built-in (at least that's a good way to think about it) which makes it cost effective and clean .. when mixing analog and IP in the same system. While we are on the subject of confusing misnomers should we talk about the fact that we call linux-based systems embedded systems when in fact it is not an embedded system? ha ha I think I'll save that for another time. Cheers. Dave 3xLogic, Inc,