MaxIcon

I did a series of synthetic benchmarks on the Dahua WDR settings a little while back, and while they're interesting, they don't really reflect real-world performance. I just happened to take a bunch of driveway clips last night, using the Dahua 3300 and the Hik 1080p, and most of these will be video posts, due to the dynamic nature of the lighting and the cameras' responses. It'll take a little time to get them uploaded, and I'll probably do that over the weekend. Boiled down, IR is not your friend when it comes to reading licenses, due to the high IR reflectivity of licenses in an environment with large lighting swings. In addition, these cams don't really respond very quickly to dramatic lighting changes (my Vivotek IP8332s are much more responsive), and this complicates things. Far and away the best performance on reading plates at night came with the cameras in color mode with the IR filter in. If these cameras had better WDR, it might be different, but you need a more expensive camera for that. In addition, you'll get the best results with a dedicated license plate cam, as the people who do serious LPR have found.

Yeah, there's a lot to be said for a dedicated NVR. You gain reliability and simplicity, but lose flexibility. One important consideration with dedicated boxes is to check the maximum bit rate or MP they'll handle. Some 8 channel NVRs won't let you populate it completely with 1080p cams.

I use IP Cam Viewer Pro on the iPhone, and it works great. It's also available on Android.

The latest Hik firmware, 5.0.0, still includes the 3MP setting, but it doesn't work any more. I never trust an X.0.0 software to be bugfree, though. Whether this will be enabled/fixed in an update, or they'll get rid of this setting, is anyone's guess. This means the Swann isn't really either camera. If it were an actual Hik model, it would be a DS-2CD2022, but Hik doesn't offer a 1080p version last time I checked. I assume the Lorex is the same, but haven't tested it.

The Hik software doesn't have profiles, and I don't think you can send command-line parameters to it to allow changing the settings at different times. You pretty much have to find a compromise between the best day and night settings.

The Blue Iris community forum at cam-it.org has a lot of data on CPU vs camera load, and what affects it, so you might want to go there for more info on that solution. You can also download a demo version, but can't run direct to disk on the demo. Everything else works the same, I believe. I'm using BI on an i5-3570k, built a few months ago to allow more cams than my old system. BI recently had some upgrades that lowered CPU utilization. This is what I'm running currently: 9 cams, 11 MP total, 10 fps, motion detect with 25 pre-trigger frames. On BI 3.23.05, I see 55% CPU when it's idle, spiking up to 70-75% with multiple cameras recording, without direct to disk enabled. You're looking at 20 MP total, so this box wouldn't handle your load at 10 fps with my setup. Direct to disk recording reduces the CPU load, but i haven't tested this yet. There's a big thread on it over at the forum, though. A few cam-it members have ordered i7 boxes to see if the extra threads help performance, but there's no data on it yet. Another alternative would be an Aver NV6240 or NV6480 in a lower power PC, which would handle this load with no problem, but the cards are expensive new. I've got some detail on performance in this post: viewtopic.php?p=220180#p220180 Others will have to pitch in on other software; these are the only 2 I have experience with.

Every software NVR is different, but one general difference is whether it's just a server or a client/server combo. Decoding for display takes a good bit of CPU, and BI's decoding/re-encoding is also CPU hungry. Some packages are just recorders/servers, and the client is a separate software package. The BI direct to disk update doesn't re-encode, but it must still decode to perform motion detection (which also uses CPU power). For example, my Aver NV6480 card in my backup box records 16 channels - 8 channels of motion detect, and the same 8 channels recording 24x7. These are running 10 fps, 10MP or so, for a total of 20MP on all channels. The Aver (which doesn't do hardware encoding) is on an i3-540 box running Win7 Embedded, 4G RAM, on-chip video, and the CPU load totally depends on how many cameras are displayed: 1 camera window - 38% 9 camera matrix - 55% 16 camera matrix - 91% The same box running BI before the direct to disk upgrade could only run 6 of these cameras at 6MP total at about 60% with no recording, 85% with 3 cams recording, and would still peg out the CPU at certain times. It couldn't handle one more 1MP camera. The Aver card is expensive if bought new, but it's cheaper than buying multiple i5 or i7 boxes. The software is a trade-off as well; it's robust and supports many corporate/commercial features that BI doesn't, but doesn't have the multitude of user-friendly features and great user interface that BI has. So, you really have to test each one using your setup to figure out how it will perform, and whether it'll do the job you need done.

Yeah, I've had good luck opening up the shadows with contrast and brightness on my non-WDR cams, like my older Vivoteks. It takes some testing to get a good balance, but works pretty well. I had high hopes for WDR on these new generation cams, though. It's a shame these settings don't work better across the whole range. Maybe as the software matures this will improve. Another option that would help would be more cameras taking command-line settings to change parameters. Some higher end cams, like my Arecont AV5100 series, allow you to set a variety of parameters from the command line, and some NVRs, like Blue Iris, take advantage of this. Blue Iris now has sunrise/sunset calculations based on your latitude and longitude, so it could be able to change camera profiles based on this if the cameras would take the commands. More to add to the wishlist...

My wife has a small business, and she also has to charge in that range to make it worthwhile to do anything, due to all the overhead. People think that's a lot of money, and it would be if you worked a 40 hour cubicle job for that, but not for a service/install business. I'm not a pro, but here's what I'd break a job into: - Hardware and software costs. - Install costs - running cables, mounting and focusing cams, setting up the gear both offsite and at the customer site. - Initial training for the customer (varies a lot depending on your target customer base). - Creation of training materials and docs, if you use something besides just giving out copies of the manuals. - Support - warranty, service, fixing customer-induced problems, etc. The cost of physical install can be a major portion of the job, especially if you need to hire an electrician to run cables, for example. Lots of hours involved, even on simple jobs, and not something you'd want a minimum wage employee to do. As gencom says, you have to be a bit of an IT expert these days as well. If it were me, I'd seriously consider setting up a system at home to get my feet wet.

Hey, the main payback here is to post your own stuff once you get things up and running. My goal is to find the best inexpensive cams without ending up with a box full of junk! Luckily, I enjoy this kind of testing. I'll get some more day shots with the new firmware Wednesday; I'm off that day, and will be home during those deep shadow times. As for the smart IR, I've realized I have no idea how to tell if it's doing anything or not. There's no setting for it, and I haven't been able to come up with a good test for it, but that's a matter of trying stuff to see what works. Not only that, one of the earlier releases listed it as well, so it may have been there already. In general, IR performance is probably the hardest thing to measure overall. The more techniques I try, the more difficult it looks to be to get objective test results. This is where the side by side comparisons come in handy - you don't get a measurement, but at least you can tell which is better or worse.

Is this a Dahua dome? I've been seeing the same discoloration on my Dahua HFW3300C, though not all the time, and there was another post recently on this as well. It does look like when the IR filter is out during the day, which gives that purplish hue to everything. Unfortunately, there's no easy way to tell if the IR filter is in or out. You can try switching it manually to night mode, then back to day or auto, and see if the colors recover, but the other thread had some other changes that would fix it too. ETA: Here's the other thread - similar to what you're seeing: viewtopic.php?f=19&t=35561

Interesting - so the main/only difference between these boxes is the motherboard, which presumably includes the alarm inputs. The Hik camera uses either admin or root as the login with the current admin password for telnet. After changing the password, it would login on the web page fine, but I had to power cycle it before it would login by telnet. After I changed the admin password, the default password didn't work. I don't know if it has a hardwired telnet backdoor like the Dahua; if it does, this hasn't been published that I can find.

OK, here's another set of Hik/Swann 1080p and Dahua HFW3300C comparisons. If I had the 3200S, that would be perfect for this, as they're the same form factor and price range, but the Swann is holding up quite well against the more expensive Dahua. All these are at the default settings, 1080p, with the Dahua zoomed out to the same field of view as the Swann. Both are set for a 1/30 sec minimum exposure to prevent motion blur, as that's how most people will use them in real life. As I mentioned in another thread, the Swann lens appears to be a bit less than 3.3mm, not the 4mm it's listed as, and the Dahua's at 3.3mm. I'll continue to use the Blue Iris split screens, which are great for comparing identical time snaps, as well as some Hik-only shots. Overall, the Hik has more pleasing and detailed shots, but less night sensitivity. The settings can be tweaked to bring the images closer to each other, but for starters, I'm leaving them at the defaults. Also, the night shots have IR on for both cams, so IR performance isn't really a good comparison here, since both benefit from each other's IR, but the night sensitivity is a good comparison. Also, gotta love that Hik adaptive overlay lettering. This is definitely the future on better cam firmware. First, here's the overview shot - heading into dusk, but still a good bit of light out. The Dahua's a good bit brighter (which contributes to both night performance and its failure to switch to night mode in many situations), but the Hik is much more attractive. The Hik also doesn't suffer from the soft patches in the grass like the Dahua does, but shows some minor sharpening artifacts. Both the Hik and the Dahua will show the colors off into the purple end now and again. Here's a Siemens focus star, 10' away from the cameras, zoomed in all the way in BI. The Dahua has slightly better focus or resolution, but the Hik is on factory focus, as it's a pain to get into it. You can see from the brighter halos around the Dahua star that it's more over-processed, and it's a lot less dark. Switching to BW while there's still light out, but no IR on yet, shows that the Dahua lens has better IR correction. The star center focus is about the same on the Dahua as before, while the Hik loses some definition, though not much. The over-processing matters less in BW mode, and the Dahua's better night sensitivity makes the star brighter. When it gets a little darker, the IR switches on on both cams, for this view. The Dahua's still brighter, but the Hik still looks better - crisper and cleaner. Here's how the stars look zoomed in with the IR on. The Dahua's not looking so good against the Hik. Now, it's pretty dark out. Again, the Dahua's got better sensitivity, and shows a lot more at the fringes - above the fence, and the IR splash on the building at the far right (from a Vivotek IP8332) is much brighter. However, the Dahua's a lot softer. Here's a zoom on a darker area that shows the trade-off in night performance between the two pretty clearly. The Dahua's still brighter and shows more highlights, but is softer with sharpening artifacts from noise. The Hik is a lot sharper and more detailed. Which is better? It depends on your needs. The sharpening artifacts contribute a lot of apparent motion in the actual video. Here's the same shot with a pair of 150W halogen motion detector lights on. Again, the Dahua looks good from the big picture perspective, but the Hik is a lot sharper and more detailed. Now, there's also been some talk about the softness on the right side of the Hik, so I thought I'd grab a few shots of that as well. I did these over the weekend, and they helped me figure out how to get the shots I needed for the comparisons above. Here's a focus star at the far left of the Hik field of view. I measured 10' out, then took the star to the edges of the field of view, so they're further than 10' at an angle. This is all in the daytime. Same shot, but at the soft right edge - yuck! That looks like a lens element problem to me, but if that were the case, you'd expect different cams to show the problems in different areas, rather than all on the right side. Maybe the reason will be clear with more testing. So, does the IR filter possibly have a problem along one edge? I switched it out, and tried again. Left side, not nearly as good as with the IR filter in. Ugly right side with IR filter out - even worse! How does this compare to the Dahua? Here's a shot of both in color mode. It was windy that day, so I have some bags of PVC fittings holding the star onto the stand so it won't blow down. Switch to BW mode, and like the tests above, the Dahua's crisper and sharper in daylight, showing that the IR correction on the lens is better for the Dahua. It also demonstrates how the edges of a lens have worse resolution in general than the center, which is typical on inexpensive lenses.

Me too. Better firmware, better support, all around better cams for less money, it seems. So, the Hik lens is glued into place, and I wasn't up to wrestling with it, so no lens swap tests. They really don't make it easy to get into this camera - it's not designed to be opened by the end user.

Nice info! I know size isn't everything, but between the larger MB and the fan on the CPU, I wonder if the Swann system has more horsepower. There's a program on many Linux systems called top that shows the amount of CPU each program uses that's currently running. If you can telnet into the system and run it, it would give a good idea of how much the processor is working for a given setup. That appears to be the same power supply in each box. Is there another hard drive connection on either one?

You may need to drop the bitrate if your video is jerky or if it has network dropouts, but the latest BI update (3.23.05) appears to have gotten rid of the Hik/Swann network dropouts (though not on the Dahua). Excellent support!

So far, it appears this firmware is for all the cameras based on the Hik Raptor platform - 1.3 and 3MP. I haven't seen a 2MP version yet, but it's the same firmware footprint as the faux 3MP Swann update, so that one appears to be covered too. I'm not sure the NVR has access to all the settings of the camera. Best bet is to access the cam directly via the web page and see if it has WDR enabled.

That's actually pretty good night video quality, in general. I don't think you can get a lot more out of these cameras without improving the lighting. There are a couple of problems here, and they're the basic tradeoffs of layout, lighting, and resources: - Field of view vs detail. This is a problem on surveillance cams in general, and is much improved over the analog days. To get better detail, you'd need to be zoomed in, but then you're more likely to miss out on parts of the action. More pixels per foot give better recognition, but either give a smaller field of view or require more expensive gear. - Lighting. Your lighting is good in general, but he's in the darker areas. If he had come to the front of the truck, with more light and more pixels per foot, you'd have an excellent shot of him, but he didn't. - Your coverage is quite good, too. You have 3 cams that caught the action. Again, better lighting would be your single best improvement with the existing cams. There are a few ways to improve results with this kind of monitoring. - Better cameras. These are great cams for the money, but you'd have much better results with high end cameras, more MP, better low light sensitivity, 1/2" sensors, etc. Most non-corporate people aren't up to spending $1000 and up for each camera, though, and having higher MP resolution requires more expensive NVRs, more disk storage, whatever. If you do go that way, Avigilon has some great technology, but for a lot more money. - More lighting. If the other end of the driveway had been lit as well as the house end, you'd have better shots. This would require more lights, maybe motion detect at the end of the driveway, and too many lights can create tension with the neighbors, as well as increasing power usage and costing more to install. An easy change that would have helped in this situation would be to take the camera that's focused on the side of the truck and rotate it more towards the street. The light on the house is causing it to increase the exposure speed, which darkens the rest of it. Since you've got some focus on the front of the trucks, this would give you better exposure and coverage of the rear of the trucks and the street. You might want to chop that tree down that's in the way of the camera watching the front of the trucks . That, or re-position the camera. Not only does it block some of your view, the lit-up tree affects the exposure for the overall scene. Again, this is a tricky set of tradeoffs, and you may already be in the best position. Of course, even if you had an awesome picture of the guy, it wouldn't help unless the cops recognized him or he was someone local and recognizable. The good news is that you know exactly what happened, so there's no guessing about what went on. It may not help unless they can catch the guy and use your footage in the case, but it's better than nothing.

If you have the WDR setting on your cams, this helps improve the night mode. I set one at 50, and it really helped the night shot a lot. Brightness and contrast can help night images or day images with lots of shadow, but are almost always a trade-off. For the infra-red, you can put tape over some of the LEDs, but how effective this is will depend on the distance and angle to the lines. Since the remaining LEDs have a good bit of spread, they may still hit the lines, but it's worth a try. This is something the next generation of cameras needs, now that WDR is becoming more common - adjustable IR, or better yet, intelligent IR, where it adjusts it automatically depending on the image. ETA: Q sent me a link to the lastest Hik firmware release, and lo and behold, it includes this: Support SMART IR (only for IR featured models) So, it's too early to say if the Hik firmware is really that much more awesome than the Dahua, but for now, Hik is seriously kicking Dahua's butt. Add that you can get the software, as well as a list of changes and bugfixes, and Dahua is, like, totally last week. I just hope I can sell mine to some sucker, ummm, n00b, errrr... adventurous but ill-informed soul! Here's the rest: 1. New Features Support three new date types: (month/day/year), (year/month/day), (day/month/year). Support DDNS for No-IP; Support anonymous login; Support UPnP; Support IP filtering (totally 48 IP addresses can be managed.) Support one way alarm I/O, audio, local output and auto Iris functions for SD card featured models; Support volume control. (DSP functionalities) Support the OSD font setting (ActiveX required); 16*16, 32*32, 64*64, self-adaption Support motion areas highlighted by frame in live view mode. (Core and relative drive) Support AK4954 audio codec drive; Support amba emac and phy hardware fluid control. (Front-end parameter and ISP) Support SMART IR (only for IR featured models); Support more AWB settings: manual WB(R, B gain adjust), AWB1, locking BW, filament lamp(3000K), warm light lamp(4000K), nature light lamp(5500K), daylight lamp(6500K). Support D/N switch timing mode (ICR start and stop timeline can be configurable); Support D/N switch alarm input triggering mode (only for alarm featured models). (Standard protocol and ActiveX) Support HTTP PUSH MJPEG encoding for sub-stream (only Firefox supported); Stream URL: http://ip/Streaming/channels/ID/httpPreview Support SDK Input in ActiveX; Support ActiveX anonymous login; Support 13 languages including Bulgarian, Hungarian, Greek, German, Italian, Czech, Slovak, French, Polish, Portuguese, Spanish, Russian, Dutch. 2. Improved Features (Application) Changed RTSP authentication as default setting; Changed the audio input default value as Line in; Optimized the smoothness of live view. (Front end parameter and ISP) Solved the bright spot issue. (DSP relative function) Optimized the OSD characters overlay in proportion; Modified QP parameters.

Those sound like some great features. We're entering a golden age of inexpensive IP cams. The problem's going to be keeping track of them!

That I understand but HOW MUCH of the 100mb pipe does it uses when transmitting at say full quality. Let say the switch is only 100mbps, the uplink to the router maxes out at 100mbs. The uplink is a single port on the switch that will be carrying the video stream from ALL the cameras at the same time. If say one swan cam uses max of 25mbps and the foscam uses 10mbps (4x 25mbps) + (2x 10mbps) = 120mbps combine. bandwidth of all the cameras at 120m it will overload the uplink speed going back to the router. and you will have either frame drops or camera not showing up. or some garrbled video. So now I would need 1000gb switch to handle the uplink going to the router. Due to the nature/location of where the cameras are wired to and the placement of switch I don't want to use anything expensive if I don't have to. *** SOMEONE correct me if I'm wrong on how it works, as I currently run 6 foscam off a full gigabit network. So, this gets back to what you set each camera for. If you have 8 cameras, each set for 8192 kb/s, you'll be sending 65.5 Mb/s through your switch. Throw in some overhead for grins, call it 80 Mb/s, and you'd be fine. If your Foscams are D1/VGA, they'll be fine at 2048 kb/s, which is a bit of overkill. 8192 kb/s is fine for any 1080p camera out there, unless you're sending MJPEG or similar. A super-sharp camera could benefit from a higher bitrate, but we're not talking about anything like that. If you're approaching the limit of your switch, the higher backplane and fabric speed of a Gb switch may help avoid congestion, but this depends on the design of the switch.

So, I started to get some night shots, and realized at the end that having the cameras so low was messing up the exposure, especially on the Hik, due to the bright IR on the ground in front of the cameras. I'll have to mount them up higher for meaningful results. I got some low-light color shots instead. I now believe this is why the Dahua appears to have more sensitivity in the previous series - the Hik is speeding up the exposure above the minimum. The following shots show the Hik to be more sensitive overall. Here's a good shot of the two cams in BW with IR on for both cams, so there's more than usual. The interesting part about this is the cat, who has a big shaved spot from a recent vet visit. The Hik shows a lot more detail here, as usual, which is pretty critical in night shots. The Dahua blurs the shaved border so it's almost unnoticeable. OK, on to the color shots. Both cams are in color mode, with the IR filter in. The IR LEDs are still lit on the Dahua, since they aren't controlled by the processor, and are off on the Hik. This doesn't matter much, though a little usually gets through the IR filter. There's a card with some text on it for the IR testing, 20 feet away from the camera. Some of the shots are at 1/30 second minimum exposure, which is as slow as you can go and avoid motion blur. You'll still get some blur for fast moving objects, but this is a popular tradeoff. Others are at 1/6 second minimum exposure. This will blur someone walking at a normal speed, and will also limit your frame rate to 6 fps max by definition. This setting gives much better low light performance if you don't care about motion blur. These shots have a light in the house on, spilling out the sliding glass door. There's not really much light there, but there were people working at a table, so I didn't turn it off for all the pics. These are 1/30 second minimum. This shows the noise that the Hik develops in very low light, which shows up as horizontal lines as well as normal noise. The Hik is clearly much more sensitive at these settings than the Dahua, and the image is on the edge of being usable. This is the same shot at 1/6 second exposure, making the light from the house look much brighter than it really is. Here's a closeup of one section at 1/6 second exposure, showing some interesting differences - lots of imagery from the Hik, but with lots of noise and sharpening artifacts, while the Dahua loses all that shading, but you can almost read the paper where it says "144 Point". Same shot, 1/30 second, but with the light in the house turned off, so there's no light at all. The Hik is still more sensitive, but neither has a usable image. Same shot, all lights off, with minimum exposure at 1/6 second. Again, not very useful, but shows the superior Hik sensitivity. Finally, here are the same shots with the 2 x 150W halogen motion lights on. Again, the Hik is more sensitive, leading me to believe that the earlier shots where the Dahua shows more in the background and fringes is due to the Hik changing exposure, rather than the Dahua being more sensitive. Same shot, 1/6 second exposure. As expected, more detail and a better looking image from the Hik. I believe the Dahua has brighter exposure due to the Hik compensating for the bright lighting on the grass; that is, it's probably not at the 1/6 sec minimum, but this is hard to tell for sure. Here's a closeup of the 1/6 second exposure. With decent lighting, the Hik shows more detail on the paper, while the Dahua shows more overall image, again due to longer exposure time. So, hopefully over the weekend I'll get some meaningful IR shots, as that's really where it counts. Based on these results, I expect the Hik to outperform the Dahua, but we'll see.

Here's the Swann with WDR set at 50, following up on a suggestion by Buellwinkle to equalize the night exposure more - definitely an improvement in the shadows, and much better than the Dahua in sharpness. The Dahua WDR really isn't very good, and seems to be more of a contrast adjustment. The more I test the Swann, the less I like the Dahua. I guess I'll try to sell it to someone who doesn't read these boards! As seen in the following post, I think the better Dahua night performance at the fringes is due to exposure times, rather than sensitivity.

can you link to it? it's a PITA searching the thead with my phone and also I can't tell if some clip are from the NVR overlay or the actual time/date overlay from the camera itself. thanks for the quick reply. Sorry, forgot about this. Here's one of the pics with the Swann overlay. The time/date info in the top left and the word Hik are from the camera; the time/date below the word Hik is from Blue Iris. You can see the time/date is in white and black, depending on how the camera perceives the average background.

Yeah, this kind of thing is a good bet, and helps address the whole problem of "how good is my IR?". This is a tricky thing, since IR reflectivity isn't obvious until you put something in front of the camera. A printout of a face, for instance, may not show up nearly as well as a real face, while the BW face on the figure above will show more strongly than a real face. I like the human size test figure - that covers a lot of bases at once. That would be a good thing to make in modular form so you could take it apart and store it without giving yourself a fright every time you walk in the garage!