MaxIcon

For the door reinforcements, go to Amazon and search on door reinforcement. There are lots of reviews of them, and it makes it easier to pick the ones that will work for you.

The cameras operate at whatever you set them to. I believe you can go over 8192 kb/s, but in general, there's no need to. This setting is independent of FPS, resolution, etc, but it can affect image quality. Too low a bitrate setting will cause pixelation during fast movement or when a lot of the image is changing at once. Also, if you're going to use Blue Iris, it has a problem with these cameras (and the Dahua 2MP/3MP cams) as you go above 4096kb/s, and starts having network dropouts if it's viewing a well lit area with alot of texture (like a yard full of grass and trees). The problem starts with the Swann at lower settings than with the Dahua, and I believe it's because the Swann has a more detailed image, with less softness. I'm gathering some data on this, and have some theories, but that's another post.

The Swann puts up overlays, and has a nice feature where it switches between black or white text, depending on the background color. The only thing I don't like is that it's pretty big text and is a bit far from the edge of the image. There are some examples of the overlay in the evening and focus comparison post I put up, with the BI overlay enabled as well.

I haven't opened up the Hik yet, but will do that this weekend. I have a few spare lenses that will show if that's the issue. Even if it's not a better lens, it'll be different, which is all we need. It's not really made to be opened easily, unlike the Dahua, but a challenge is always good!

So, this is part of what I'm trying to do here - come up with a repeatable way of testing that people can compare from different locations, using different cameras. I'll post the other tests here shortly, which use some more traditional techniques. The Siemens star is a good tool for this. It's not easy to get an absolute measurement of camera quality, like TVL, but you can easily get a relative measurement like above. This is harder to do comparing different focal length lenses, but we should be able to work up some standards for this. For testing the lens quality side to side, or the Swann side softness, the tests above work great. All you need is a decent quality printout of the Siemens star and a way to hold it up straight (three would be ideal, now that I think about it), then put them at the edges of the screen and compare how they look. It doesn't even have to be the Siemens star, but any similar shape that will make it easy to tell when quality shifts. I always use the star for focusing, and it makes it really easy to tell when your focus is the best it can get. If you google siemens star, then click images, you'll see a huge assortment. This one from Wikipedia is very good - high resolution, clean and sharp. It's a .png file, but most graphics programs, like Microsoft Paint, can handle that these days. http://upload.wikimedia.org/wikipedia/commons/thumb/5/50/Siemens_star.svg/2000px-Siemens_star.svg.png Print it to fit on a piece of paper (or three), mount them on cardboard or foamboard, and tape it to something like a big nail, a stake, whatever you can use to hold it upright, if you'll be using it outside. I use a folding stand for my tablet. For indoors, just tape it to whatever is in the right place. Measure some distance straight out from the camera (I use 10'), mark it, then go in a straight line to each side and put the stars at the edge of your field of view. They should be about the same distance from the camera at this point. Alternately, for indoors testing, just tape up one star and move the camera so it's at one side or the other. Take a snapshot at your highest resolution, zoom in on each star, and it'll be obvious if one side is different from the other, just like above. For even more info, put a 3rd star at the center spot, since that's where your lens quality is best. Good lenses won't degrade too much from the center, while cheap ones will be worse. Next up is an attempt to figure out the true effective distance of a camera's IR, which is not an easy task. Again, hard numbers are hard to get, but relative numbers may be easy.

OK, updated to the local forum version! I realized I should throw in the firmware versions, since this can affect results. You never know what changes in updates unless you have a comparison from before. Dahua IPC-HFW3300C: Software Version - 2.103.0001.0.R, build : 2012-09-14 WEB Version - 3.0.0.0 Hik/Swann: Model - SWNHD-820CAM Firmware Version - V4.0.9 130106 Encoding Version - V4.0 build 121228 Interestingly, the Swann's label says CONHD-A1080X4 (part of the 4 pack, maybe?), and the serial number on the label is completely different from the serial number the software reports. The Dahua is the same - it's got a K&D label, with model number KDW-HW47RC83, and a different SN than the firmware, but it's always had Dahua firmware.

Typical IR LEDs are 850nm, which is the right on the edge of the visible spectrum. You can also get 940nm LEDs, which are pretty much invisible, but have a few drawbacks: - They're more expensive, because they're not as heavily used as the 850nm LEDs. - Camera sensors are more sensitive to the 850nm part of the spectrum, and less sensitive at 940nm, so it takes a lot more power from the 940nm LEDs to get the same illumination. Here's a chart of the typical spectral response of a CCTV sensor. The details will vary from sensor to sensor, but the general response curve is similar for all of them.

For the timeline issue, there is a timeline mode, but I haven't used it much. What has helped is the setting for whether each clip is a standalone video, or part of a longer clip. By setting it to break every 8 hours (or whatever), you get the clips embedded in a much longer file. Once you've started playing one, you can scroll through the entire 8 hours, which is much better than the individual clips. I used to use the standalone clip setting, and ended up with many 10s of thousands of clips in my folder, which bogged down my low power box when BI scanned them every 10 minutes to archive older ones. The new setting has reduced the number of clips greatly, as well as making it easier to scroll through them. It does limit the start/stop precision when you're exporting a clip, so I just export a larger time window, then open that and trim it down. I like the way the Aver software handles the timeline. In the 1 hour timeline display, it shows green at the locations where there was motion recording, so you can skip to the key spots easily. BI doesn't give a visual indication like this, but it's still pretty easy.

For those interested in Blue Iris, but put off by the high power PC required to run a lot of MP cams, the developer recently added direct-to-disk recording. This records the cam stream directly without re-encoding. It has some downsides, like BI won't be able to add it's own overlays (time, camera name), but most cameras do that themselves these days. I'm sure there are other limitations, so whether the trade-offs are worth being able to use a less powerful PC is up to the user. I haven't tried this setup yet (no time), but initial reports are very encouraging, with pretty dramatic drops in CPU usage. There were a few bugs early on, but these were squashed quickly by the developer, who's always responsive to bug reports. Anyone interested can follow the threads over at www.cam-it.org, the BI community forum. Please don't take this as shilling - I don't get anything from this, but I'm a satisfied BI user, and this is a major improvement in an already impressive product.

The second connector on the pigtail is a standard 2.1x5.5 mm 12Vdc power connector. I fired up the Swann on one of my old analog cam supplies and it worked fine.

Hey, Stanislav is back! Always good to see posts from an expert! To expand a bit, sharpening is similar to edge detection, while noise reduction is closer to blurring or averaging. Most Chinese cameras oversharpen as the default these days, and some don't allow it to be adjusted. This is a big problem in low light, as it brings up a bunch of rectangular artifacts around the normal camera noise, increasing bandwidth and image size. It also can trigger motion detection, which is a problem on one of my Dahuas. Around dusk, before it switches to night vision, it detects the oversharpened noise as motion for 15-20 minutes.

The TP-LINK TL-PoE150S PoE Injector is a good bet for one camera, but if you're going to run more, you should look into getting a Zyxel switch. The datasheet means it will run on either a local 12Vdc supply with it's own dedicated plug, or on 802.3af POE, which is 48Vdc over the ethernet cable with negotiations to enable it for appropriate devices. http://www.newegg.com/Product/Product.aspx?Item=33-997-219&ParentOnly=1&IsVirtualParent=1

You'll need another POE switch, or a set of POE injectors (which would cost more than the POE switch). Silly design decision, but there you have it.

I've posted some comparison pics on the 3MP setting over at cam-it: http://www.cam-it.org/index.php?topic=4248.0 I'll toss them up here when I get the time. Anyway, as was pretty clear already, the Hikvision flash doesn't give the Swann true 3MP capability, but scales it from a lower resolution, as Tom said, which I estimate to be about 2.2MP. Also, this doesn't appear to be a 4mm lens, with 70 deg FOV, which is what the Costco spec claims. The horizontal FOV at 10' is 18', which calculates out to approx. 90 degrees. Traditional 4:3 lens calculators claim this is 2.7mm, and scaling that for a 1080p FOV gives a rough estimate of 3.6mm. However, it just happens to nearly match my Dahua HFW3300C at full zoom out, which is 3.3mm. If anything, it's slightly less, as the FOV is slightly wider than the Dahua, but 3.3mm seems like a good estimate of the actual lens focal length. More 3300 to Hik/Swann comparisons to come later. Here's the Dahua/Hik comparison overview - Hik on top, Dahua on bottom, from a BI screenshot.

So, the Hikvision/Swann keeps the same field of view at both 1080p and 720p, which I believe is a smart move for a consumer oriented camera, since explaining this kind of thing to people without IP cam experience can be frustrating for both parties. Which MP IP cams keep the full field of view at lower resolution, besides Hikvision and Vivotek? Dahua crops when you switch to lower resolution, and there have been some threads complaining about this. What about other popular cams?

No, you're right. I was mis-remembering the complaint thread, which had to do with the 3MP vs 1080p settings, which have to be different FOVs. Never mind, carry on...

You can check if Blue Iris supports the Swann DVR. It's very flexible and not too expensive.

A few things you should check before ordering a lens: - The lens should match your sensor - 1/3" lens for 1/3" sensor, 1/4" lens for 1/4" sensor. Otherwise, you'll get a different field of view from what you expect. - You'll need an IR corrected lens. Otherwise, whichever light source you focus with (white or IR), it'll be out of focus on the other. - A glass lens will give better quality than plastic. The cheapest lenses are often plastic, but not always. Best to find one that specifies glass. If you have a 1/3" sensor, www.m12lenses.com has had good reviews from other members. If you have 1/4", it's mostly ebay or www.dealextreme.com, which are both a gamble, unless someone else knows a good 1/4" lens source.

What version of firmware did this one come with? Those sound like some good improvements. It doesn't have WDR enabled, does it?

Sigh. Look, I understand your point of view. This is your business and your living, and you don't want to give away free expertise; it's completely understandable. Nobody wants to work for free. I'm just a hobbyist, and I'm also a troubleshooter and maker. I like to help people who don't have my technical expertise. I do this for fun. I never mention TFTP except when someone has *already* bricked their device. If they had Dahua support, they'd have used it. These users who have already made a mistake and now have unusable devices now have no other option. Dahua doesn't offer them anything, that's for sure. Most people don't have the technical skills to TFTP their firmware, and many of them don't have the files they need even if they did have the skills. Still, this is what Dahua would do to fix their gear if they could get Dahua to help. They ask "Is there anything else that can be done?" So far, nobody's ever come back and said "Hey, that TFTP upload fixed my problem, and here's how it worked!" If they do, well, there's a useful data point. If I ever get the opportunity to make it work, I'll publish the details, because that's what I do. Until then, yeah, the advice may be useless, but it's the only thing they have. If Dahua offered them support, it would be a different matter, but they don't. If someone else offered useful suggestions, that would be great, but they don't. When your only option's the trashcan, where's the harm?

For what it's worth, this compares identically to the same rev firmware uploaded by ilkevinli in another thread. Thanks to both of you for helping out. It's sad that it's come to this, but it's good that the community is stepping up to the plate. Those of us without access to Dahua's inner circle appreciate it.

Well, that's amusing. The 720p and 1020p are exact same field of view, just different pixel counts. So, that brings up the next question: Is the 720p a downsampled 1080p capture, is the 1080p an upsampled 720p capture, or are both scaled from some other resolution? I know lots of people like interpolated images, and are disappointed when a camera crops as the resolution changes, which many IP cams do. There's an argument to be made for either - interpolation lets you keep the same field of view as you change resolution, while cropping gives a more precise image - 1 sensor pixel equals 1 image pixel. If the sensor has the right number of elements, an interpolated image can actually be an improvement. For instance, a 1/4 resolution image can be made by combining 2x2 elements into a single pixel, which gives you better low-light performance as you drop the resolution, but if your elements aren't a multiple of the final resolution, you can end up with aliasing errors (http://en.wikipedia.org/wiki/Aliasing), moire patterns, image distortion, etc. Software can include anti-aliasing filters, which require more CPU power and costs more, so it's a trade-off, as usual. It'll be interesting to see where this all falls out. Q, if you get a chance to print an ISO 12233 test chart, tape it to the front of the monitor, and take the same 3 snaps, that would be illuminating. That's assuming you want to keep tilting at this windmill, of course! http://www.graphics.cornell.edu/~westin/misc/ISO_12233-reschart.pdf (from http://www.graphics.cornell.edu/~westin/misc/res-chart.html)

Yeah, that's why I checked it, because many packages don't. I have a Messoa that only shows 4CIF resolution on the web page and I needed a screen shot, so I took it from BI, but had to verify that it was correct. So, it looks like the main difference between the 1.3MP and 3MP versions of this cam are the sensor size. Hik doesn't actually offer a 1080p version that I'm aware of. DS-2CD2032-I (3MP) - 1/2.5" sensor = 0.4" diagonal DS-2CD2012-I (1.3MP) - 1/3" sensor = 0.33" diagonal There is no 2022 version listed. This is assuming they're not being imprecise with their specs, which many vendors do. Both of these sensors have 4:3 format at max resolution, so if the Swann is truly hardware limited to 1080p, it'll have a 16:9 sensor, which is much more rectangular than the traditional old-school 4:3 almost-square sensor. This still doesn't prove anything, but if the Swann has a 4:3 looking sensor instead of a 16:9 looking sensor, and truly does 1080p, then it would stand to reason that it's a 3MP capable sensor.

Both of the snaps posted are in the correct resolutions for the settings chosen (3MP and 1080p), so it appears this is what the camera's putting out. The question is what resolution the sensor output is before the firmware adjusts it to these resolutions. One or the other of those snaps is not at native sensor resolution. Blue Iris captures snapshots at the video resolution. It also won't allow you to set a different resolution than the incoming video; if you select something different, it switches back to the native video resolution when you click OK. I took a few unrelated snaps from BI the other night and verified this. Bitrate and fps won't make a difference in still captures. This isn't actually an easy thing to figure out without access to sensor details. Q, can you post a snap of the 720p resolution? Are there only 3 resolutions available, as it appears from the firmware shots earlier?

Does the Hikvision web interface have a snapshot button? I like to use a camera's web interface for snapshots, to minimize the chance that the NVR software is doing something with the image. Nevermind, I see from an earlier post that you did capture from the web page. As long as those stills are the same resolution as the BI stills, they should be comparable.