What Advantages Still Left For IP Megapixel ?

In the current market where HD Analog (be it AHD, HD-CVI, HD-TVI or else) beats IP Megapixel on price and many features, what benefits are still left for choosing an IP Megapixel solution ??

- Cat.5e/6 Cheaper Cabling: one can use $7 - $10 bucks pretty-good quality video baluns with RJ-45 jacks built-in to transmit Analog HD over network cabling without any problem (forgoing coaxial RG59/6 completely). For short distances (200 ft./60 Meters) one can just buy a $22 bucks siamese cable for video + power and forget about the Cat.5 entirely. No need for PoE using siamese cables.

- CPU and OS Licenses: Genuine Intel processors and Microsoft Windows licenses take a huge chunk of the cost of an IP NVR. SD Analog had managed to work since forever with Linux and Embedded processors. Of course HD Analog has followed the same course without needing Intel or Microsoft.

- Software features: honestly in my years of experience, operators in their daily duties don't use much more than: looking at Live Video and Search for recorded events. The FREE VMS
available to manage standalone HD Analog DVRs do this perfectly of course. Video analytics on IP Megapixel VMSes, I still haven't seen users exploiting this. Some users are even scared of them. A half-baked video analytic software is worst than having none I think.

- Special features like FishEye panoramic cameras dewarping: even though it gives users a real panorama view with a single camera replacing 3 or 4 HD Analog fixed cameras, many customers when they crunch the numbers, find that even putting 6 HD Analog fixed ones is even cheaper than a single FishEye.

- And let's not even talk about IP technology's unavoidable TCP/IP latency even on a local LAN without any routing to the outer Internet. Analog clearly wins on this one with less latency and real time video.

Maybe the only 2 things that comes to mind where IP Megapixel wins is MORE STORAGE space available under "one roof so to speak" and catering to the IT crowd:

A customer could still argue that stacking 4 standalone Analog HD DVRs on their server rack with 4x SATA HDD drive bays inside each one; and still get it way cheaper than a SINGLE 4U Rackmount NVR with 16 SATA HDD Drive Bays. If analog HD DVRs include hot-swappable features in the future, that would be the final nail in the coffin for IP Megapixel.

Catering to the IT crowd (which many of these guys see the standalone DVRs as cheap un-sexy boxes) may be favorable, but is not really an advantage when they have to justify with numbers to their upper management the extra cost of IP Megapixel. I've yet to see an executive approve a budget on "subjective feelings" from the IT crowd rather than hard numbers. In my experience, management only see the objective numbers and if the system can do what it does, specially in tough economic times with budget cuts here and there.


Are things that bad? :)

I think HD analog is doing very well (e.g.HD Analog Usage Statistics), but there's definitely still a number of advantages for IP.

(1) Form factor / feature options: Not just fisheye, but minidomes, smart IR, covert, true WDR, super low light, etc. For professional applications, these are important.

(2) 3MP, 5MP, 8K, 12MP, etc. Many (you can debate how many) want more than 1080p.

(3) PoE is attractive to many.

(4) VMS Software: This is the one point I disagree with you significantly on: "The FREE VMS
available to manage standalone HD Analog DVRs do this perfectly of course." To the contrary, I find most HD analog DVRs have limited functionality and poor usability.

(5) Lots of big brands still do not offer HD analog. To name a few, Axis, Bosch, Pelco, Panasonic, Sony. This is not a technical element but has real market force, as their large sales and marketing budgets impact many buying decisions.

To name a few, Axis, Bosch, Pelco, Samsung, Sony.

I meant to say Panasonic, corrected.

Hi John,

Last Friday I had a presentation on the new HD-CVI products coming out from Dahua in the next 6 months. They seem to address your points 1, 2 and 3 directly:

(1) Form factor / feature options: They are introducing every single form factor that is offered in IP (and they have a LOT of IP form factors!) - mini-ptz, full-sized PTZ, mini-domes, WDR, etc... the HD-CVI catalogue will be as thick as the IP catalogue.

(2) 3MP, 5MP, 8K, 12MP, etc. : they showed me a beta release of 2K (4MP) HD-CVI (released before the end of Q4 2015) and showed me the roadmap for 4K HD-CVI (that they have at 12 MP) which will be avilable before the end of Q1 2016.

(3) PoE is attractive to many. They showed me, from their roadmap, power-over-coax for HD-CVI. 1 wire to do it all...

They seem to be going all-out on this technology.

4K over coax! At 1/2 the price of IP...

I am sure they are working on this, and at some point I expect many of those to be available. I severely doubt all of this will be in production in 6 months.

That being said, to the extent they can deliver this it will severely hurt the case for IP. However, I am simply answering the question for what's available now.

Software features: honestly in my years of experience, operators in their daily duties don't use much more than: looking at Live Video and Search for recorded events. The FREE VMS available to manage standalone HD Analog DVRs do this perfectly of course. Video analytics on IP Megapixel VMSes, I still haven't seen users exploiting this. Some users are even scared of them. A half-baked video analytic software is worst than having none I think.

Yea I don't think so. The FREE VMS are really bad when it comes to speed and usability.

Do you agree it is a matter of time ? as in HIKVISION software used to be crap, now the latest iVMS4500 mobile app has improved a great deal and their iVMS 4200 desktop software as well. HIKVISION and DAHUA certainly have the financial resources to do so ...

Btw, last Spring Hikvision internationally released its next gen offering, IVMS 5200, (excluding the U.S.).

This is a video they made even before that. Emphasis on analytics.

Yea I have been testing 5200 pro for a couple weeks now and its better then 4200 but still not fast or intuitive. One thing that drives me nuts is you can't manage the system from the client software. There is a separate web GUI to admin the system which just adds unneeded complexity IMO. Most customers are not a fan of this either from what I have seen.

I was also advised by Hik AU, that it's a VMS for Hik products. Don't have a camera with edge analytics? Too bad so sad, you can't use it with the VMS if you want anything other than record liveview and playback.

I get a system to play with later today.

I think you can summarize most of the IP vs. Coax-HD solutions as "choices".

The coax-HD stuff is still very much segregated into incompatable technology stacks. IP is, for the most part, interoperable (eg: ONVIF, RTSP). It's not perfect and there are edge-cases like Genetec not supporting Avigilon cameras, but for the most part all the mainstream products can work together.

With IP cameras you have many more choices for form factors, resolutions, and camera options (audio I/O, digital I/O's, SD-card backup storage, etc.).

The latency difference has never been a selling point, IMO. Very little video is observed live, most of the time users are accessing recorded video where the latency is no longer an issue. In cases of live video, we're generally talking 1 second or less. This would be important if we were doing remote robotic surgery with IP cameras, but not for typical applications. Also, if the user is not using a monitor/keyboard//mouse directly attached to the DVR they will experience that same network latency. So, it's really not a univeral advantage.

Coax-HD is, IMO, optimal when your customer is looking for a simple no-frills budget system. It's easy to install, easy to use (generally because of lack of advanced features), and lower-cost. Most of the mainstream players in the IP side of things hadn't been particularly focused on those smaller cost-driven systems so it's a nice alternative.

Also, if the user is not using a monitor/keyboard//mouse directly attached to the DVR they will experience that same network latency.

Still, wouldn't the Analog HD system have one less network hop, so slightly less network latency?

Of course not so pronounced because of the additional encoding latency, but...

IMO, this doesn't matter, especially not in the market where the coax-HD systems are being used.

We're not talking about 5 or 10 seconds of latency difference, it's usually 1 second or less. This matters if you're doing lots of real-time PTZ controlling, like maybe for a gambling vertical market. But these systems are still WAY too limited to get any significant penetration in that kind of a market.

For the applications where these products are being used, I think anything less than 3 seconds is effectively "real time". If someone is monitoring the video live and looking for a particular action or object you still have some lag for the operator to *do* anything, or to contact another person. If the application is so critical that 3 seconds is too much lag, the customer is most likely not going to rely on what is right now mostly Chinese import product with limited direct support (at least in the US).

And let's not even talk about IP technology's unavoidable TCP/IP latency even on a local LAN without any routing to the outer Internet. Analog clearly wins on this one with less latency and real time video

Do you have any documentation or testing that backs this claim up? Latency on a local Lan is generally 1-2 ms with modern gear. That is literally one thousandth of a second.

Testing IP Camera Latency

In a simple, lightly loaded local system, latency from IP camera to VMS typically were 300 - 400ms.

To be sure this is primarily due to the added time from camera-side h.264 encoding + client-side h.264 decoding, not the LAN overhead itself. But they are unavoidable delays in IP surveillance camera systems.

Both of these though can be eliminated when monitoring live on the machine where the Analog HD transcievers exist.

I do. Here is a real installation. See video here.

Time/Action/Channel:

- 14:34:37 / Vehicle barrier lifts up / Upper Right (UR)
- 14:34:42 / Vehicle barrier lifts up (but 5 seconds later) / Lower Left (LL)
- 14:34:43 / Vehicle starts moving / UR
- 14:34:48 / Vehicle starts moving (but 5 seconds later) / LL
- 14:34:57 / Security guard closing the gate / UR ... but the vehicle is still crossing on LL as you can see ...

And all of this footage was extracted locally with my laptop connected to the local Gigabit 10/100/1000Mbps switch with a patch cord. NOT remotely, NOT via Internet,
none of that.

So, my question is: Why so many seconds of latency still on a local LAN without further routing/switching ??

What system are you using? Some systems are worse than others so it's hard to take one example of an unknown system and apply it universally across a whole category.


... using an Intel Core i7 with 16GB RAM Hot Swap 20-HDD Bays Rackmount-4U NVR with a 128-Channels VMS which comes with an integrated Apache Server service; all of them connected to an ALL Gigabit 24-Ports PoE switch.

IMO, you're seeing software lag from the DVR to the laptop, not the camera to the DVR. Could be several reasons, but most likely revolves around lack of processing resources on the DVR itself.

Lack of resources ? ... it already has a Quad Core i7, 16GB RAM and plenty of hard disks with sufficient swap file. Are you implying that upgrading to a Xeon-based server with Windows Server 2012 OS would be better for reducing latency ??

I'm not implying anything, your video and post really don't provide nearly enough data to make any concrete recommendations.

I am simply saying that my personal opinion is that what you are seeing is more of a DVR to Client PC problem then a Camera to DVR problem.

The network will add latency up to a point (typically 1s or less), but at some point excessive network latency will inevitably cause lost packets, giving you choppy or pixeled video or other problems. In your case the video looks excellent in terms of general sharpness and continuity of frame rate. That tells me the video is arriving to the DVR fine, but something in the DVR and/or PC client component is adding in the delay you are seeing.

I have been experimenting with a number of IP based solutions for live view, all, including my beast of a desktop, have a lag even on the local network. I initially

You have to look at a monitor connected directly to the NVR to get the true live view.

I think integrator 2 hits it on the head. You are looking at encoding/buffering delays from your gear and not delays from the underlying network. Think about Skype video conferencing. It certainly doesn't take 10 seconds for video to reach the person on the other end, and those calls can be across the country.

Recommendations for reducing these delays ??? Change the VMS ?... Change video card, CPU ?? Still add more RAM ??

Hmmm...

...latency...

..... latency.....

........ latency ........

Todd, is that you? Seriously, I think I've seen Rocky use the word "latency" ten times as much as everyone else combined, ever.

Like the others have said, there's not enough info given for an accurate diagnosis of the problem, but offhand I'd start with the fact it looks like an old GeoVision system - not what I'd call a shining example of a high-performance VMS, regardless of what overblown hardware it's running on. Clunky software will always be clunky even on the best gear. Probably 32-bit software, and I wouldn't be shocked if it's running on 32-bit Windows, meaning you're only using about 3.5GB of that 16GB RAM.

Second, this is obviously playback of recorded video, and looks to me like just poor playback sync, which could be due to any number of causes. Others have suggested server-to-client issues, but I suspect the system is just having problems pulling all three streams in proper sync. I don't care how super-special the hardware is, if your VMS has poor file management or you're using a lousy file system, the VMS WILL have trouble pulling multiple streams simultaneously, especially if it's still recording at the same time. Every VMS I've ever used has had issues with multi-stream playback sync at one time or another.

Also on the hardware performance front, if you're playing this back on your laptop, the power of the NVR itself is almost irrelevant - all it's doing is pulling the stream from storage and sending it to the client; the actual decoding of the video is being done by your laptop. If your laptop isn't powerful enough to process all three simultaneous streams, well... that could be your problem as well.

Is it possible there's massive latency between the cameras themselves and all three were actually recorded with that much lag between them? Sure, anything is possible... I'd call it highly unlikely, though. Unfortunately none of the cameras are showing their own timestamps, so there's no way to know.

Recommendations for reducing these delays ??? Change the VMS ?... Change video card, CPU ?? Still add more RAM ??

As noted, it's impossible to determine how to fix the problem, without first determining where the problem lies, and it's impossible to do that without more information. For that matter, one needs to determine, IS there actually a problem? If the NVR is recording everything properly, and you can export video clips from it directly and THOSE are all in sync, then there really isn't a problem except with your perception - the video LOOKS out of sync when you're playing back on your laptop, but it ISN'T really.

Of your suggestions here, I think the most helpful would probably be to change the VMS. Your hardware is already at the overkill stage, and again, unless you're running a 64-bit OS, you're not even using its full power.

Recommendations for reducing these delays ??? Change the VMS ?... Change video card, CPU ?? Still add more RAM ??

The problem actually shown, out of sync playback of multiple recorded streams, can be fixed by by synchronizing the individual camera's time to that of the recorder.

Also make sure that the VMS is set to timestamp the incoming video based on the camera's timestamp, not on arrival (or get one that allows this), as discussed here. Then even if there is a 5 second latency, everything will playback seamlessly. And it is always easier to watch, even when the latency is minimal.

If your beef is really about live video though, you'll need to show a live clip.

I had similar problem. Solution was to ad a management gigabyte switch and to create a hardware separate network for outgoing connection to clients.

Scalability is the biggest advantage along with resolutions beyond 1080p. A multi-building facility would be ill-served with 15 different DVRs in random closets. The SMB business should benefit greatly from the various analog 1080p options, but I just cannot see enterprise accounts adopting it without additional maturing of the product. Encoders in order to tie these cameras into more advanced VMS is one step that needs to occur soon.