Home > TechNews > Megapixel Nonsense

Megapixel Nonsense

I’m not sure I get the whole megapixel craze sweeping the IP camera market.  Maybe it’s not a craze.  Maybe it’s a few very visible marketing initiatives.  Don’t get me wrong.  I LOVE the way megapixel images look.  For watching live video feeds, nothing beats a megapixel camera.  But is it really useful?

Last time I checked, DVRs could not record megapixel feeds without dumbing down the image with lossy compression.  (Maybe someone can educate me).  It seems to me that if the camera is going to be deployed in a way that the images may be needed for forensic analysis, say, for checking identities of bad guys in a casino, or shoplifters at Target, then recording is a must. 

I have no idea how much it would cost, but the Avigilon system combines an 11 megapixel IP camera with a high capacity storage array designed to store lossless hi-res video (transmission over gigabit ethernet only).  It looks great.  But it’s the only system I’ve seen that preserves the hi-def image for forensic analysis.

CoVi always has an impressive display at the trade shows with its high definition video system.  As I understand it, the CoVi Crystal HD IP camera plugs into a CoVi recorder (blade or appliance) that can support a couple IP or analog cameras at a time.  It sounds like it would work well if the recorders were distributed wherever the cameras are.  That way, you’d still get high resolution from your HD camera, and very high quality from your analog camera.  But once you needed to view that camera or several others from a distance over coax, you’d have to lose image quality.  The coax simply couldn’t move that much data, it seems to me.  Gigabit ethernet would certainly help, but the CoVi specs on the website don’t mention gigabit. (To be sure, the CoVi solution is superior to regular camera deployments for many specialized uses).

ioimage just announced its newest encoder.  The specs look as if this is the lowest latency encoder on the market.  So I guess it will be particularly good for live PTZ response.  Has anyone tried it yet?  I’ll see if I can get my hands on one and test it out.  Comments?

Categories: TechNews
  1. Anonymous
    January 18, 2007 at 8:35 am

    Your review of megapixel cameras assumes that all pixels are equal and so doubling the number of pixels requires doubling the storage. While that is how typical DVDs work, human perception treats the “interesting” parts of a scene differently from the boring background part of the scene. Adding intelligence to the system before the storage allows selection of the interesting objects for high resolution recording while making only periodic updates to the background that doesn’t change. This concept allows high resolution images of all of the people without burning through the hard drive recording the same brick wall 30 times a second. (Search the USPTO site for 20060165386)
    Disclosure notice: I’m the inventor for that patent and the concept was implemented in Cernium’s Perceptrak several years ago.
    Maurice Garoutte

  2. January 18, 2007 at 12:04 pm

    Thanks, Maurice. I’ve seen some attempts at applying intelligence or an algorithm to video transmission. CoVi does something like that. You are telling me that Cernium does. So why do I still hear that megapixel images are degraded when viewed remotely or recorded?

  3. Anonymous
    January 18, 2007 at 12:39 pm

    To get the terms straight, any compression involves the use of an algorithm to reduce the storage/bandwidth. That’s not unique to Cernium. Spatial compression schemes such as JPEG compress every frame independently and don’t benefit from constant backgrounds. Temporal schemes such as MPEGX take into account any changes in pixel values between frames so can reduce the storage in video where some of the scene is constant. Even with the best compression, the largest disks, and biggest network pipes there remains a compromise between resolution and storage time or frame rate over the link. Most designers make the compromise in the direction to increase sales, often lower recorded resolution for longer advertised storage capacity. When pipes and disks get very big, or users are happy with ones of hours of stored video, then video from megapixel cameras will look great.
    Object Selective Recording is just the next step in the progression. Video is analyzed to determine and classify objects and spatial compression is used to compress all objects. Periodically (the temporal part) the background is heavily compressed and stored. I was never very good at compromise so the original design of Object Selective Recording left resolution and frame rate as user configurable.

  4. January 18, 2007 at 2:00 pm

    CoVi was founded to bring the benefit of High Definition video to the surveillance industry. Your recognition of our distributed architecture as the solution is correct, but you have missed the point of difference between our solution and the rest of the ‘megapixel’ industry. As you noted, streaming giant firehoses of megapixel video over IP networks is a trainwreck, and essentially useless in a surveillance context. By distributing the storage, we are able to capture and store HD video with NO burden on the network, no matter how many cameras are in the system. In fact if the network goes down, we are still recording. Further, we record up to six streams simultaneously in a variety of resolutions and frame rates, so that we supply only the amount of data that the user needs and can accomodate. In our solution, one need only pull video across the network for viewing. Therefore, the maxium load per viewer is 12 megabits per second. This compares to 15 to 48 Mbps per camera for the average ‘streaming’ HD camera. EFFECTIVE Video is the goal. We deliver.

  5. Steve Surfaro
    January 19, 2007 at 4:30 am

    Megapixel nonsense? Interesting point, Steve. You actually make a good one, although in a different way for both for the security and professional digital photography industries. When we purchase a consumer digital camera, we look first at the # of pixels, the brand, the reviews and plunk out the wallet. I have both Nikon D2h and D200, which differ in pixel quantity by almost 6 million. Why, then, do they take (close to) the same picture for a substantial size like 8×10? Yes, there are subtle differences, but they use different imaging tech to achieve the same results. Pixel pitch, density and the space between the pixels and other factors play a role in resolving images.
    Similarly, in our industry, it is not necessarily the number of pixels that will get you a great image.
    Also, the lens plays a great role in resolving images consistently, with or without distortion.
    I have seen a dangerous trend where we forget about the lens and some companies are including optics that just barely match their imager’s performance.
    My favorite Nikon lens is a 17-35mm f/2.8, which works well with both 35mm and digital imagers. A dual performer has to be incredibily consistent to do this! I can only hope that users will rediscover the benefits of great lensing…

  6. Wes
    January 19, 2007 at 6:06 am

    Most installations of mega pixel cameras I have dealt with require high quality video to be recorded so if they need to backtrack and view recorded video, they have a high quality image to zoom in on the perpetrators face or check out the cars license plate. In situations when you can’t watch the video live and zoom in on the action, mega pixel works great. As previous posts pointed out, bandwidth issues can come from the high amount of data transmitted via ethernet.
    Most installations I deal with are just recording to a PC, I don’t deal much with DVR’s so my view is a bit one sided ;).

  7. January 23, 2007 at 7:50 am

    We have been manufacturing IQeye Megapixel Network cameras since 2001 and at least 90% are used for their forensic value, not live viewing. They are not recorded with DVRs since most DVRs are limited to low resolution like NTSC. You record Megapixel Network cameras using low cost network storage and Network Video Recording (NVR) Software.
    As for cost, the only fair way to compare low resolution CCTV/DVR systems to a megapixel based system is to look at quality from a pixels/sq. ft. perspective. NTSC resolution is only 1/3 of a megapixel (today’s cheap cell phones have more resolution than that) so if all things are equal, you would need 10 CCTV cameras to cover the same area as a 3 MPix camera. Granted a 3MPix image is large but not 10x the NTSC images so installation costs, storage costs and bandwidth costs will be lower with a megapixel based system.

  8. February 23, 2007 at 6:19 pm

    Contrary to your assumption, Steve, coaxial cable CAN move that much data. It is actually easily capable of transmitting multiple mega-pixel cameras over a single cable. We provide a product called HIGHWIRE which is connected at each end of an existing coax cable. It provides totally transparent 100BaseT Ethernet over up to 1200ft of coax. Most of the mega-pixel camera manufacturers (including those mentioned here) are familiar with it and it is now widely used by integrators in real-world projects.
    Mega-pixel surveillance will dominate the industry within a few years. Mega-pixel resolution at last provides the compelling reason to switch to IP video which has previously been lacking. The problems of high-bandwidth transmission over typical analog camera cable distances have largely been solved using HIGHWIRE. We even have customers fitting NEW coaxial cable in new installations just to get the transmission distance they need for their IP cameras, as standard CAT5 and Ethernet is limited to 100m between devices. Several of the mega-pixel manufacturers are successfully running up to four 3Megapixel cameras down a single coax cable. RG59 cable will allow about 850 ft at full data rate with HIGHWIRE, or 1200ft at a slightly reduced rate. RG11 cable can allow it to go up to 1600ft or thereabouts.
    So in conclusion, good old coaxial cable has a high bandwidth capability – so don’t go ripping out your old coax cable – just exploit it !

  9. February 25, 2007 at 8:40 am

    Thanks Alastair. And thanks Paul, Steve, Maurice and Barry. Your comments are appreciated.
    Let’s keep this discussion going on the Dreamer Forum. click on the link at the top of this page!

  10. February 25, 2007 at 8:41 am

    …and Wes. 🙂

  1. No trackbacks yet.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: