Quality of recorded images

[chewingyu 0]

Can someone share on what factors determine the quality of recorded images? How does one define the quality? Is CIF specification alone sufficient? Or monitor resolution?

 

I am asking because I had tried changing the configuration of the DVR setting for some of my CCTV cameras to 4CIF and at 'Best' quality, yet I do not see significant improvement in quality in the recorded images.

 

Anyway, what is the resolution of CIF and what does it stand for?

[chewingyu 0]

Came across this website. ttp://www.aventuratechnologies.com/dvr_tutorial/default.asp?index=11

 

Looks like a credible explanation. Anyone having different views?

[scorpion 0]

http://scorpiontheater.com/lab.aspx

 

CIF

http://en.wikipedia.org/wiki/Common_Intermediate_Format

 

Resolution

http://en.wikipedia.org/wiki/List_of_common_resolutions

 

http://en.wikipedia.org/wiki/Display_resolution

 

 

 

http://en.wikipedia.org/wiki/Lens_makers_equation

http://en.wikipedia.org/wiki/Optical_coating

http://en.wikipedia.org/wiki/Back_focal_plane#Focal_points_and_planes

http://en.wikipedia.org/wiki/F-number

http://en.wikipedia.org/wiki/Lux

http://en.wikipedia.org/wiki/Fabrication_and_testing_of_optical_components

http://www.mrpinhole.com/calcpinh.php

 

 

__________________________________________________________

 

Wow! We are opening up a can of worms with this topic!!

 

I had a guy yell at me about a bad DVR that I sold him. He told me what he wanted, and I delivered big time!

 

He was mad because it did not look like watching a DVD!

I showed him that you can count every leaf on a tree if this was outside. He needed good facial recognition as this was in a hallway of surgical area of a hospital.

 

He did not like the snap shot effect!

 

There are too many variables with camera selection, DVR selection, wiring selection, lens selection, compression format, and budget issues.

 

I do not read the specs of products. I just plug them in. If it works, then it works, if it does not then it does not. Some equipment may suck in one environment, and sail in another environment. Interesting!

[Soundy 1]

Came across this website. ttp://www.aventuratechnologies.com/dvr_tutorial/default.asp?index=11

 

Looks like a credible explanation. Anyone having different views?

 

That covers it pretty well, I'd say... but as Scorpion's other links show, there's a lot more involved in the final quality of the video, primarily the type and level of compression used, and the quality of the camera/lens system itself. Scorpion's hospital client, for example, should probably keep in mind that the DVD's he's thinking of were probably created using cameras worth at least a hundred-grand *each*, as well extensive post-production image processing... while he's probably using cameras and lenses that combined are worth maybe $300 each.

 

As to adjusting your DVR settings, try looking for settings for compression type and amount (without knowing exactly what DVR is it, it's hard to say where to look).

[normicgander 0]

Seems we have been down this road before...

 

720x480, CIF, 2CIF, 4CIF, D1 or whatever means very little. The problem with our industry or perhaps the weak link is with DVR OEMs. They have failed to address poor image quality with too much image/video compression. They need to incorporate dual codecs, namely one for recording and one for network streaming. This way you can load up the DVR with TBs of low-cost capacity to record higher quality images, while remotely viewing lower quality video via the network. Of course the nework parameters should be adjustable based on bandwith.

 

TeleEye seems to have the right approach in this regard, but they could also reduce video compression on the record channel.

 

Then there is the issue of recording two fields when there is motion in the scene. Progressive scan cameras (i.e. Dallmeier) with intrepolated segmented fields for compatibility with typical analog input DVRs is required.

[Thomas 0]

Seems we have been down this road before...

 

720x480, CIF, 2CIF, 4CIF, D1 or whatever means very little. The problem with our industry or perhaps the weak link is with DVR OEMs. They have failed to address poor image quality with too much image/video compression. They need to incorporate dual codecs, namely one for recording and one for network streaming. This way you can load up the DVR with TBs of low-cost capacity to record higher quality images, while remotely viewing lower quality video via the network. Of course the nework parameters should be adjustable based on bandwith.

 

TeleEye seems to have the right approach in this regard, but they could also reduce video compression on the record channel.

 

Then there is the issue of recording two fields when there is motion in the scene. Progressive scan cameras (i.e. Dallmeier) with intrepolated segmented fields for compatibility with typical analog input DVRs is required.

 

 

Installers and customers want to cram as many cameras on a machine as they can with the least amount of processing power used. Running two encoders is a huge processor hit. Plus end users (and to a limited degree installers) believe that the ultimate codec is the ultimate codec, not understanding that different codecs do different things well. And people don't want to pay for storage.

[normicgander 0]

We are starting to see dual codec DVRs and storage is much cheaper. Once we used see 160 to 320gb HDs in stand alone DVR offerings and now 1T is not unusual. So my point is reduce the level of compression to increase image quality because larger HDs are now available...

[Thomas 0]

We are starting to see dual codec DVRs and storage is much cheaper. Once we used see 160 to 320gb HDs in stand alone DVR offerings and now 1T is not unusual. So my point is reduce the level of compression to increase image quality because larger HDs are now available...

 

And at the same time the demands for length of storage went up with it. Five years ago, people were pretty happy with 30 days. Now I read RFPs for 90 days or 120 days storage requirements.

 

And in the end there is a cost issue. Using two codecs at the same time has a major processing cost. This means better processing power. In some cases more ram. It requires more coding. All of which lead to significantly increased costs. And the fastest way to make a dealer howl is increase their costs. We do it on the IP side with our products but there is room for it in the costs. On the Analog side, price is extremely important.

[normicgander 0]

I would not discount your assertions Thomas. However, we have been using TeleEye DVRs with multi-steam video coders with success. Compared to the cost of GE DVMRe series we used to install, the TeleEye DVRs are significantly lower in cost and we can support the HDs etc....

[Soundy 1]

Put it all together, and you get the same thing: it's ultimately a balancing act between quality, retention, and price. Or to coin another old phrase: "good, quick, and cheap; choose two."

 

People want to be able to buy an eight-camera DVR system for $600 from Costco, and have it do all the same things they see in the movies and on CSI - go back to six months ago, find a clip of the bad guy taken from behind in the dead of night, and draw out a perfect image of his face from his reflection in a door knob. You can't always make them understand why that's just not possible.

 

Storage may be relatively cheap these days, but when you're looking at the need for 10TB or more, it can add up pretty fast at ~$200/TB. Add to that the need for some sort of enclosure and interface for all those drives, and the cost spirals. It's easy to say "space is cheap", but when you get to that level, it can start ramping up pretty fast.

 

And then there was the old Diebold DVR we pulled out of a site a couple years ago... built on a P166 with a whopping 128MB RAM, running NT 4.0 and a plethora of custom frame-grabber and other cards. Thing took a JPEG still something like once per second, but only one camera at a time, and only at 1CIF... and had over *six months* of footage stored on an 80GB drive. Quality, though, left something to be desired - it was about comparable to VHS at 24-hour time-lapse speed.

[normicgander 0]

Good points Soundy. I'm still searching for that near perfect DVR. I tend to lean toward image quality first and archiving, because obtaining a good forensic snapshot of a subject of interest is critical. We need to educate the customer that 2 weeks of quality images is better than 2 months of low-quality "stuff." I would rather not deal with the future problems and disappointed customers. Catching a nice batch of perch through the ice can be more productive .

 

We all see the poor quality images and video clips on TV. I have always recommended recording at the highest-quality settings, but even the $6000 DVRs come up short at times.

 

Then there's the deinterlace issues of 2 fields which most DVRS don't even attempt to process. You will see more analog progressive scan cameras which will solve this problem, resulting in better image quality.

 

The question is what can we do, besides progressive scan IP cameras? Are the ADC chip sets used in DVRs bounding quality as well or is it just over compression?