Compression

[Michelangelo 0]

Hi guys. My first post, here it goes.

 

Can anyone explain to me how a megapixelcam in 4CIF can show a better image than an analog 540 TVL camera? Doesn´t the compression take away the exessive pixels so that the image should be basicly the same?

[survtech 0]

Hi guys. My first post, here it goes.

 

Can anyone explain to me how a megapixelcam in 4CIF can show a better image than an analog 540 TVL camera? Doesn´t the compression take away the exessive pixels so that the image should be basicly the same?

It shouldn't. 4CIF is 640x480 (307k pixels), while analog color is 768x494 (380k pixels). Megapixel resolution should be at least 1024x768 (3/4 megapixel) or 1280x1024 (1-1/4 megapixel). Anything less wouldn't really take advantage of all of the pixels.

[rory 0]

What he said, and 4 CIF is not megapixel.

 

If you are viewing an analog camera on an LCD Monitor, such as from a stand alone DVR, or a Composite to VGA converter, then the quality will be much lower than it coming from a PC Card. Perhaps that is what you meant?

[photys 0]

Because most IP cameras are progressive scan, analogue cams are Interlaced.

In theory, IP cameras start with double the quality before compression.

 

 

.....kk, tear my theory down.

[survtech 0]

Because most IP cameras are progressive scan, analogue cams are Interlaced.

In theory, IP cameras start with double the quality before compression.

 

 

.....kk, tear my theory down.

Why is that? If looking at a static, unmoving object, there should be no difference between an interlaced and a noninterlaced picture of the same resolution. When motion is introduced, that is where you would see a difference since there would be different data on successive fields.

[CollinR 0]

No you still have only half the lines from each frame, motion just magnifies it. Progressive is superior, how much so is very debatable.

 

The real issue here is megapixel is not 4CIF unless you got screwed by the DVR salesman.

[rory 0]

Yep, just to confirm, 4 CIF is VGA, which is 640x480 NTSC.

[survtech 0]

No you still have only half the lines from each frame, motion just magnifies it. Progressive is superior, how much so is very debatable.

 

The real issue here is megapixel is not 4CIF unless you got screwed by the DVR salesman.

Not at all. 480 lines is 480 lines. It doesn't matter if there is a 1/60th second delay between the even and the odd lines (interlaced) or not. there are still 480 individual lines of information.

 

What you are saying is that the information in successive fields is the same. It is not, even in a static image. By your definition, there is only 240 lines of information in each frame; each repeated once. That is not the case.

 

The sensor chip of an analog delivers 480 (actually 494) independent lines of data. With interlacing, the camera delivers the odd lines on one field and the other even lines in the next field. The two fields combine to form one 480-line frame. A non-interlaced, or progressive scan camera delivers all 480 lines sequentially.

 

The two technologies deliver the same number of lines (480) of distinctly different data; usually at the same number of frames (30) per second.

[CollinR 0]

The fact the two fields occur in different time is less then desirable, I have no (0) customers who bought my services expecting to capture a static image. They wanna see motion, they wanna see guy using trash bin or girl cheating or whatever but it moves.

[survtech 0]

Nevertheless, it is a fact of life with analog. That's why analog interlaced cameras work best with analog interlaced monitors. The 1/60th second delay between fields is partially compensated for by the 1/60th delay between the scanning fields of a CRT display.

[photys 0]

Can someone explain this?

 

Geovision units.

Recording at 640x480 no de-interlace gives a far superior static image compared to 640x480 de-interlaced.

Is it because the geovision units only show half the lines per frame?

[Michelangelo 0]

Ok, but besides moving objects, does the megapixel cam provide better image quality? What I have seen most NVR also record in 4CIF or D1 at the highest. Right?

[survtech 0]

Can someone explain this?

 

Geovision units.

Recording at 640x480 no de-interlace gives a far superior static image compared to 640x480 de-interlaced.

Is it because the geovision units only show half the lines per frame?

The problem is that the deinterlacing process is lossy. Because of the 1/60 second difference between fields, the deinterlacing device tries to extrapolate the position of objects from field to field and reconstruct that into one non-interlaced frame. Even on static images, where there is no movement to correct, this is a hit-or-miss proposition.

[survtech 0]

Ok, but besides moving objects, does the megapixel cam provide better image quality? What I have seen most NVR also record in 4CIF or D1 at the highest. Right?

A megapixel camera requires higher resolution and bit rates to record its signal properly. Usually that means recording the digital signal directly to a hard drive, either with or without transcoding (converting from one compression codec, say MJPEG, to another, say MPEG4). Only a few DVRs have the capability to do this. True NVRs, on the other hand, are more often able to record the megapixel camera's data directly, at higher resolutions: 1024x768, 1280x1024, 1600x1200, etc.

 

If the digital signal from a megapixel camera has to be converted to analog to record, as with most DVRs and NVRs, the higher resolution of the megapixel is wasted and the result is no better than an analog camera would deliver - 4CIF (640x480) or D1 (720x480).

[CollinR 0]

The problem is that the deinterlacing process is lossy. Because of the 1/60 second difference between fields, the deinterlacing device tries to extrapolate the position of objects from field to field

Because the interlace removed them!

and reconstruct that into one non-interlaced frame. Even on static images, where there is no movement to correct, this is a hit-or-miss proposition.

 

 

No the problem is the interlacing process is lossy. That was the point of doing it, to reduce bandwidth back in the 1930s. This is 1930s technology we are still having to tolerate.

 

Some deinterlacing engines do a much better job then others, either way though when it gets to a component, RGBHV or digital interface display it must be deinterlaced. When you see a difference between Geovision capturing and deinterlacing the capture vs not deinterlacing it internally what you are really comparing is the display adapter's ablity to deinterlace vs Geovisions. If you use nice display adapters they may well have hardware support for deinterlacing and thats usuallt higher quality. If you capture and then playback on an old W95 machine with on board graphics it will look as saw toothed as a modern $150 Chinese DVR.

 

 

Good reading.

http://www.100fps.com/

[survtech 0]

Semantics.

 

Still, the deinterlacing process itself is lossy. When combined, the 60 alternating fields that make up a frame of video in interlaced analog can contain as much information as an equivalent progressive scan 30fps camera with the same pixel count. The primary difference is that it takes 2 scans to fill in the complete picture when interlaced versus only one scan with non-interlaced video.

 

Still, megapixel technology can not reach its full capability until the signal maintains its full digital attributes through the whole chain from camera, through recorder, to display.