help me understand this camera's compression

B

blazingoat

Carbonite Member
Let me see if i've got this straight. This camera shoots at 1080 resolution. But it's still 4:1:1 compression, and then mpeg 2 compression on top of that. Is that correct?

Anyone know what the bit rate of the mpeg 2 compression is? Does the camera run a 60min mini-dv tape @ normal speed? If so, then it's limited to 25 mega bits. (The limit of mini-dv tape). - 3.6 mega bytes/sec.

So if a normal ntsc dvd is @ 480 resolution (close to half of 1080) and is typically at a bitrate of no more than 9.6 mega bits per second (the limit of DVD) then this camera is recording with about the same compression as a typical dvd movie.

Yuck.


Doe this make any sense? I'm stoned. Tell me I'm wrong. Try projecting a dvd movie on a large screen. Yuck.
 
Re: help me understand this camera's compression

[quote author=blazingoat link=board=sony;num=1103844428;start=0#0 date=12/23/04 at 15:27:08]Let me see if i've got this straight. This camera shoots at 1080 resolution. But it's still 4:1:1 compression, and then mpeg 2 compression on top of that. Is that correct?[/quote]
It images at 960x1080, and uses pixel-shift to get that up to 1440x1080.  Then it gets chroma-decimated to 4:2:0 (not 4:1:1) and then the results undergo MPEG-2 compression.

Anyone know what the bit rate of the mpeg 2 compression is?
Click to expand...
25 megabits per second, using a 15-frame Group of Pictures (GOP).

Does the camera run a 60min mini-dv tape @ normal speed? If so, then it's limited to 25 mega bits. (The limit of mini-dv tape). - 3.6 mega bytes/sec.
Click to expand...
Yes, that's exactly what it does.

So if a normal ntsc dvd is @ 480 resolution (close to half of 1080)
Click to expand...
No, it's less than 1/4 the resolution.  HDV is twice the horizontal pixels and a little over twice the vertical pixels, so a frame is about 4.5 times as large as a DVD frame is.

and is typically at a bitrate of no more than 9.6 mega bits per second (the limit of DVD) then this camera is recording with about the same compression as a typical dvd movie.
Click to expand...
Quite a bit more compressed, actually.  If you divide 25mbits by the 4.5x larger frame size, it gives you an equivalent of about 5.6 megabits per DV-sized frame (which, by the way, I have no idea if that's a valid way to look at it or not...)
 
Re: help me understand this camera's compression

Barry,

You're the man. Thanks for correcting my numbers. I suspect it's not a vaild way to look at it, but it's my way of looking at it.

I doubt anyone would ever intentionally take their pristine HD footage, (no mater what format) and dumb it down to 5.6Mb mpeg 2 per dv sized frame before it goes out to film.

High rez or not, that compression sucks.


Barry, one more question, could you please explain to me what chroma-decimated to 4:2:0 means.

I'm limited in my understanding of sampling. I was under the impression that it's Luminance: Red - luminance :Blue - luminance (Y:R-Y:B-Y). It's a ratio, so for 4:1:1 it means the chrominance is sampled once for every 4 luminance samples.

at 4:2:0 does that mean that there is no B-Y sampled? That doesn't seem to make any sense. How does the signal still have rgb color? There's no blue sample and no way to calculate the green value......

How does 4:2:0 relate to 4:1:1 and 4:2:2 and 4:4:4...

Thanks barry!
 
Re: help me understand this camera's compression

"decimated" is simply a term to describe that it's reduced from four down to one (which is not technically the right term, since "decimated" would imply reduction to 1/10 or by 1/10, but... well, close enough...)

The 4:x:x ratio implies the # of samples per even and odd line, per 4 pixels of luminance. They're ways of measuring the # of chroma samples in a 4x2 pixel block.

Both 4:2:0 and 4:1:1 have the same # of chroma samples for a 4x2 block of pixels, but in a different configuration. 4:1:1 has one color sample on every line. 4:2:0 has two color samples on every other line (which means that on horizontal scan lines you'll have one scan line with two color samples, and the next scan line has zero color samples). So in 4:1:1 the color sample is long and skinny and applies to each 4x1 block. In 4:2:0 the color sample is basically square, and applies to each 2x2 block.
 
You must log in or register to reply here.
Back
Top