S-video No Compressed?

i have read...
S-video out when read from dv tape 4:1:1 resamplig chroma in 4:2:2 , when shot live is 4:2:2.
Realy?
 
bullamrk said:
i have read...
S-video out when read from dv tape 4:1:1 resamplig chroma in 4:2:2 , when shot live is 4:2:2.
Realy?

S-video is an analogue output, so I don't believe you can refer to it as 4:2:2 on 4:4:4. I am not very good with analogue formats though so I can't tell you much more.
 
bullamrk said:
s-video out not is compressed ?
S-video is an analog signal. It does not have digital compression but it does have analog compression (if there is such a thing). The chroma signals are blended together, irretrievably.

A DV signal is a component video signal, and it's going to look better than any S-video signal under all circumstances (that I can think of).
 
If you use s-video out when you shoot, then it is uncompressed analog output.
If you use s-video to play back your DV tape, then it is an analog output that was converted in real time from DV 4:1:1

...usually, digital to analog converters are not so good in a camcorder.
 
A S-video signal is 4:2:2 ..just as is composite and Component (Y/Cb/Cr)

Component is full Y (luma) and half 'samples' of Cr and Cb (the red and blue colors, green is always calculated from teh difference) hence 4:2:2
The Cr and Cb signals are modulated togethere as phase dieeference to a carrier signal in S-video (also known as Y/C) The Y is the same as Y from component (full luma) and the colors are C, they are modulated together preserving the 2 'samples' each. So S-video is 4:2:2
Composite video is the Y and C of S-video all modulated into one signal, it's still 4:2:2 though! - there is just less bandwidth and therfore more possibilty for errors. But it's one signal. This is what analog TV is; just a Composite signal sent over RF 4:2:2 all the way into your home! ..So yes, even VHS is 4:2:2! ("wha!....")

"4:4:4" is actually a slightly false statement, as it is actually refering to RGB ..which is used for example in a VGA signal, but it is not the 4 of Y as in 4:2:2 and 4:1:1 ..but the patern fits so it's used.. this is the only truely uncompressed signal!

SDI (and HD-SDI) are 4:2:2 ..except for true uncompressed SDI, which is 4:4:4 and requires a Dual-Link SDI (2 paralell SDI cables) to carry the signal.

If your source is a raw feed, then it will be 4:2:2 over Component, S-video, or Composite. But if you are originating form somethin lower, like DV at 4:1:1 (4:2:0-4:0:2 for us PAL folks..) then of course "Garbage in = garbage out" applies and you will only be carrying 4:1:1

This is why BetacamSP (an analog Componet format) is still broadcast quality, but DV is not!

This of course only refers to color sampling per pixel (or 4 pixels actually..) then there is still pixel number and pixel depth compression, and time compression (like MPEG) ...but that's another story!

- Mikko
 
mikkowilson said:
This is why BetacamSP (an analog Componet format) is still broadcast quality, but DV is not!

Have broadcasters given up with the term "broadcast quality" yet?

If the BBC ever told me something wasn't "broadcast quality" I would ask them to check out the hours and hours of camera phone footage they broadcasted during the Iraq War or after the 7/7 London bombings!

We all spend our lives working towards visual perfection, and, at the end of the day, no one gives a ****! :cry:
 
haha, i definatly get, and agree with, your point mmm.

however the term is still thrown aroudn a bit.. and it still holds true for most situtations, except one: news.
And that's because the content is much more important than the quality at times.

But for planned program making, the term "broadcast qaulity" is still viable... ...simply and plainly beacuse the (analog) broadcast signal is 4:2:2 (of course there's resolution/bandwidth inssues in there that keep VHS, for eample, off the list ;) - it's a basic technical description.

But as you said: CONTENT IS KING.

- Mikko
 
Analogue does indeed have compression - except it gets called "noise" and "bandwidth".

I don't know what the analogue bandwidth of a typical s-video connection is, but I'd suspect it's below that of 4:2:2 on chroma and luma, but not by too much. To get a precise answer, you'd have to measure the connection and see for yourself.

Although 4:4:4 is predominantly for RGB, there's nothing stopping you having 4:4:4 Y'CbCr.

"This is why BetacamSP (an analog Componet format) is still broadcast quality, but DV is not!" Highly debatable! DV is at least as good as SP, and although has slightly lower chroma sampling, it's luma resolution is much higher, and that's visible to the naked eye whereas chroma sampling is not generally visible without some kind of analysis. That means that BetaSP is most definately not capable of recording and playing back 4:2:2 without mushing it down to about 3:1.5:1.5.

Graeme
 
Well yeah.. of course the numers are a *little* flexible.. this is the main atribute of Analog.. nothign is absoloute.

4:4:4 Y/Cb/Cr? how?

haha an yes the BetaSP vs DV debate is an old one (about as old as DV...) but i'm just discussing the color sampling and compression, as that is what was eluded to by the question of this thread in tegard to 4:x:x sampling.

- Mikko
 
OK,

but in the end, even if the analog is "equivalent" to 4:2:2, if your footage has been captured in DV 4:1:1 color space, you'll output nothing more than 4:1:1 quality from S-video.
It won't upsample the chroma space just because it's using a 4:2:2 output., the original info IS 4:1:1.

You'll have to use a chroma upsampler in post to close the gap between 4:1:1 and 4:2:2
 
"4:4:4 Y/Cb/Cr? how?"

Well, it's got to exist as an intermediary step between RGB 4:4:4 and Y'CbCr 4:2:2, ie, before you undersample the chroma, but after you convert to Y'CbCr. For instance, FCP treats all Y'CbCr video as 4:4:4 for effects etc.

"haha an yes the BetaSP vs DV debate is an old one (about as old as DV...) but i'm just discussing the color sampling and compression, as that is what was eluded to by the question of this thread in tegard to 4:x:x sampling."

But by it's very nature, analogue video is not described easily or natively in x:y:z terms, and as pointed out, analogue video is compressed, just by different means to digital by way of noise levels and bandwidth, and bandwidth can be directly equated to x:y:z sampling, but it's not going to work out as whole numbers.

Graeme
 
ok, yeah 4:4:4 is an intermediary, but it's very rare, as it imples full data on all colors.. and that's nor true because in YCbCr the Cg is calculated, and there for compressed.

the whole 4:x:x system was concived in teh analog days to describe the formats.. the numebrs are based on simple approximations used ot quickly describe a quality. Digital only helped made these vales more absoloute.


Anyway, S-video will always be sending '4:2:2' but as you correctly said, it will only be casrying 4:1:1 data. Any upsampling done in post is then of course interpolation, nothign more than a mathematical guess. - which can be good, but still a guess.

- Mikko
 
Right,

But when i use G Nicer (chroma upsampler) in a DVC Pro50 timeline (Final Cut Pro 5), then i can actually SEE that overall image quality improves, even on my old 21" TV.
More, i still see the improvement even after burning to a DVD.

So, i fugure it closes the gap between 4:1:1 and real 4:2:2.
How much ? I dont know...but enough to see the advantages of using it.
 
"ok, yeah 4:4:4 is an intermediary, but it's very rare, as it imples full data on all colors.. and that's nor true because in YCbCr the Cg is calculated, and there for compressed."

Well, not compressed, although there might some mathematical imprecision and rounding errors in it, but that's not compression per-se.

AFAIK, 4:x:y designation came in with digital video. I never saw any reference to it back with analogue, and indeed, 4:2:2 was first know to me as a designation for digibeta.

Thanks Antione for your comments RE: G Nicer! A lot of work went into that algorithm.

Graeme
 
oh yeah, synthesized information is aditional information, and it will look 'better' but is it *correct*? there is no way to tell!

A CCD in even an analog camera (a 1CCD cam that is..) with a Bayer color pattern will result in 4:2:2 sampled colors.

Interesting discussion guys, thanks.

- Mikko
 
Back
Top