S-Video better than composite

[marklewis]

Hi

I've probably asked this before, sorry but cant receall the answer.

S-Video output is considered better quality than composite and there is clearly a better quality image coming out of S-Video.

How can the two be compared?

[jagabo]

Find a device with both s-video and composite outputs and look at them?

[marklewis]

Nohhh! I mean why is s-video better quality than composite? Something to do with shielding or separating colours into different cables.

[victoriabears]

It is not necessarily true that s-video will give you a "better" picture quality, its often subjective and a personal preference.

[marklewis]

But there is a technical difference ...

[smithburger]

S-Video is theoretically better quality because less processing is required to transmit/receive it. Composite requires modulation/demodulation before transmission and after reception. Noise is added and information is lost during most processing.

The price you pay for S-video is it requires two discrete signal cables.

[usually_quiet]

Comb filters are needed for composite video to reduce dot-crawl noise. Because S-Video separates luma (luminance) and chroma (color), it does not produce dot-crawl noise.

[SatStorm]

SatStorm's totally unofficial & non proven theory, based on his personal practical knowledge about capturing over the years:

composite: 640x480
S-Video: 720x576

So, composite is OK for NTSC. For PAL you need that extra more that S-Video gives you.
+ this annoying dot-crawl noise, especially on the red colour. No such a noise on S-Video.

Exception: Laser Disc capture. Composite is the best option there.

[edDV]

Composite video has the chroma infomation encoded on the same wire at 3.58MHz (NTSC) or 4.42MHz (PAL). The luminance shares the same bandwidth up to the channel cutoff. A clever trick of mathmatics allows the chroma to fall into unused frequencies between luminance harmonics. An expensive comb filter is needed to properly separate luma and chroma at the receiving end. Cheap receivers simply notch out the higher luminance frequencies around subcarrier thus reducing picture detail.

S-Video has luminance Y and encoded subcarrier on separate coax cables (4 pins). Source devices such as camcorders convert RGB to Y, U, V. The Y is sent down one coax, U and V are encoded to NTSC or PAL then sent down the second coax as "C". A composite output is made by summing Y+C.

Broadcast VTR/VCR recorders used to record the full composite signal (e.g. analog 1" type C or digital D2). All analog TV stations transmitted composite video. Laserdisc also encoded the information as composite video.

Home VHS/SVHS, 8m/Hi8 and Betamax recorders by contrast recorded luma and chroma separately. Incomping composite video from the internal tuner or composite input goes through a luma/chroma separation process. VHS/8m/Betamax VCRs simply low pass luma at ~3MHz (NTSC) or ~3.5 MHz (PAL) then FM modulate RF to tape. Meanwhile C is downconverted to below 1 MHz and AM RF modulated. This recording process limits luminance resolution to about 240 "horizontal lines of resolution" x 480/576 scan lines and reduces decoded U and V to about 0.5 MHz (about 40 color changes over a horizontal line). This class of VCR usually adds Y+C to composite for output. Downstream Y/C separation is relatively easy because recorded luminance has been cut well below subcarrier frequency.

SVHS/Hi8 and later home recorders plus most capture cards record the full luminance bandwidth. These recorders normally use a comb filter to separate C from Y for composite inputs thus preserving the full luminance bandwidth. Chroma is downconverted to less than 1 MHz for SVHS/Hi8 same as for VHS/8mm.

The output stage of SVHS/Hi8 recorders usually allow for separate processing of Y and C to an S-Video output. This avoids the distortions that would result by re-mixing Y+C to a single coax. In this case, S-Video should always be better than composite. It represents what is on the tape.

[edDV]

SatStorm's totally unofficial & non proven theory, based on his personal practical knowledge about capturing over the years:

composite: 640x480
S-Video: 720x576

...

I agree with your capture settings but both represent an oversampling of the actual VHS/S-VHS recorded bandwidth. The following references NTSC-M.

For VHS, 3MHz in theory can be sampled at 6Ms/s which would create a 320x480 raster for NTSC.

For S-VHS, luminance is cut above 4.2 MHz by the FM recording process. Minimum Nyquist sampling for 4.2 MHz is 8.4 Ms/s which would result in a 448x480 raster.

Broadcast NTSC is also band limited to 4.2 MHz.

Cable and satellite DBS send SD video between 480x480 (DirectTV) to 528x480 (Comcast and Time Warner).

Capture card sampling at 640x480 should be adequate for all of the above. 352x480 is an alternative for VHS/Video8.

[jagabo]

For VHS, 3MHz in theory can be sampled at 6Ms/s which would create a 320x480 raster for NTSC.

And the color resolution of VHS is roughly 40x480.

For S-VHS, luminance is cut above 4.2 MHz by the FM recording process. Minimum Nyquist sampling for 4.2 MHz is 8.4 Ms/s which would result in a 448x480 raster.

What is the color resolution of S-VHS?

[edDV]

For VHS, 3MHz in theory can be sampled at 6Ms/s which would create a 320x480 raster for NTSC.

And the color resolution of VHS is roughly 40x480.

For S-VHS, luminance is cut above 4.2 MHz by the FM recording process. Minimum Nyquist sampling for 4.2 MHz is 8.4 Ms/s which would result in a 448x480 raster.

What is the color resolution of S-VHS?

The same as VHS. Only luminance bandwidth (detail) differs.

[jagabo]

Thanks edDV.

[edDV]

I should add that S-Video outputs from digital sources (e.g. DVD players, HD cable boxes) can have significantly better specs than SVHS/Hi8.

DVD recorders/players and DV camcorders are capable of 6.75 MHz luma and 1.7 MHz CbCr in digital. Most of the 6.75 luminance can be passed to the Y coax for S-Video or analog component. The NTSC/PAL encoding process will limit chroma C bandwidth to about 1-1.5 MHz but that is more than double VHS/SVHS. Analog component PbPr can carry up to the full 1.7 MHz (4:2:0) chroma bandwidth.

HD cable/sat tuners can go higher than that for luminance but most will low pass filter S-Video luminance to around 7 MHz. Analog component YPbPr outputs are not low pass filtered so can carry the full HD bandwidth (i.e. up to ~18 MHz luma and ~4 MHz PbPr chroma).

[marklewis]

OK, thanks for that. I get the general drift that composite contains all the signals which then have to be separated at the receiving end and in S-Video the signals are on separate cables.

Also that the resolution possible is higher with S-Video. Which means for me that its worth looking for the right model of player to do the job.

Although much less technical I'm looking for the best way to capture VHS to DVD and its clear that I need an s-VHS player with S-Video output. There are many models which have BNC connectors also which can be hooked up to S-Video, I understand.

I've looked at several models but having great difficulty finding something like this in PAL or SECAM format. Most on offer (on ebay) are in the US too. The most widely available are medical application models like the Sony SVO-9500MD. I'm also comparing with combi VHS / DVD models such as the Toshiba D-VR80KF, but difficult here to understand if the quality is there or better to get a stand-alone.

[marklewis]

This also seems a quite useful guide to the differences between composite and s-video

[jagabo]

The images in this post show one of differences between good and bad luma/chroma filtering:
https://forum.videohelp.com/threads/313735-Capture-card-for-Laserdisc-and-VHS-Good-card...=1#post1940519
Note the rainbow artifacts.

And dot crawl artifacts:
http://www.michaeldvd.com.au/Articles/VideoArtefacts/VideoArtefactsDotCrawl.html
http://en.wikipedia.org/wiki/Dot_crawl

Be aware that the low resolution of VHS means you don't need to worry much about the max bandwidth of the capture device. Much more important is the VHS player you use.

[2Bdecided]

Also that the resolution possible is higher with S-Video.

No, that's not right. S-VHS has higher luma resolution that VHS.

S-video has the same luma resolution as composite - it's just that composite has chroma mixed in (at higher frequencies) too.

Some DVD players will happily output the full 6.75MHz via composite. I know. I've measured them. Some notch composite luma at 3.58MHz or 4.43MHz (to reduce interference with chroma), some don't.

It's irrelevant for VHS, unless your capture device is very poor at separating luma and chroma on composite (some are), because from VHS they don't even overlap (the luma doesn't go far enough up in the frequency range!).

Cheers,
David.

[turk690]

I still have a JVC S-VHS-ET deck with that wonderful S-video jack alive and kicking. I would have luved to use that, but none of the current LCD TVs between Sony and Samsung I was considering buying have an S-video jack. In fact it's hard, if not impossible, to see any current LCD TV model with S-video. BUT, they all have your requisite composite video jack.
Why is that?
Is it because their comb filter circuits are so fabulous they don't have dot-crawl even with only a composite jack? Or are S-video jacks and associated circuitry expensive? Or is the fuss about the merits of S-video now quite quaint in the face of HD, HDMI, and component Y-Pb-Pr??
I wonder: if a current 1920x1080 analogue HD signal had to be forced over an S-video or even composite connection, what possible chroma modulation operation/circuit would have been used? What carrier frequencies would be viable?? Would color-bleeds and dot-crawl be even worse?

[edDV]

I still have a JVC S-VHS-ET deck with that wonderful S-video jack alive and kicking. I would have luved to use that, but none of the current LCD TVs between Sony and Samsung I was considering buying have an S-video jack. In fact it's hard, if not impossible, to see any current LCD TV model with S-video. BUT, they all have your requisite composite video jack.

Why is that?

Is it because their comb filter circuits are so fabulous they don't have dot-crawl even with only a composite jack? Or are S-video jacks and associated circuitry expensive? Or is the fuss about the merits of S-video now quite quaint in the face of HD, HDMI, and component Y-Pb-Pr??

They are cheaping out. They want to phase out analog ASAP to save the cost but ofcourse this obsoletes all the customer's analog equipment. The "Buy More" stores support this. They know the customer won't catch on during the HDTV purchase, then will need to replace the DVD player and other analog interface equipment on the next trip.

Also, the media companies are trying to kill any remaining analog holes.

[edDV]

I wonder: if a current 1920x1080 analogue HD signal had to be forced over an S-video or even composite connection, what possible chroma modulation operation/circuit would have been used? What carrier frequencies would be viable?? Would color-bleeds and dot-crawl be even worse?

Our current (ATSC/DVB/HDV/AVCHD/Blu-Ray) 1920x1080 system is all compressed digital to the home. The analog 1920x1080i output is created in the set top box or player after D/A.

In the late 1980's, Japan launched an analog "HiVision" HDTV system which used MUSE digital compression scheme for recording and transmission. Some might consider Hi-Vision as the ultimate SuperDuper SVHS. It was 1125 lines with 20MHz Y and 7MHz chroma (at input). Various transmission schemes were tried including FM modulation as used in SVHS and Hi8. See history here.
http://en.wikipedia.org/wiki/Multiple_sub-Nyquist_sampling_encoding