S-Video vs. Composite, when the capture card matters.

[swiego]

I have a Theater 550 Pro-powered capture card being fed from a JVC 9911U. As you may know, the 550 Pro chipset, besides its atrocious software support, lays claim to some unique capabilities. Possibly the most touted of these is the 3D Comb Filter which ATI climbs will match that in most other top-tier consumer devices.

The nice thing is that while the software support of the 550 does suck, using Graphedit you can flip between the 2D and 3D comb filters! The improvement in quality between the 2D and 3D filter is pretty noticable on a number of different kinds of video, definitely for stills and somewhat less so for motion. So I've done some captures and screenings of tapes I'm familiar with, over s-video (AR "Master Series" s-video cable) and composite (cheapo Monster dealie that was sitting in a box).

What I've found is that I think the PQ is actually better over composite. I see fewer edge artifacts, a somewhat sharper picture and less color bleed around edges. What this tells me is that maybe the comb filter in my ATI card is far better than the one in the 9911 SVHS player, so that for this particular combination it's better that I use composite.

I guess my question is... what are your thoughts? Is this possible, or am I missing something obvious? If it's true that you need a comb filter somewhere (I think that's the case) then should we be trying to determine whether to utilize the one in our playback device vs. the one in our capture device? Or is this not true at all?

[gshelley61]

The comb filter in your 9911 is not in the signal chain during playback using s-video out. The Y and C signal components are actually separate on VHS tape. The comb filter comes into play when recording or passing through a composite source, or using the s-video output for the 9911's built-in TV tuner.

When you use the composite output, the Y and C signal components are combined. This can sometimes soften the picture a bit. It would be unusual for the composite output to look better than the s-video output when playing back a tape. What you may be seeing is the effect the comb filter in that capture card has on the composite signal. Most modern digital multi-line comb filters are excellent compared to Y/C separation filters from even just a few years ago.

[edDV]

Are these S-VHS tapes or VHS?

Y and C are recorded separately on VHS/SVHS tape through the color under recording process. Color separation takes place during the recording process if the source was composite.

On the playback side Y and C are read off tape and are fed separately to the S-Video connection. At the 550 side, Y and C from the S-Video input bypass the comb filters and go through A/D conversion. I'm not sure which other filters are in the Y or C path.

Also, on the playback side Y and C are read off tape and are added together to form the composite output. At the 550 end, the composite signal is fed to the comb filter. It is the job of the comb filter to try to separate Y and C again. In theory, the best you can expect is to match what is already on the S-Video input.

There may or may not be other filtering affecting what hits the A/D converter.

[Suds-N-Spuds]

It's possible that the chroma and luma were not separated properly when the tape was recorded. Unfortunately this is a common problem even with commercial tapes.

If this is the case you may be getting a better picture using the composite connection because your capture card is doing a better job at re-separating the chroma and luma.

[edDV]

It's possible that the chroma and luma were not separated properly when the tape was recorded. Unfortunately this is a common problem even with commercial tapes.

If this is the case you may be getting a better picture using the composite connection because your capture card is doing a better job at re-separating the chroma and luma.

If this were VHS, there would be no residual subcarrier left in the Y since Y is low pass filtered below 3MHz. There may be some Y crosstalk into the separately recorded C. In that case it would seem you are better off keeping that away from the luminance path rather than mixing it back in with a composite connection.

If this were S-VHS with luminance response out to 4MHz, I'd need to hear the arguement that a digital comb filter can remove pre-existing chroma crosstalk from the Y path. I can see where a 3.58MHz notch filter in the S-video Y path could work.

Any upstream Y crosstalk into C would be part of C before before recording. I don't see how a digital comb filter can distingush between real color and crosstalk color on the playback side.

[swiego]

Thanks edDV for the very informative post. I apologize for the late reply but I did want to do some more testing. I think I traced this down to the 9911U, there seems to be more artifacting or other "stuff' going on with the s-video output that I simply don't see with the composite. I tried another player with both outputs and could not see any difference at all between video and s-video.

[anitract]

Lately I've been wondering if my prosumer jvc vcrs benefit more from using s-video jacks or the rca jacks (and thus the comb filters) when playing STANDARD VHS tapes. I have a hard time telling one way or the other though. edDV, any thoughts on this?

[SingSing]

See diagram posted by edDV, if the Y/C seperater can seperate Y and Cr/Cb very well, than the result will be the same.

[edDV]

Lately I've been wondering if my prosumer jvc vcrs benefit more from using s-video jacks or the rca jacks (and thus the comb filters) when playing STANDARD VHS tapes. I have a hard time telling one way or the other though. edDV, any thoughts on this?

A comb filter in the JVC VCR will only affect recording, not playback. If you mean a comb filter in the capture card, in theory it will never match the S-Video direct connection quality.

For VHS playback, the only information above 3MHz will be mix chroma and noise. The comb filter doesn't have much to separate. There is no overlap.

see http://members.aol.com/ajaynejr/vidcomb.htm