Whats the best encoder in terms of quality for mpeg2 (dvd)?

[Nelson37]

Resolution of analog signal is a tricky area; however-

I have done a fair amount of research on this and SFAIK, there is NOT a linkage between cable type and resolution, at least in real-world terms.
Component, S-video, composite, and coax - listed in order of quality of signal retained - all can carry resolutions higher than current equipment is capable of displaying. I have personally displayed an 800x600 signal over both S-video, composite, and even coax, onto a standard TV. Color seperation shows a variance, but the full picture is displayed.

Where I think the confusion comes in is that the better-quality equipment tends to have as a built-in option the better quality connection, as with S-VHS decks having S-Video connections (I have even seen S-Video cables labeled as "S-VHS cables"). The better cable/connection types do NOT "enable" a higher resolution, they just loose less of the signal, which is in turn more obvious/objectionable at higher resolutions.

On the Vdub question, it sure as hell is not vaporware, what it IS is the most valuable single video prog I have used, far better for editing than many $100.00 plus programs, and it is free.

[DJRumpy]

It's not whether or not the cable can carry the signal. It's the quality retained. There is no way a composite connection can retain the original values in a component connection. It's just not possible once all the values have been thrown together. If an SVCD cable is rated to carry a signal at 352x480, then trying to grab a source signal of 720x480 via that cable is going to give you a poor copy of the original due to signal degredation.

[Nelson37]

Saying that a cable CAN CARRY a certain resolution, but with some degradation, (which is what I said), is an entirely different thing from saying a cable IS NOT CAPABLE of a certain resolution.

There is a difference between SUBJECTIVE evaluations of quality and OBJECTIVE hard numbers of resolution capability.

[DJRumpy]

Fair enough. Let me put it this way. Why would you want to capture a signal that is poor due to signal loss, when a relatively clean signal could be captured using a cables' rated capacity and resolution?

Just because you can do a thing, doesn't mean you should. We'll just have to agree to disagree.

[Nelson37]

In everything we do in this hobby, there is loss. All we can do is try to minimize it. By all means, use the best connection type and cable you have available. I was not suggesting to use S-Video in place of Component, or at any time to use a lesser type if a lower-loss alternative is available.

My point was that to limit capture resolution due to the cable type available is non-sensical, and to underline the continued fallacy in your last answer, Cables do not have a rated resolution, or at least not one that limits the resolutions that are available to us within current standards.

This is similar to some people saying that MPEG-1 is limited to 352x240, when the actual limit is 4096x4096. Whether MPEG-2 may or may not be better is not the point, there may be a legitimate reason to use MPEG-1 instead of MPEG-2. The point is that the stated limit is NOT accurate, and that the decision should be made based on other factors.

Even Coax can carry HDTV resolutions. Recommended? Hell no, but if coax is all you have it is still possible, the attempt should not be abandoned because of a "limit" which does not exist.

[Thorn]

The situation is a little more complicated for analog signals. While a VCR may not have a very high resolution, it is NOT exactly 352x240 or any other computer-related resolution. You can say in general if one device has a higher resolution than another but the assumption that you want to capture at low-res just because the analog source is low-res is a bad one. You see a difference in quality if you capture at 720x480 vs. 352x240. The reason is that the capture chip has to do some on-the-fly resizing (in a sense) when you capture to get the image size you want. By capturing in as large of a frame as possible you get the best possible image to work with. The difference was visible to me when encoding VCDs from analog captures, so I always cap at full resolution with Huffyuv. Some might say that's overkill, but there really isn't any substitute for quality, is there? If you're dealing with constraining circumstances then you deal with it. Otherwise go for the best.

Now, umm... are you people correlating the best capture resolution with the cable used??? The two have NO relationship whatsoever. Choose your video connectors based on which provide the best video quality (or more typically, the only one you have available). There is no resolution locked to a cable and you certainly don't choose the resolution to send over the cable. While there is some argument, I belive the pecking order is component, s-video, then rca. It's all decoded into an analog signal to be encoded by the capture card, so even if one cable type has a higher "resolution" than another that would be secondary to the image quality that the card gets.

[medievil]

Composite is limited to about 352x240 resolution.

This I disagree with. Composite can Also carry the full 480 lines (Unless you are using very, very cheap cables).
I am also on the verge of disagreing that it's only 352, I think it's more along the lines of 576. We know that VHS is definately lower (352X360 or so) and there is a huge difference between Composite from Satellite and VHS.
Using Satelite there is no difference bewteen Composite or Svideo as far as resolution goes. he only difference technically between the 2 is the coloron Svideo being on it's on seperate wire and combined at a different stage in the video circuit on the reciever. They did it to help prevent cross over with the color signal interfering with luminance.

I work in Electronincs (VCR's, TV's all the repair stuff) and am positive The resolution difference between the 2 is 0. The quality of the signal (Being clean and such), however is increased by using Svideo.

[DJRumpy]

Thorn, no one is saying VCR is 352x240. I'm talking usable formats as the end result. I would never bother to capture VHS to anything higher than 352x480. We're talking these resolutions because the captures we're talking about are probably going to end up on VCD, CVD,SVCD, or DVD. We also how each cable will perform. We're simply talking about which will deliver good signal at which resolutions. Capturing a source at a higher resolution than the source gains you nothing. This is the same as trying to re-encoding an MP3 at a higher bitrate, to get better quality. Pointless. You can't improve the quality by enlarging it. You only exagerate what's already there (including the flaws). Last point: of course you would use the best medium available (component in this case). Anything less can introduce loss in the analog signal. Your stating the obvious.

Medievil, a couple of points:
Composite is a standard cable used for broadcast television. Of course it can carry the full 525 Lines (although not all 525 lines are used).
As for VHS & Sattelite, now your talking SOURCE. Of course satellite looks better. It's capable of HDTV resolutions, which in turn, looks better on your TV as an end result, no matter what cabling you use. The source signal is better, and digital for the beginning of the trip.

Why do you think they use S-Video cables to begin with? By placing all luminance and chroma values into one signal (composited signal), you have color bleeding/loss, and dot variance as a result. S-Video keeps these values separate, allowing higher resoutions, while retaining quality. I can hook up my progressive scan DVD to my HDTV via composite, and force progressive mode, but the end result is crap. Why? Because the cable, although it can carry the signal, it is not designed for it. Does it display the full resolution? Yes. Is it watchable in any way? No.

[DJRumpy]

Was I cranky this morning or what?

Bottom line. If you exceed the recommended capability of the cable your using, you would end up with a lesser quality capture, than if you had captured the same source at a resolution that the cable handles better.

A bit more research didn't yeild much. The only useful link I found suggests S-Video cable tops out at about 640x480, and component tops out at about 800x800. This makes sense, since S-Video isn't used as an option for progressive scan on a DVD player. To enable it requires component, since it wouldn't produce a good quality picture at 720x480.

No idea where composite tops out, but it would have to be somewhere below S-Video, but meeting the minimum typical broadcast requirments (440x480 NTSC).

Capturing a source via composite connections at 640x480 will still only net you the resolution of your source (440x480 for NTSC), which is then scaled after it's captured, giving you a magnified view (including the flaws).

By the same token, capturing the same broadcast source via S-Video at 640x480, would still net you the same resolution (as our source hasn't changed), but with a cleaner signal because of the cable.

Same situation, but this time assume a higher res source (say 720x480 for example). You would not want to capture this source at 720x480 if it's passing through a composite media (remember, your hardware is hard coded to produce the same signal out of a composite connector, regardless of what your source is, your never going to exceed it's design output) to get to your capture device, because your still getting the best that cable can produce (since we don't know the exact number, lets assume a full 525 lines horizonal), and then stretching it to fit your capture requirements. It makes more sense (at least to me), to shrink your capture instead of stretching, so artificats and flaws are not exaggerated. The next closest compatible format is half D1 for me, since I no longer burn SVCD/VCD.

The MP3 anology is the closest thing I can think of for what your trying. Simply raising the resolution isn't going to net you a better capture, if your source is limited in quality by the pipe it's passing through (or for the MP3 anology, since the MP3 already has a hard coded cap of kbps defined by it's encoding, re-encoding at a higher bitrate will not net better quality).

[ronnylov]

There is another reason to capture at as high resolution as possible and then resize to the destination resolution. The resizing itself works like a denoising filter. Let's say the destination resolution is 352x576 (I'm in PAL land). Then by capturing at 704x576 you can average 2 pixels for each final pixel during the downsizing. You have two source samples for each destination sample. I beleive this is similar to oversampling? Also when you use denoising filters you have more source pixels to work with and may not loose as much details as when you apply the filters on a lower resolution image.

There is a method to reduce noise which is done by capture the same source video clip several times and make an average of each frame from the different clips. This reduces much noise with low loss of quality. But instead of making two half resolution clips and average them you can make one full resolution clip and downsize to half resolution.

Maybe we are off-topic now? Anyway this is a method to get better quality for mpeg2 (dvd) by using preprocessing before the actual encoding.