Primarily, what I am seeking to do is to do is to transfer relatively recent, fringe, high action cartoons on VHS to some CD format.

I have managed to create 1 CVD (not vcd) that works over the 4 months since I started reading about capturing, encoding, and burning, but I think I have the basics of it down. I do have some questions though.

What resolution does a VCR record at? Is it a function of the speed? (SP, EP, SLP?) Or is that like setting CQ or the bitrate lower in your encoder---just a hit on quality.

I have read a bunch on Luke's page about telecine (apparently a method from going from 24 to 29 fps by adding basically duplicate frames) and interlacing (apparently TVs pictures are built of two pictures which are displayed at the same time but alternated line by line). I understand that in theory, if you had your decoder do an inverse telecine, (like NTSC VCD film) you would end up with a file about 80% the size of the original VIDEO and if you de-interlace, you eliminate about 50% of the size that video. As I understand (please correct me if I'm wrong ) that is an ideal situation if you want to watch the video on your computer as the monitor handles images way differently than a TV and doesn't need that extra stuff.

But isn't that a problem if you are then going to burn this stuff so you can play it on a TV? I understand it is less of a problem with cartoons since the entire concept is EMULATING movement.

Since disk space is still unfortunately an issue, and since it seems the source material is not that great, I have been capturing directly to MPEG-2 and then reformating using TMPGenc and then burning with NERO. My thoughts were to burn the stuff in CVD format (352 x 480 NTSC) so I would be able to burn that to a DVD in the future without reencoding.

As I understand it, capturing at 480 vertical means I am capturing both images in the frame. To de-interlace (take out one of the interwoven images), that vertical resolution (in NTSC areas) would be required, right? (If you capture at 240, does it drop a picture from the frame automatically, or because it is handling it in dots, will it ultimately just kill your resolution?)

Now as I understand the benefits of CVD, capturing at 352 horizontal is no big deal because your TV shows lines, not dots like your monitor. To your TV, there is little difference between 352 and 480 Horizontal. It is just another 27% space savings. And it is a DVD compatible format.

Now is there a quality bump in recording stuff at a higher horizontal resolution, or would that only be the case if you were only going to play the resulting file on your computer (dots vs. lines)?

In capturing from video, what kind of quality difference do you see in capturing an avi in a lossless codec like Huffy and then TMPGenc it down to a CVD resolution vs. capturing directly into MPEG-2 in that resolution? (Unfortunately my recently departed capture card -an I/O magic PC PVR- never seemed willing to capture AVI, so I don't have a first hand compairison. Speaking of cards, any recommendations? I am thinking about a Radeon all in wonder 7500? Good/Bad choice?)

I have tried de-interlacing and inverse telecine with mixed, mostly unsatisfactory results. Do those need to be processed in a certain order (multiple encoding sessions), or if you select both in TMPGenc does it automatically do the IVTC first (since the telecine frames would be more obvious that way)?

I also find it a bit confusing about WHERE in TMPGenc I should be specifying to de-interlace or IVTC. I feel like when the 5 screen wizard is talking about the source, I should be discriptive of what I think I have, right ? Anyone who wants to thow up a screen shot or two, I would be immensely greatful. Also should that be 1:1 VGA since I have ALREADY captured the video.

I know I bounced all over the place, but I really appreciate any help, advice, corrections, or explanations you guys can give...

Well digging through one of zooter's links I found some info---not hard answers but maybe they help a bit...I'll post them and maybe someone can add to this...

from http://www.divx-digest.com/articles/vhs_capture.html

"If you search through the internet looking for information on the VHS resolution, you will find basically 352x240 at 29.967fps (Frames Per Second) for NTSC and 352x288 at 25fps for PAL-M. Yeah, right. ....So all those documents were wrong? No, they were inaccurate, or maybe just incomplete. The VHS video works with the interlaced system (not progressive), which means it draws every frame twice on the screen. ...If you capture at 352x240 or 352x288 you'll be skipping every second field, which results in only half of the resolution."


"If you want your capture to look just like the VHS source (unfortunately it can't look any better), capture at 640x480 at 29.967fps...."

Concering resolution to capture from vhs I found Kwag said the following in response to another post.

"Always capture at the resolution you intend to encode.

Note:
If you are capturing from a VHS tape or TV ( Satellite, etc. ), you capture at 352x480, not at 352x240.
This way you get both fields.

Even if your intention is to create a 352x240 VCD.


There is no quality gain by capturing above 352x480 if your target is 352x480 or below. "

That is supported in nicky's guide
http://nickyguides.digital-digest.com/video-capture.htm

It doesn't strongly advise, but offers this explanation from the ligos company... (truncated for space)

"Video Format Horizontal Resolution
Standard VHS 210 Horizontal Lines
....

We suggest capturing at a resolution that most closely matches the resolution of the video source. For video sources from VHS, Hi8, or Laserdisc, SIF resolution of 352x240 will give good results. For better sources such as a direct broadcast feed, DV, or DVD video, Half D1 resolution of 352x480 is fine."

So I guess that's clear enough on resolutions---if I don't want to mess with de-interlacing or telecine, I might as well capture at 352x240. Since that only captures one feild of a frame, no interlacing to worry about. If I want to try IVTC to remove telecine (and then I guess encode it as a film?), it is easier to detect the pattern if both feilds are captured.

Well there are still a bunch of questions I hope you guys might help me with.

I found this by Kwag on inverse telecine. Basically I think I get what this is saying. If you can successfully do an inverse telecine, it SHOULD remove the feilds generated in doing the telecine. When those feilds are removed, you should have pretty much the matching feilds in each frame---no combing effect. Now is that really the case, or does even the process of creating feilds generate the combing effect (to a lesser degree)?

"Then I use AviSynth with decomb.dll to do IVTC automatically.
There's no need to do De-Interlaceing, because decomb takes care of producing the original progressive image and discards the duplicate frames. "

I have to agree with finiteman.

capture at: __x480@29.97fps

If you want VCD - 352x240, capture 352x480.
If you want SVCD - 480x480, capture at 480x480.

I find my drivers for my card, limit me to 640x480, so that is what I cap.

1. Cap and convert with HuffyUV.

2. Edit/filter with VirtualDub. Have a look at the VirtualDub site and download "flaXen's cartoon filter".

3. Convert with TMPGEnc - SVCD 60 minute template, tweaked to "2 pass VBR" and Motion Search "High Quality".

Hope this helps.

__________________
--
Willy