SVCD Conversion Questions

[puar]

There are a couple parts to my question.

First, I've seen on some other msgboards and posts that sometimes people who capture in 704x480 will resize it in VirtualDub to 480x480 before encoding it in TMPEnc. Is this additional step necessary or just a waste of time? Does it do anything? Do you lose quality?

Next, I've been slowly ironing out problems with my capture and conversion thanks to much help here, but I do have another problem now. I've captured an SVCD at 704x480 and encoded it in TMPEnc using highest quality for the motion search. Played back on PowerDVD it looks fine. However, playing back on my dvd player, there is choppiness in movement, especially whenever slow camera pans are done (moving the camera scene from one side to the other). I didn't have this problem earlier when capturing first at 352x240 (altho those svcds are definitely not as sharp as the 704x480).

From what I've gathered from some other posts I've read, it seems that important settings here to consider are the field order, and interlacing/non-interlaed/de-interlacing settings. Can anyone give me the low down on these (both in VirtualDub and TMPGEnc)?

Thanks,
Bobby

[spiderman2k1]

i think you have to set the field order to A or B for encoding with TMPG I have had the same problem when ripping DVD then encoding

[Kdiddy]

"sometimes people who capture in 704x480 will resize it in VirtualDub to 480x480 before encoding it in TMPEnc. Is this additional step necessary or just a waste of time?"

This depends on who you ask, the people who do that prefer the resize filter of Vdub over the resizer that TMPG uses, they say Vdub resizers maintains a better quality than TMPG's. I dont have an opinion on that.

"I've captured an SVCD at 704x480 "
assuming your bitrate is standard SVCD 2.5 mbps since you didnt say, then here your resolution is high, try standard SVCD resolution 480x480..if this doesnt help, then yes I would move on to seeing what the deal is with "the field order, and interlacing/non-interlaed/de-interlacing settings."

[puar]

You were correct in your bitrate assumption. Do you have any suggestions as far as the interlace/etc. settings?

Should I deinterlace? Should my source/stream be set to interlace or non-interlace(progressive) in TMPEnc? The source file is an AVI capture in 720x480 from a broadcast signal or vhs. The resulting SVCD will be played on my TV/DVD player.

Bobby

[Kdiddy]

I dont do any capturing, Im not sure how the interlacing works with that, I only do DVD progressive force film, so interlacing is not an issue for me....did you try 480x480 first, since you say 352x240 did fine, I realy think lowering to 480x480 really would solve the problem....before you go worrying about the interlace subject.

[puar]

Are you saying I should capture directly into 480x480? Or convert to 480x480? Unfortunately, I get a weird error when capturing at odd resolutions (e.g. 352x480 or 480x480). What happens is my VirtualDub (or anything else) will kind of flicker and then somewhat resize the image to the window. I know this isn't clear, but to elaborate...

Imagine an image of normal aspect ratio (let's say 352x240 or 704x480). Then, just crop it to the resolution of 480x480 so that the right side of the frame is missing. This is what happens when I try to capture in 480x480. It starts out fine, but then resizes. I think it might be a driver issue (with win2k specifically). I'm not sure yet. I'm using a Matrox Rainbow Runner G-series for my tv capture.

I'm currently conducting some tests (encoding at different settings). They actually all look identical to me so far, so I'm not sure... Maybe it's a problem with my original capture.

Bobby

[Kdiddy]

like I said I dont do capturing so I wouldnt know how to advise there properply, off hand, I would think that capturing at 480x480 and encoding at the same would be the best....but if you are having rpobs with that, then capturing however you were capturing before...then just resize in TMPG down to 480x480.

[adam]

Its always best to capture to a higher resolution and convert down. As far as which resizing method you use, it probably depends on who your asking. I think vdub's is better than TMPGenc but I don't know if its worth the extra step. You know, you can frameserve from vdub to TMPGenc so that you can do it all in one step. Frameserving is slow but it will be faster than resizing then encoding.

As far as higher resolutions playing choppy... You might be dropping frames. Its just something that can happen even with good quality hardware. You actually might be forced to capture to a lower resolution, even though its not the preferred method.

As far as the interlacing. Do not deinterlace, svcd supports interlacing so if you do nothing else, just set both the source and output to interlaced.

If your source was properly telecined then you can perform an inverse telecine. This is the single most thing you can do to improve the quality of any video. It improves the quality of your video by about %30. If you are unfamiliar with IVTC then just try using TMPGenc's auto IVTC function, check the advanced tab. Set it to 24fps and auto setting, then hit enable while encoding. Depending on your source this may or may not work correctly.

When you do this set the source to progressive and on the video tab set the encode mode to 3:2 pulldown when playback.

If you need more information on what IVTC does and more information on when it can be used do a forum search. There have been many discussions about it. Its definitely worth a try though if your looking for a way to get higher quality encodes.