VHS capture question

[Baerchen]

uh no, that old topic again. Anyway, I have some specific questions:

A quite widespread opinion is, that when capturing from VCR or TV, a resolution of 352 x 576 is favourable, since I have more details as when capturing with 352 x 288. Any higher resolution is said to be useless.

However, 352 x 576 means interlaced. Now I need to apply some filters, like rmPAL, TV Chroma up and 2D cleaner. But for these filters, the frames must be deinterlaced so the filters don't mess up. Here I have to sacrifice quality: debob'bing (interpolating lines) is said to be best (all other methods have their downsides too, such as motion artefacts etc.). But when I have to debob, I could also capture only at 288, since that means the same amount of quality.

So I'm running in circles, can anyone tell me the mistake in my argumentation or otherwise, suggest a (better) solution?

[zeb7]

If you have enough capture space , capture at 704x576 then use virtualDub to scale to 352x288
then if you feel the video is not clean enough use one of the noise reduction filters
and putting a 3x3 colour blur at the top will fix weak colour signals

If this doesn't work for you , tell me more about what you want to do
( VHS to Divx /VCD / or best format for Archiveing ?)

[Brian Ellis]

Baerchen

It will not take you much time to do an experiment: capture a 2 or 3 minute clip in 704 x 576, then in 352 x 576 and then in 352 x 288 (or the other way round if you prefer). After each capture, record the result in the medium you prefer. Carefully compare the results on the TV screen, looking for artifacts. Make sure you set the right field order for your system, though.

I am currently transferring/editing some 100 VHS-C cassettes taken from about 1990 through to when I bought my mini-DV in 1996 (yeah!, 6 years ago, already) onto SVCD. I capture in 704 x 576. In my case, I found a very distinct improvement in quality capturing at 704, probably because there was more data to work on when rendering to 480 x 576 MPEG2, even though, theoretically, the original video probably had a horizontal resolution considerably lower than 352 equivalent lines.

In other words, lose as few data as possible until the final lossy rendering.

FWIW...

[Baerchen]

zeb,

thanks for your replay.
My target is XVCD for playback on a DVD player.

Capturing at 704 x 576 would seem much exaggerated to much in regard to quality issues .. i mean .. it's VHS .. Besides that, I actually do not have enough HD space. Let me outline again my understanding problem:

1. Capturing to 352 x 576 gives me full details, but results in interlaced frames.
2. I have to deinterlace since VCD does not support interlace and I want to apply filters.
3. Deinterlacing leads to quality deterioration => why capturing interlaced ? => gives me full details => go to 2.

@brian: Thank you too. Yes, I will do that, but I was interested in the basics behind doing the way I described, since I don't understand it.

[zeb7]

352x576 (interlaced capture) to 352x288 (XVCD) :
There will always be loss of detail .
----------
352x576 (interlaced capture) to 352x576 (interlaced XSVCD) :
If this is what you are doing then in virtualDub use these settings:
3x3 TV
Temporal cleaner 0.5

http://home.earthlink.net/~casaburi/download/

-----------

Actually Brian or Doctor M would be able to describe interlacing better than I .
Here are some facts (almost)

704x576 at 25fps (interlaced) (is the same as) 704x288 at 50 fps(this is what get displayed on TV at 50hz)
50fps + Interlacing = 25fps interlaced
Interlacing means ( weaving 2 frames together )

Encoding to Mpeg2 has other problems which can be fixed later