PDA

View Full Version : Can you justify using anything other than VCD for VHS backups?


FlimsyFeet
28th August 2003, 13:18
I've seen quite a few people saying they have backed up their VHS tapes to DVDR.

Now I know that VHS should be captured at full resolution for the noise filters, etc. to do a better job, but I was under the impression that there's no benefit in using anything higher than VCD resolution for the final copy.

Is it because, if the original tape is interlaced, then a format with only 240 lines will not retain the 3:2 pulldown and the VCD will end up with a jerky picture?

r6d2
28th August 2003, 14:48
This article may be of help. Get to the part that says "clarifying a myth":

http://www.dvdrhelp.com/forum/userguides/98177.php

Regards,

FlimsyFeet
29th August 2003, 13:32
That article is discussed in a thread (http://forum.doom9.org/showthread.php?s=&threadid=60313) that was posted a few hours after this one.

The opinions expressed in that thread generally agree with what I understood from previous discussions here: you should always try to capture in full resolution.

I'm also pretty confident that if you are capturing from VHS (or 8mm), there is no point in using anything other than VCD for the finished backup copy.

The problem I envisaged, (quoting myself):
Originally posted by FlimsyFeet
Is it because, if the original tape is interlaced, then a format with only 240 lines will not retain the 3:2 pulldown and the VCD will end up with a jerky picture? is not a valid reason. Captures from a telecined tape should be IVTCed to a progressive 23.976 fps video, because the VCD format supports "23.976 NTSC film". When this kind of disc is played on a standalone player, the player itself does the telecining automatically.


It seems to make sense to me, but PLEASE somebody tell me if I'm talking rubbish!

vhelp
31st August 2003, 21:14
Hi FlimsyFeet,

I don't think that there is anything wrong in what you wrote, but
as for VCD's, there is no support for IVTC, hence 3:2 pulldown won't
work.

VCD (MPEG-1) is stuck at 23.976 or else 29.970 ...
...but not 23.976 w/ 3:2 pulldown, making it 29.970 ...as that is
going to cause you studders and things :)

Hmm.. would be nice if there was such an app that could tell you what
your's Frame is consisiting of.. some ie:
* 23.976
* 29.970
* 23.976 w/ 3:2 pulldown (29.970)

This way, we know what's actually encoded !!

-vhelp

FredThompson
1st September 2003, 06:59
I'm also pretty confident that if you are capturing from VHS (or 8mm), there is no point in using anything other than VCD for the finished backup copy.
VCD isn't DVD-compliant. For VHS, I use CVD with 48 audio rate and a sufficiently high bitrate to not have artifacts. For 8mm I'd got with SVCD.

VCD is MPEG1, CVD and SVCD are MPEG2. MPEG2 is far more flexible and has better compression. Is your source interlaced? MPEG2 supports interlaced source, MPEG1 does not.

Any time you convert from interlaced source to progressive you will LOSE quality unless the source was film or you want double the frame rate. Regular interlaced source is NOT 25/30 half-pictures. It is 50/60 half-pictures with no true 1:1 correlation (because the scan lines record a different physical space.)

A decent software player will display interlaced source just fine on a computer monitor.

Only you can decide what's right for you. Me, I'd like to do it once and not have to do it again, nor have inferior quality just to save a few pennies of storage medium.

Try it yourself. Make a capture at full resolution and create 2 test discs, one full-disc VCD and one full-disc CVD. Filter the source the same way (but remember to separate the fields for the CVD then weave them together after the filter.) Play both and see which looks better.

fellaw
1st September 2003, 19:10
Sorry for me to barging in, but
VCD isn't DVD-compliant.
this isn't correct. VCD IS DVD-compliant! See here (http://www.mpeg.org/MPEG/DVD/Book_B/Video.html).
Try putting a VCD mpg into an authoring app without patching it.

Nevertheless, I don't agree with this article on dvdrhelp, especially the chapter "Clarifying a myth". I captured VHS via composite both at 768 and 384. There where no missing columns in the capture. Despites that, the higher the resolution, the more information you have for filtering VHS, e.g. TBC or CNR. As S-Video has better picture than composite, you can also apply this theory on it, can't you?

FredThompson
1st September 2003, 20:20
Sure looks like I was wrong. The audio needs to be transcoded but the video is supported.

Nonetheless, VCD isn't that attractive now. A few years ago it wasn't that bad and option and there are people who are some interesting ways it's been tricked out (KVCD, etc.)

Yes, S-video has a higher resolution than VHS. The deck I use is an upper-end JVC S-VHS with image stabilization and other filters that provide for a wonderful capture. Black is black, not swirling red and blue, for example. Better to start with a good capture than try to re-create data through filtering.

Take a look at the description again. It doesn't say an entire column will be missing. It's talking about missing data in a scanline. It seems reasonable to an extent but I've never seen damaged tape give horizontal streaks of color. Dropouts, yes, streaks, no.

TerraForce1
12th September 2003, 20:32
quote:
----------------------------------------------------------------------
I'm also pretty confident that if you are capturing from VHS (or 8mm), there is no point in using anything other than VCD for the finished backup copy.
----------------------------------------------------------------------

Well, I disagree. You only think in video. There is also an audio signal. Where with DVD you can go for Dolby Digtal or DTS. When properly encoded you definetely won't say there is no point in using anything other than VCD for VHS captures.

FredThompson
12th September 2003, 23:20
VCD is also low bitrate and unsuitable for lots of motion.

Darksoul71
12th September 2003, 23:20
@FredThompson:

Nonetheless, VCD isn't that attractive now. A few years ago it wasn't that bad and option and there are people who are some interesting ways it's been tricked out (KVCD, etc.)
That depends on your source. VCD or at least VCD resolution can be also "attractive". Esp. for VHS backup.

Have you ever tried to encode your old (and I mean OLD :)) VHS tapes to DVD ? For me it was obvious that no bitrate or resolution in the
world could anything to 2nd generation / LP-copies I owned from rare movies that will never find their way to DVD.
After a lot of experiments I found my personal solution:
-Capturing in a Full D1
-Deninterlace
-Heavy denoising
-Resize to VCD Res

Even MPEG1 can do "wonders" when you use above VCD bitrates and on TV when played back from my standalone, I mostly canīt tell a difference between those "ulgy-second-LP-copies" and VCD-res MPEG1. Often (may be due to denoising) they look even "better" than the original.

@FlimsyFeet:

Now I know that VHS should be captured at full resolution for the noise filters, etc. to do a better job, but I was under the impression that there's no benefit in using anything higher than VCD resolution for the final copy.

Not to offend you by any means, but why didnīt you find out by yourself ? I mean: Capture a movie in Full D1 and encode to VCD.
Do the same with a CIF (352x240) capture and encode to VCD.
Mostly youīll notice that the first movie will look much "sharper".

-D$

FredThompson
12th September 2003, 23:38
Yes, I have done that. The only time I deinterlace is if the source was film. This is because I work with NTSC. My understanding is there is progressive PAL source but I don't use PAL. Deinterlacing NTSC throws away half the temporal information.

If you deinterlace, filter heavily and don't see a difference from your source, your source is very poor. Many people try to remove the "look" to tape and, in so doing, destroy much of the source itself.

Granted, the quality of a lot of older recordings is...uh...mushy, but not all. If you're working with 1940s newsreels, VCD is ok.

vhelp
13th September 2003, 03:16
Hi guys..

We ARE talking about strickly Standard VCD ie,

* 352 x 240 resolution
* 1150 bitrate
* 224k audio
- - - - Standard VCD - - - -

or xVCD ie,

* 352 x 480 resolution
* 2150 bitrate
* 224k audio
- - - - non-Standard xVCD - - - -

w/ respect to above Standard VCD, this just cannot compare to VHS
source quality. VHS source quality containes both fields, and w/
sharpenss you can't achieve in Standard VCD format authored CDRs.

Say, you did capture from Standard VCDs (above) you'd have an even
more blurry source file, because your source's true (based) resolution
was 352 x 240.

But, let us say, you created a non-standard VCD (aka xVCD) ..
Then, you WOULD have a better quality source file captured from VHS.

We keep comparing VCDs to VHS but we are not noting the format specs!
So, VCD vs. VHS will loose, cause stVCD would have 352 x 240 res. but
an xVCD vs. VHS, now we're talking a better comparison, because we are
now incorporting both fields or x240 vs. x480 .. and the x480 would
give you the VHS quality.

So, the "myth" is somewhat invalid to a point, because we first have
to establish WHICH format we are using..
* Standard VCD or non-standard VCD (aka xVCD)

And, in all honesty, the "myth" would only have a relationship w/ the
x480 route.. not the x240. So, when you discuss "myth", make sure
you are talking about x480 (aka, both fields)

Hope this makes better sense of the topic :)
-vhelp

FredThompson
13th September 2003, 03:36
Yes, that's what I was saying.

If you want interlace, why are you chosing MPEG1?

For (S)-VHS source I save as CVD with audio at 48, not 44.1. It's DVD-compatible and support interlace.

FlimsyFeet
6th October 2003, 14:07
OK thanks guys for all your comments.

Since I first posted this thread I have been doing some testing. This is a summary of my current (simplistic) thinking - note of course this is my own subjective opinion.

VCD resolution of 352x288 is adequate for VHS or 8mm captures - yes, the picture is softer, but this softness helps to mask the problems you see when capturing a lower-quality analogue source. However - as Fred says - the low bitrate of VCD-compliant MPEG-1 causes undesirable blockiness during motion. I'll not be using VCD for captures I want to keep.

Personally, I'd like to avoid using xVCD or KVCD formats, my standalone player is pretty old, and even if it could play these formats (I've not checked) if I get a new player, these discs might not play.

Since I don't have a DVD writer, it looks like SVCD is the way to go (or maybe CVD - I understand this is a fairly widespread format?)

But I still have a lot to learn. I'm still struggling with dropped frames, and there are several things I need to look at to minimise that. Also would like to improve the source, I'm envious of Fred's JVC VCR - although when my Toshiba deck first came out at 300 GBP it was not exactly a budget model, I don't think it has the image stabilisation required for really good captures. Same goes for my old camcorder. Also I need to learn how to use AVIsynth and its filters.

But thanks anyway for all the comments!

Joe Fenton
7th October 2003, 06:27
Originally posted by FredThompson
Take a look at the description again. It doesn't say an entire column will be missing. It's talking about missing data in a scanline. It seems reasonable to an extent but I've never seen damaged tape give horizontal streaks of color. Dropouts, yes, streaks, no.

You generally only get the streaks at the start and end of the tapes for two reasons. First, at the start, the tape is under the most tension and close to a splice point. To avoid streaks and dropouts here, let a minute or two of tape pass before starting to record (when recording on tape of course). This doesn't effect commercial tapes much since that's where all the adds are and we skip those. :)

Second, you get streaks at the end of the tape because it's under the least tension and is again close to a splice point. Try not to record to the very end of a tape. Dropouts and streaks here aren't very noticable because this is usually credits. A streak here is lost in the mostly black picture.

Dropouts result from loss of magnetic material from the surface. Most tapes use granulated material for flexibility. As time passes and the tape is played, the material gradually falls off. This affects high-frequencies the most since magnetic penetration is inversely related to frequency. High frequencies don't penetrate very deep and are therefore the first to go. That's why older tapes look softer - they are losing the high end of the spectrum. The occasional streak in the tape is to be expected on home recorded tapes, even high quality ones. My tapes of Babylon 5 show an average of one streak (not dropout, a streak) every episode. This is on high quality SVHS tapes. Expect worse on low quality VHS tapes. Commercial tapes can be better, or worse depending on the batch. My Star Wars Ep 1 commercial tape is complete crap with streaks and dropouts all through the entire tape.

A high-end tape player will often have a built-in "de-streaker." It keeps a line of video in memory (part of the HQ circuitry) and when it detects a streak, it replaces the portion where the streak is occuring with the data from the line memory. That is why some people never notice streaks - their VCR eliminates them. I have one deck that does that and another that doesn't. There are filters for AviSynth to try to remove dropouts. Given a deck that removes streaks and the proper avisynth script, you can get a decent source from your tape.

I personally feel that whatever resolution you will eventually encode, you should capture at the highest resolution you can. Take an analogy from audio recording: the higher the oversampling rate, the lower the quantization noise. In video, high res capture doesn't give you any more info on a VHS source, but is does give you less noise from sample quantization. A simple low-pass filter can then be used to almost completely eliminate this noise without effecting the picture quality.