Sensible resolution for VHS captures

[Raga]

What's a sensible resolution for VHS capture if I want to archive them? Does half DVD resolution get the most out of, or should I go with full? Since there doesn't seem to be a clear-cut answer for an exact equivalent of the VHS analogue signal is in pixels, what'd be an educated estimate? If I'm not mistaken, capturing in anything much bigger than that would blur the image.

[gshelley61]

Full D1 (720x480 in NTSC land) gives you the best picture quality and retains the most detail. However, many feel that half D1 is adequate for VHS transfer. There is a visible difference between the two, with full D1 being a little better, but it's a small difference. The difference is more noticeable on large screens.

[Raga]

Any thoughts on the blurring effect with more interpolation?

[greymalkin]

i have one angle to add to this..i noticed when I capped my wedding video I would get macroblocks when trying to do full 720x480....even at a bitrate of 8mbps..352x480 seemed to take care of that...with it's lovely "softening" effect (aka blurring)....

[gshelley61]

When capturing hand held camcorder footage directly to MPEG2 at full D1, if you exceed 1 hour per DVD you will likely have macroblocks and other compression artifacts. This is due to the constant image movement inherent with hand held video footage (eats up lots of bitrate). If you are using a computer, you might try capping at 15Mbps then re-encoding to compress.

[thecoalman]

i have one angle to add to this..i noticed when I capped my wedding video I would get macroblocks when trying to do full 720x480....even at a bitrate of 8mbps..352x480 seemed to take care of that...with it's lovely "softening" effect (aka blurring)....

That's a system or encoding problem though....

[lordsmurf]

Read this:
http://www.digitalfaq.com/dvdguides/capture/understandsource.htm#analogsource

352x480 is perfect for most sources.

[junkmalle]

Full D1 (720x480 in NTSC land) gives you the best picture quality and retains the most detail. However, many feel that half D1 is adequate for VHS transfer. There is a visible difference between the two, with full D1 being a little better, but it's a small difference. The difference is more noticeable on large screens.

I agree with this. But it also depends on your capture card. Some cards do better than others at half D1. I remember seeing caps from a Philips chip based device that looked as good capturing at Half D1 as capturing at full D1 and reducing later (and as good as any other card doing the latter). Best to do the experiments yourself with your hardware and software and then decide.

And if you're really serious about quality you'll have to spend about $300 on a good SVHS VCR with noise reduction and a line based TBC, and another $300+ on a full frame TBC.

[vhelp]

TGIF everyone

Yeah, but the problem with 352 x 480 is when you finally get
a large screen tv. That's when you realize *it is* better to
capture 720 x 480 and *KEEP IT* at that resolution.
.
If you try to inlarge a 352 pixel width to a large screen tv's
dimension, the changes of one seeing artifacts is much greater
than, if you were to start out with a 720 pixle width and *keep it*
and then view it on a large screen tv to compare w/ 352 width
MPEG's.. weather source is VHS or not.
.
As was stated in so many words above, you will notice the difference
when you move up to a larger screen tv. I don't think that a
good interpolation is going to save a 352 width though, for large
screen tv's. That's why its so important to start and end with a
720 pixel width.. and use a high'er bitrate. And, just because
one's source might be from VHS does not mean, "..because it's VHS,
it doesn't really matter.." is just plain false, IMO. In fact, I
would say that *its more important so* that you use the highest
[720] pixel width possible.
.
Another side to this issue is with respect to bitrates and per/disk
projects. Many try to squeeze in so much data onto a 4.3g disk,
and add to that, that they probably understand that because most
Commerical DVD's are 2.5 hours in length, that *they* should be able
to obtain the same quality from their encoding projects if they
do the same, using the same or less (say aprox) 2 hrs onto a dvd
disk, and expecting DVD quality. But, what is missing here, is
the sort of illusion going around. You see, most Commercial DVD's
are DL, holding aprox 8 gigs of video data. So, in understanding
this, a user will try and put 2+ hrs onto a 4.3 gig disk thinking
that they should obtain the same level of quality, but miss that true
mark here. That's why I say (make suggestions) that for 4.3 gig
dvd disks, you should shoot for 60 minute dvd size. Remember, that
most everyone here wants "quality".. even when they miss the true
mark (read above) here.

So, for quality, the receipe (IMO) should be:

* For 4.3 gig disk, plan for 60 min. per disk, at high bitrate,
though prefereable example scenario of: CBR / 9000 bitrate. You
can tweak the bitrate to your given source S/N/R.

* For 7.8 gig disk, plan for 2.5 hrs. per disk, at high bitrate,
again, though prefereable example scenario of: CBR / 9000 bitrate.
But, you can tweak the bitrate mode with VBR, depending on your
MPEG encoder and your level of skills/experience, etc. and per your
sources S/N/R.

Remember, I said per MPEG encoder, because of the differences in
the methods used for the compression/encoding algorithems performed
by these encoders.

Note..
If quality is not your main concirn, and but size/per/disk is, then
use a VBR method for fitting onto one 4.3 gig disks. But be warned,
that quality results will not be matched to that of the original
source. In other words, expect to see macroblocks (pixelation) in your
final MPEG's.

-vhelp 3435

[thecoalman]

<everthing above>

That's not concise enough. Could you go into a little more detail.

[Raga]

If the actual amount of data you get off the VHS is much closer to half-D1, then what difference would it make if one captured in half-D1 and it was interpolated on the screen, and if one captured full-D1 and it was interpolated while being captured? The latter obviously eats more space.

[junkmalle]

If the actual amount of data you get off the VHS is much closer to half-D1, then what difference would it make if one captured in half-D1 and it was interpolated on the screen, and if one captured full-D1 and it was interpolated while being captured? The latter obviously eats more space.

Read up on digital signal processing. It's not easy...

[SatStorm]

I'm not agree with vhelp regarding this subject.

A good 352 x 576 mpeg 2, when the source comes from VHS tapes or DVB broadcasts (from non 704/720 x 576 channels) looks as good as a good 720 x 576 on a big TV Screeen

If it does not, then there are other reasons. Usually the problem focus on the cheap mpeg 2 decoders all manufactures use in general ont the DVD players. It's all about the cost

I'm able - because of my job - to test Hi-End LCD/Plasma screens combined with Hi-End DVD players. On those -impossible to buy in real life- units, a 352 x 576 vs a 720 x 576 from the same VHS source, restored the same way, encoded with the same encoder, look identical

On a "mainstream" LCD screen with a "typical" DVD player, 720 x 576 looks noticeable better 352 x 576.
Is that a fault of the framesize? No!

[lordsmurf]

Good point. Your typical $30 DVD player has a rather poor decoder in it, compared to one that runs about $75-150 or so.

[Raga]

In other words, current mainstream decoder quality aside, the basic question here is whether I do the interpolation at my end or whether the end-user does the interpolation at his end. Correct?

[edDV]

In other words, current mainstream decoder quality aside, the basic question here is whether I do the interpolation at my end or whether the end-user does the interpolation at his end. Correct?

It mainly depends on how the capture card actually processes 352x480/576. Some cards (cheaper BT cards and others) capture to square pixel 640x480 and then interpolate to non-square pixel 720, 704, 352x480/576. The quality of the hardware and/or software interpolation filtering can be awful. Drivers are often written by clueless computer geeks not video engineers. Even when the hardware supports filtering, driver writers may take shortcuts and not use the filters properly.

In theory, 352x480/576 is enough for VHS but in practice 720 may give a better result depending on capture device and encoding software.

[junkmalle]

For what it's worth:

I printed out some black/white vertical bar test patterns. I used an old B/W analog video camera to film it, recorded on a crappy old VHS deck (composite cable). I then captured from the VHS tape (composite cable) to my Hauppauge WinTV PVR-250. This is a hardware MPEG encoder device. I captured once at 352x480, and again at 720x480, both at 15,000 kbps CBR.

Here are some 1:1 crops of a typical frame (exact same frame from both captures):

352x480


720x480


The top bar represents about 310 lines across the full width of the image, the middle bar 295, and the bottom bar 280. Both images were drop frame deinterlaced to remove any interlace artifacts (I was holding the printed image by hand). All lines were clearly visible when the camera was connected directly to a TV.

[SatStorm]

I own Hauppauge 250 PVR and never really like it. It produce 'quality" picture only of full CIRR-601

We miss some points here:

Capturing any kind of source to your PC, is only a first simply step of the chain. Overall, capturing the higher framesize you can, can be from overkill to neccessary.

A second VERY neccessary step, is filtering. Not for the things we see, but the encoders do see. Because of the nature of mpeg, it is better to have average values of areas. Less macroblocs that way. A minor filtering, not affecting visually the picture, helps a lot later during the encoding

A third step is the correct resizing method. Do the wrong resizing and you have a F#$%@ result

Then, it is the encoding part. The technique you use, the encoder you prefer, etc

Finally you have your mpeg from your VHS. You author / burn it and you believe it is over

Wrong!

To manage to watch your 352 x 576/480 file the best possible way, you need to have a DVD standalone player with good mpeg decoding parts, and a good CRT TV. But we love those LCD / Plasma screens, right? Well, then, we need an LCD / Plasma screen, capable to "emulate" what CRT TVs do. And this is something technology has not yet manage to do.
Plasma screens still look awfull compare CRT TVs. And LCD screens only recently manage to present technologies that actually help the situation with sources from analogue to digital convertions (DVDs from VHS tapes, etc). Imagine: Those "Flat" TVs, have to emulate stuff "wrong" for the digital world, like interlace, overscan, the missing fields for broadcast, the phosphor of the CRT TVs we so much are used for decades... We expect to see on a flat screen even the ...flaws we so much "love" to see on a CRT screen, and that is impossible...

The EASY alternative to all this mess, is very "simply".
Just capture and encode the higher possible framesize with the higher possible CBR bitrate. That way, your mpeg encoder do the job easy, the DVD standalone decoders do their average job, the "flat" screens don't have to emulate anything, the bitrate is more than enough, etc...

I see all this as a hobby, so the "hard" way, is a challenge for me. But when you simply wish to grabb and encode some VHS tapes without becaming an "expert" on this, better follow what most people do or buy a DVD standalone recorder.

I hope I helped a bit...