Log in

View Full Version : How do I get a VHS tape to AVI?


Semmemon
25th August 2002, 00:10
There are probably posts about it but...I'm still somewhat :confused: .

I want to take a VHS tape get that content on the computer because I haven't found any single picture about the show and it's just a cool show that I might make a website for. :D What exactly do I need and how much money do you figure I'm going to need to get it all?

What the recommended things I should buy; ie: What the cheapest digital camcorder that will do what I want.

avih
25th August 2002, 01:15
if you want to start experimenting, find the cheapest tv-tuner card you can find ($10-$50 depending where u look), preferably brooktree (= connexant) 878 or 848 based chip to keep it as standard as possible (that's most cards out there). since vhs quality is relatively poor, it's most probably that any tv-card u buy will be more than addequate for the task. naturally, you'll also need a standard vcr for the playback, and some cables to connect it to the tv-card (some cable might come with your tv-card, and some u might have to buy for few bucks).

basically that's all u need. the rest if freeware.

and start reading the guides, starting with the capture faq of this forum.

good luck ;)

ps, next time DO start with reading the faq, that's what it's for, and you prob guess that your question is a pretty standard one, right? ;)

Semmemon
25th August 2002, 05:48
Yeah, well I did read the FAQ's and the threads about VHS to DVD..

But you say I don't even need a digital camcorder?

BaronVlad
25th August 2002, 08:32
No you dont, you just need a VCR (to play the tape) connected with your computer (videoIn of your graphic accelerator or a cheap TV Card as avih said).
Now you are able to capture, it depends on your computer, your time and how much quality you want to have, in which resolution etc you capture. But - as avih said - the FAQ should be a quite good starting point for all the questions. ;)

bb
25th August 2002, 08:44
If you want to use a TV card for video in, then make sure the connectors support video input. My Hauppauge WinTV PCI has an S-Video connector, but there seems to be no way to input a signal. Output works fine, though. The only way I can input video in this TV card is through the antenna connector - but this way I loose quality.

bb

wotef
28th August 2002, 12:33
i have settled on an involved but good procedure for dealing with my old pal tapes

i use svideo-out from my panasonic svhs to svideo-in on my wintv fm pci

i make TWO captures to huffyuv 640x576 using virtualdub
read http://www.lukesvideo.com/

then i use an avisynth script to

first layer in some noise for dithering on one of the caps
read poptones guide http://arstechnica.infopop.net/OpenTopic/page?a=tpc&s=50009562&f=67909965&m=4020952194

then layer the other clip in on top of that
read poptones guide http://forum.doom9.org/showthread.php?s=&threadid=28438&highlight=averaging

then make the video progressive with deinterlace smooth
and apply some final denoising with convolution3d
read www.100fps.com
read http://forum.doom9.org/showthread.php?s=&threadid=32243

then 2 pass xvid with 160kpbs mp3 audio

i'm very pleased with the results and the trade-off with processing time is OK cause i do care about the shows i'm transferring

in terms of hardware, definitely get an s-vhs video and a quality s-video cable!!!!!!!

Ookami
28th August 2002, 15:51
Wotef, it seems I really should be reading the Avisynth forum more :) .

Especially the averaging seems very interesting (and time consuming :( )...

Thanks, for your posting.

BTW, why do you capture at 640x576? Why not full PAL? I've started to convert all my captures to SVCDs, as I don't have to deinterlace and I have a standalone supported format... Your thoughts?

Cheers,

Mijo.

avih
28th August 2002, 16:13
iirc, bt chips do the horizontal resizing adequetely (i.e. probably bilinear), so it shouldn't matter whether you capture 720x576 and resize to 480x576 or cap directly to 480x576, since the chip should resize quite well.

regardless, i've read on many places that practical vhs horizontal res is about 480, so it doesn't harm if not capping full pal.

of course, if you intend to filter then sometimes it's better first filterring, and then resizing, and then you'll have to cap at higher res than 480x576.

wotef
28th August 2002, 23:29
hi Ookami, i am still new to the forums in general but i have learned loads of unusual things in the avisynth forum

there's certainly lots of expertise here and there, though!

i would characterise the difference as you indicated; there are many folks that like to cap to svcd / to play on standalone, master of arcane gop structures and stuff like this

and then there is another crowd that likes to work in the pure avi space, who are very technical, with lots of developers and people who mess around with advanced filters

specifically, re: layer, actually it is a much faster and pragmatic way of denoising than any temporal or spatial filter - i really recommend you try it on a short clip for proof of concept and i would be happy to provide a walkthrough!

re: 640 - well, i settle on resizing to 640 x 480 so don't see the need to waste time on the horizontal resize (already scaled up by the wintv card)

Ookami
29th August 2002, 03:44
@avih

>iirc, bt chips do the horizontal resizing adequetely (i.e. probably bilinear), so it shouldn't matter whether you capture 720x576 and resize to 480x576 or cap directly to 480x576, since the chip should resize quite well.

BT chips don't even do the resizing correctly (aspect ratio wise)... More infos (german):

http://www.dvd-svcd-forum.de/phorum/ikonboard.cgi?s=3d6cde380a6bffff;act=ST;f=22;t=155

Adequatly? AFAIK, only the new Radeon cards are doing HQ hardware resizing (bicubic?).

> regardless, i've read on many places that practical vhs horizontal res is about 480, so it doesn't harm if not capping full pal.

Hmm... Time to dig in my articles and saved discussions. I've read far too many contradicting things about such things.

> of course, if you intend to filter then sometimes it's better first filterring, and then resizing, and then you'll have to cap at higher res than 480x576.

:)

@wotef

LOL, nice groups you have there... I would call myself "the master of the three dots" ;) .

I've switched a few days ago to SVCD, as I don't want to convert my captures to non-standalone formats anymore. Altough I don't have a standalone, it is more useful to convert to such formats. I will also check into the ISO MPEG 4 stuff soon...

-layer (averaging) I will try it, but I fear that the dropped frames could cause some trouble.

Cheers,

Mijo.

wotef
29th August 2002, 15:28
dropped frames, between two caps - not really been a problem for me -

i tend to drop about 4 frames for every 45 minutes of the old shows on vhs, and they always tend to be duplicated rather than completely lost

i check the dupes out in virtualdub - if they're on a scene change, then i'll use avisynth trim(blah,blah) to insert the clean frame from the other cap, problem solved

if the dupe is not on a scenechange or it's a fast motion scene, i'd never humanly see it anyway and so i continue to layer in on top regardless

this is what i meant when i said layer was fast; as long as clip1 and clip2 starting frame zero and end frame are identical, and assuming you only have a handful of dropped/duped frames, you can get right on with it

here's an example - the key thing is syncing "frame zero" and the "last frame" over the different caps:

let's assume "frame zero" in alpha.avi is actually frame 1, and the "last frame" i want in alpha.avi is frame 11

but that the equivalent "frame zero" in beta.avi is frame 3, and the "last frame" i need in beta.avi is frame 13

clip1=avisource("alpha.avi").trim(1,11)
clip2=avisource("beta.avi").trim(3,13)
clip3=clip1.layer(clip2,"fast")
return clip3

that's it, hope it's clear

ronnylov
6th September 2002, 14:19
I tried the layer method at a VHS capture (704x576 25fps PAL huffyuv) and it gives great results! But I have some problems with dropped frames but it was not hard to find where the frames were dropped. As soon as a frame drop occurs the output of the averaged clips have serious ghosting. But with trim it was possible to correct it. I guess I could capture without audio in AVI_IO but I think the audio can be useful.

Now I wonder if this method can be applied to the audio also? My old VHS tapes are noisy in the audio so averaging two audio tracks may decrease the noise. Can this be done directly in AviSynth or do I need a sound editor? How do I average two wav files in Cool Edit?

ronnylov
6th September 2002, 18:07
Originally posted by wotef

first layer in some noise for dithering on one of the caps
read poptones guide http://arstechnica.infopop.net/OpenTopic/page?a=tpc&s=50009562&f=67909965&m=4020952194


Can you explain how you can remove noise by adding noise.
I don't understand the theory behind this and the link did not give me the answer.

Thanks!

ronnylov
17th September 2002, 08:42
Now I have completed my first VHS - DVD conversion. I got great quality by averaging 3 identical clips (I captured the video three times and averaged the frames into one clip with avisynth). I captured at 704x576 resolution 25 fps (PAL). The averaged clip was denoised with STMedianFilter and cnr2 filter in avisynth and resized to 352x576 resolution (with 16 pixels black bars at sides, top and bottom). I encoded to MPEG-2 using Canopus Procoder at mastering quality 2200 kbit/s avergae video bitrate, 3100 kbit/s maximum to be able to fit 2 movies on one DVD-R.

It was not possible to average the audio. I used the audio from one of the captures and encoded to 128 kbit/s mono MPEG audio. I noticed that I got less frame drops if I captured in stereo and afterwards converted it to mono in the post processing compared to capture in mono directly.

The result was great! It looked like the original but less noisy.
So I have to say that averaging multi-captures is a good method to get rid of the noise! It was a little bit tricky to synch the clips because of some frame drops but after a while it was easy when I got used to the method. But it took long time. I need 6 hours to capture 2 hours of movie three times and then I need almost the same time to synch the clips and then the computer need several hours to save the averaged clip and add the encoding time to it also... I think I'll use this method only on old home videos I want to convert to DVD.