Help with VCR -> CD conversions

[guthrie]

We are trying to transfer S-VHS video tapes to CDs, and are new to this, so would like some expert advice!

* Goal:
We have a number of lectures, from 1-2 hours each, in SVHS source. We want to replicate the quality of VCR viewing, using CDs on a computer. We do not currently want to do editing, only transcription.

* Situation:
We bought a Matrox G450eTV card, and are digitizing to MPEG2, then converting to AVI, and then to RealVideo (RV) format. We need more capacity per CD than raw MPEG2 will fit.

We are using high resolution (704x..) because we found the quality was too low with low res (320x240). We also found that the Matrox card would drop frames at HiRes even with a 1G P-III and 512K memory, 20MB IDE disk. They said it 'should' work, but ... "all systems vary".

[We are now getting a faster (AMD 1.2G dual Athalon DDR 512M) motherboard, so it should solve the frame drop problem.] For now we dropped back to LoRes capture; but even there get abou t2-5% frame drops.

* Problem:
But, we still find that the final quality of the images if very poor. Doesn't matter if we start with Hi or LoRes data. During capture even the LoRes looks pretty good! Is this intrinsic to RV format? Are we taking the best path for this? Recommended target formats (DivX, ASF, SMR, ..?) Seems like this should be a FAQ; any /all recommendations welcome!

* Specifically;
-- does the capture board matter much in this? Ours does MPEG encoding in software, so does require a fast CPU; but are there quality differences as long as they can keep up?

-- Are we chosing best formats for capture and encoding? Is Quicktime good?

Thanks,
Gregory Guthrie
guthrie@mum.edu

[kinneera]

If you need good quality, high resolution, for computer playback only, you're best bet is definitely Divx. You'll be much better off capturing to uncompressed AVI, or a lossless codec like Huffyuv before converting to any of the highly compressed formats.

[kinneera]

Oh yeah...and even your high-end system is probably having frame drop problems if you are trying to do realtime capture to MPEG2. If you're going to try that, you should at least use a very high bitrate and I-frame only capture.

[holistic]

kinneera has nailed it. DivX is your best bet.
If harddrive space permits capture in AVI.

There is a good load of useful info here :
http://www.divx-digest.com/


Suggestion : Capture at 640*480
Encode Mpeg4(DivX) 3.11 codec @640*480
Make a auto play CD that offers to load the codec if needed, the have it resize your screen to 640*480 before playing file .

[guthrie]

Thanks to you both, we'll try this ASAP.

What do you mean by "use a very high bit-rate, and I-Frame only"?

[holistic]

Bitrate is the 'heart and soul' of digital video

To show an example

VCD uses a fixed (Constant) 1150 video bitrate where are a typical DVD may have a VBR (Variable) bitrate from 4000 to 9000. The more bits allocated per unit of video the better the image. Thus the higher the bitrate the bigger the file. Making digital video is a compromise of file size VS bitrate.

Am happy to answer all questions , but please learn to do your own research . It is fun and you will learn so much more.
http://www.google.com [search engine]

I,P,B frames are part of the GOP (group of pictures) that make the 'mpeg' experience

Read on .............

http://www.cisco.com/warp/public/614/20.html

[guthrie]

Thanks.
We have done LOTs of searches! but there is a lot of context and details here. Your notes help to tie all the parts together.

I get the main ideas of bitrate, etc, but our digitizing tools and card do not have parameters directly for these. They have options for resolution, and in some modes timing of reference frames. I presume that the actual encoding bit-rate is a derivative from these. The reference frames seem to be the same as the I-Frame terminology.

I've seen a spec of how often to take a refernce frame, but never an I-Frame only mode, seems like this would loose some of the value of predictive compression methods.

[kinneera]

Well, you're correct... capture I-frame (or reference frame) only eliminates a majority of the compression employed by MPEG, it's sort of like a psuedo-MJPEG capture. The idea is that it is not your final product, but rather a stream to be re-encoded at a lower bitrate with full GOPs after you've captured the video. The goal is to allow your computer to focus on capturing, since compression is also very processor intensive. Doing both simultaneously is likely to cause significant frame drops (as you discovered) and on-the-fly compression is usually much poorer in quality.

[guthrie]

Very interesting, thanks.

We have an infinitely fast CPU now (well, nearly..!), so will try encoding directly to DivX.
Or, to AVI, and then convert. By your note the second might be better; presuming we use different codecs?

I'm just looking into VirtualDub, seems like it gives a lot of options, and may do the task.

Thanks!