Should I capture VCR, 8mm and Hi8 tapes to MPEG2 or MPEG4?

[dorenhagen]

My understanding is MPEG4 is more efficient than MPEG2. Should I therefore capture directly to MPEG4?

Whichever method you recommend (MPEG4 or MPEG2) what capture device and workflow gives you the best quality for time and money invested? I am not going to edit other than trimming and would like to get to the point of diminishing returns in my approach. I do not need to squeeze out every last drop of theoretical benefit out of my tapes.

Thank you in advance for any assistance.

[edDV]

I recommend DVD standard MPeg2 at greater than 8000 Kb/s (aka 1 hour mode).

If you don't want to edit of filter, a DVD recorder is the easiest and overall probably least expensive solution.

Example
http://www.walmart.com/ip/Magnavox-160GB-DVD-Recorder-Tuner/10104532
http://www.walmart.com/catalog/product.do?product_id=15080509&findingMethod=rr

A hard disk simplifies cutting.

If that is too expensive you can try this with added risk.
http://www.walmart.com/catalog/product.do?product_id=15032475&findingMethod=rr

[dorenhagen]

Thank you edDV. Any particular reason not to go with MPEG4?

Also, does the Magnavox unit allow you to determine bit rate and other parameters?

[olyteddy]

DVD recorders usually have several pre-set 'quality' settings. These are rated typically by 'Hours' (1 Hour per DVD being the 'highest', 2 Hour 'standard' to 8 Hour 'unwatchable pixelated junk' modes).

[edDV]

And you should use the 1 hour mode for hand held, noisy camcorder source.

DVD MPeg2 is a universal exchange format and can be played in any DVD player, most computers and media players. DVDR blank media is cheap and can be easily copied.

MPeg4 is not universal and does not offer much compression advantage for noisy interlaced camcorder video. Deinterlace is highly destructive. So if you want MPeg4 for your iPod, do it separately using the DVDs as source.

There are other ways, but this is what I recommend if you don't want to edit or filter. Otherwise the learning curve and expense are much greater to do it all on a computer.

[PuzZLeR]

I would easily choose MPEG-2 over MPEG-4 for this task. MPEG-2 demands less processing, is better for edits (even with just cuts), and is better with decoders including DvD.

As well, many capturing devices that write to MPEG-4 are horrid.

Yes MPEG-4 is better at compression than MPEG-2 with everything being equal. But it isn't always at equal level with capturing. And at higher bitrates as well, which you should be capturing to, the gap closes significantly more, especially with interlaced material.

Besides, hard drive space is getting bigger all the time. Likely your VHS content is memory-filled - and if valuable, a few saved bits of data, if you CAN achieve this with MPEG-4 with capturing, shouldn't be the priority anyway.

[2Bdecided]

MPeg4 is not universal and does not offer much compression advantage for noisy interlaced camcorder video. Deinterlace is highly destructive. So if you want MPeg4 for your iPod, do it separately using the DVDs as source.

To be fair, properly deinterlaced video, encoded as 768x576p50 (for PAL 4x3) or 640x480p60 (for NTSC 4x3) to MPEG-4 using x264 at a decent bitrate will look far far better than 8Mbps MPEG-2.

Deinterlace doesn't have to be highly destructive - it can even be lossless if you choose (i.e. keeping all the original lines of video perfectly intact - so you can re-interlace and get the original back perfectly - though once you use and lossy coding nothing is perfect).

You're right that playback options will be far more limited.


More importantly, some paranoid people choose to keep a lossless copy of the original unprocessed capture. Others keep a "slightly lossy" copy to save space (e.g. 15Mbps MPEG-2, 25Mbps DV, high bitrate MPEG-4).

Whatever you do, the Hi8 footage will look better than the 8mm footage. Make sure you capture Hi8 via S-video connection. Get the levels right. Use a decent capture device. A decent DVD recorder at 8Mbps is more than good enough.

DVD-Rs aren't necessarily very good for long term storage though. The original tapes will probably last longer (though maybe not the machines to play them on).

Cheers,
David.

[dorenhagen]

Whatever you do, the Hi8 footage will look better than the 8mm footage. Make sure you capture Hi8 via S-video connection. Get the levels right. Use a decent capture device. A decent DVD recorder at 8Mbps is more than good enough.


Cheers,
David.

Do you have any opinion about using a miniDV camera analog to digital pass through for TBC?


Thanks for your very informative reply.

[jagabo]

many capturing devices that write to MPEG-4 are horrid.

Yes. Many of them perform a simple deinterlace, potentially losing half the spacial and temporal resolution.

[2Bdecided]

Do you have any opinion about using a miniDV camera analog to digital pass through for TBC?

Yes - if you have one with S-video input and a half-decent TBC it's an excellent route. Check the resulting levels are correct though - especially in NTSC land.

Cheers,
David.

[PuzZLeR]

many capturing devices that write to MPEG-4 are horrid.

Yes. Many of them perform a simple deinterlace, potentially losing half the spacial and temporal resolution.

Yes. Much of the horrid MPEG-4 captures, which include DivX, AVC, even WMV, are due to ignorance of proper settings on the part of the capture tool itself, not really with MPEG-4.

[PuzZLeR]

@2BDecided: Not disagreeing with all, but I find some of your points rather vague without explanation.

To be fair, properly deinterlaced video, encoded as 768x576p50 (for PAL 4x3) or 640x480p60 (for NTSC 4x3) to MPEG-4 using x264 at a decent bitrate will look far far better than 8Mbps MPEG-2.

Deinterlace doesn't have to be highly destructive - it can even be lossless if you choose (i.e. keeping all the original lines of video perfectly intact - so you can re-interlace and get the original back perfectly - though once you use and lossy coding nothing is perfect).

I do understand the result will be lossy though going from MPEG-2 -> MPEG-4 AVC. But I'd like to test your script for "proper deinterlaced video" with all lines intact, etc, and at for what "decent bitrate", that would be better than MPEG-2 at bitrate >= 8mbps.

As well, this doesn't insinuate capturing to MPEG-4, only encoding to it from the capture source.

More importantly, some paranoid people choose to keep a lossless copy of the original unprocessed capture. Others keep a "slightly lossy" copy to save space (e.g. 15Mbps MPEG-2, 25Mbps DV, high bitrate MPEG-4)... A decent DVD recorder at 8Mbps is more than good enough.

Maybe I'm taking this out of context, but really? A decent DvD recorder at 8mbps can indeed produce good results, but you believe it's "good enough"? Yes, the Law of Diminishing results does apply to higher bitrate MPEG-2, and even DV, using these bigger sized files - quality per incremental bitrate - but there is, at minimum, a bit more you can gain from them if you're willing to give up the extra storage space for them.

if you have one with S-video input and a half-decent TBC it's an excellent route.

I'm assuming you're not including VHS captures here. A full-frame TBC is highly more productive with VHS - I'm sure you realize that.

[poisondeathray]

But I'd like to test your script for "proper deinterlaced video" with all lines intact, etc, and at for what "decent bitrate", that would be better than MPEG-2 at bitrate >= 8mbps.

I think he's talking about QTGMC's lossless modes (lossless mode 1 , noise restore =0 ) +/- sourcematch , noise bypass etc...

My opinion that is sure the original fields are untouched , but the deinterlacing quality suffers - you get bob shimmer jaggies etc..., which IMO, defeats the purpose of a great deinterlacer

[edDV]

MPeg4 is not universal and does not offer much compression advantage for noisy interlaced camcorder video. Deinterlace is highly destructive. So if you want MPeg4 for your iPod, do it separately using the DVDs as source.

To be fair, properly deinterlaced video, encoded as 768x576p50 (for PAL 4x3) or 640x480p60 (for NTSC 4x3) to MPEG-4 using x264 at a decent bitrate will look far far better than 8Mbps MPEG-2.

Deinterlace doesn't have to be highly destructive - it can even be lossless if you choose (i.e. keeping all the original lines of video perfectly intact - so you can re-interlace and get the original back perfectly - though once you use and lossy coding nothing is perfect).

You're right that playback options will be far more limited.

dorenhagen wanted

"what capture device and workflow gives you the best quality for time and money invested? I am not going to edit other than trimming and would like to get to the point of diminishing returns in my approach. I do not need to squeeze out every last drop of theoretical benefit out of my tapes"

To me that meant quickest, easiest workflow to the desired result without much quality loss.

As said above, direct mpeg4 capture has serious quality loss issues. Capture to DV or huffyuv intermediates has serious learning curve issues, potential for error (e.g. levels) and a time consuming workflow.

Deinterlace to 50p/60p is an advanced craft, requiring knowledge of avisynth filters to do it right. The advantages over 50i/60i are small particularly for this kind of source.

More importantly, some paranoid people choose to keep a lossless copy of the original unprocessed capture. Others keep a "slightly lossy" copy to save space (e.g. 15Mbps MPEG-2, 25Mbps DV, high bitrate MPEG-4).

Whatever you do, the Hi8 footage will look better than the 8mm footage. Make sure you capture Hi8 via S-video connection. Get the levels right. Use a decent capture device. A decent DVD recorder at 8Mbps is more than good enough.

DVD-Rs aren't necessarily very good for long term storage though. The original tapes will probably last longer (though maybe not the machines to play them on).

I agree he needs a backup strategy. The DVDR disc should be replicated for backup (local and separate address) and I'd recommend hard disk backup as well (for archive and playback). The MPeg2 archives can easily be edited and transcoded to other formats for various device playback or web uplink but will play back directly on a home network.

[dorenhagen]

EdDV wrote:

To me that meant quickest, easiest workflow to the desired result without much quality loss.

Yes! That is exactly what I meant. Is the Walmart Magnavox DVD-R as good as the Toshiba and Phillips?

[edDV]

EdDV wrote:

To me that meant quickest, easiest workflow to the desired result without much quality loss.

Yes! That is exactly what I meant. Is the Walmart Magnavox DVD-R as good as the Toshiba and Phillips?

Search the forum archives using model numbers. There are many first hand comments.

There is little choice with ATSC digital tuners. Magnavox is a Philips brand.

[dorenhagen]

After a little research it appears the Magnavox 513/515 from Walmart is equivalent to the Phillips 3575. All three units provide excellent TBC and A/V Synchronization.

My question: Davideck on an older thread recommended the following approach S-VHS > MiniDV camcorder > Firewire > DVD recorder > S-Video > PVR-250

WHY?!! Why all the additional steps when he could simply go S-VHS > Phillips 3575 which would perform the TBC, A/V Sync and MPEG2 conversion? Since the Phillips 3575 is excellent at A/V Sync and TBC why the additional steps for a small incremental theoretical improvement?

Am I missing something?

Thanks for any insights.

[jagabo]

Maybe Divideck's DV camcorder has a TBC/framesync and his DVD recorder doesn't. Or maybe the camcorder's is better.

[dorenhagen]

Maybe Divideck's DV camcorder has a TBC/framesync and his DVD recorder doesn't. Or maybe the camcorder's is better.

Nope. His miniDV camcorder has marvelous TBC capability, but no A/V Sync. Still, why not just use the Phillips 3575 for BOTH TBC and A/V Sync and be done with it. The way I see it, he's introducing two superfluous steps, the miniDV pass through and the PVR-250.

Why?

He does say in this thread that the results are fantastic.