Analog > DV > MPEG2 vs. Analog > MPEG-2.

[MrMungus]

Does the DV step in-between the analog source and the MPEG-2 file decrease the quality? I'd think it would.

[trevlac]

I'd say it depends upon the equipment and process....

Most Analog source has about 1/2 the color information as luma info.

DV and MPEG both have 1/4 the color info as the luma. NTSC DV uses a different subsampling than MPEG.

So for NTSC color, Analog > DV > MPEG2 is 4:2:2 -> 4:1:1 -> 4:2:0 This means there is some interpolation in the last step.

For NTSC color, Analog > MPEG is 4:2:2 - > 4:2:0.

Now are you going to see a difference? This probably depends on the source, any steps in between, the software, and the hardware you use (including how your view the results).

Analog ->DV -> MPEG is nice for people, because it tends to be easier.

[lenti_75]

if your sistem is slow...then analog->mpg is not an option, you'll get probably a lot of dropped frames.
the safe way, which is my way too, is analog->dv, then you can convert to whatever you want, and if the system is slower....no big deal, it takes longer, but no dropped frames...
have fun

[MrMungus]

I'd say it depends upon the equipment and process....

Most Analog source has about 1/2 the color information as luma info.

DV and MPEG both have 1/4 the color info as the luma. NTSC DV uses a different subsampling than MPEG.

So for NTSC color, Analog > DV > MPEG2 is 4:2:2 -> 4:1:1 -> 4:2:0 This means there is some interpolation in the last step.

For NTSC color, Analog > MPEG is 4:2:2 - > 4:2:0.

Now are you going to see a difference? This probably depends on the source, any steps in between, the software, and the hardware you use (including how your view the results).

Analog ->DV -> MPEG is nice for people, because it tends to be easier.

I was thinking more in terms of the compression.

Analog > DV > MPEG-2 entails both DV and MPEG-2 compression.

Analog > MPEG-2 entails only MPEG-2 compression.

I hadn't really considered this until recently (and after converting about a dozen tapes).

[Kevin abq]

Keep it simple. Eliminate as many post-processing steps as you can, because they add opportunities for sync problems.

You shouldn't have to worry about dropped frames unless your processor was made in the mid-'90s or earlier :P A fairly new PC should be able to handle a hardware encoder with no dropped frames whatsoever.

[trevlac]

I was thinking more in terms of the compression.

Analog > DV > MPEG-2 entails both DV and MPEG-2 compression.

Analog > MPEG-2 entails only MPEG-2 compression.

I hadn't really considered this until recently (and after converting about a dozen tapes).

Only if you capture Huffyuv or uncompressed does Analog>MPEG not include a loss due to compression. I believe DV is considered a better method of compression than MJPEG (which is what most people use for caps). Here is a technical write-up on the merrits of DVvsMJPEG. I'd say the compression ratio is a reasonable guide to the amount of loss. Anything past Huffyuv is a bit of loss. I believe DV, MPEG, and MJPEG all add additional techniques to what Huffyuv does, but they do use Huffman encoding.

2:1 Huffyuv is lossless (if you do not change colorspaces). 5:1 DV introduces compression loss, but probably more loss due to the subsampled color. 10:1 MJPEG is much worse.

All in all, it depends on your software, hardware, and source. One clear thing about DV is that there may be loss due to the chroma subsampling. This is in effect part of the compression.

Here is a nice summary of DV.
http://graphics.csail.mit.edu/~tbuehler/video/dv.html

[MrMungus]

Keep it simple. Eliminate as many post-processing steps as you can, because they add opportunities for sync problems.

You shouldn't have to worry about dropped frames unless your processor was made in the mid-'90s or earlier :P A fairly new PC should be able to handle a hardware encoder with no dropped frames whatsoever.

No no no. Dropped frames are not a problem. I'm wondering if the visual quality of the final output is affected by having the DV device between the source and the MPEG-2 file (since it introduces another compression step).

[MrMungus]

I was thinking more in terms of the compression.

Analog > DV > MPEG-2 entails both DV and MPEG-2 compression.

Analog > MPEG-2 entails only MPEG-2 compression.

I hadn't really considered this until recently (and after converting about a dozen tapes).

Only if you capture Huffyuv or uncompressed does Analog>MPEG not include a loss due to compression. I believe DV is considered a better method of compression than MJPEG (which is what most people use for caps). Here is a technical write-up on the merrits of DVvsMJPEG. I'd say the compression ratio is a reasonable guide to the amount of loss. Anything past Huffyuv is a bit of loss. I believe DV, MPEG, and MJPEG all add additional techniques to what Huffyuv does, but they do use Huffman encoding.

2:1 Huffyuv is lossless (if you do not change colorspaces). 5:1 DV introduces compression loss, but probably more loss due to the subsampled color. 10:1 MJPEG is much worse.

All in all, it depends on your software, hardware, and source. One clear thing about DV is that there may be loss due to the chroma subsampling. This is in effect part of the compression.

Here is a nice summary of DV.
http://graphics.csail.mit.edu/~tbuehler/video/dv.html

Thanks for the info. Reading. I guess I'll be doing some testing soon, also (and put it on my site).

[Kevin abq]

if your sistem is slow...then analog->mpg is not an option, you'll get probably a lot of dropped frames.

I was just making the point that the direct analog-to-MPEG route is probably better because it's simpler. One of the replies, quoted above, mentions a concern for dropped frames. That was what I was referring to. I understand you want quality first. That's why I think the direct approach minimizes the times the file is altered, and in doing that, minimizes the chance for lip sync trouble.

[MrMungus]

if your sistem is slow...then analog->mpg is not an option, you'll get probably a lot of dropped frames.

I was just making the point that the direct analog-to-MPEG route is probably better because it's simpler. One of the replies, quoted above, mentions a concern for dropped frames. That was what I was referring to. I understand you want quality first. That's why I think the direct approach minimizes the times the file is altered, and in doing that, minimizes the chance for lip sync trouble.

That's cool. I've never had a lip synch problem and just didn't want the thread to get off on a dropped frame tangent.

[Kevin abq]

MrMungus,
Cool. You're one of the fortunate few who haven't had sync issues then. Good luck

[cerburus]

Yes it does. I had started a project involving a HI-8 source conversion with a Canopus ADVC-100; when a friend showed me a analog to Huffyuv file. I captured a segment from my source with both methods and there is a clear difference in compression quality between the two. I just restarted the project using HUFFyuv. My source is from a budget HI-8 camcorder. A cleaner, sharper source may show less of a difference. If you have the drive space for HUFFyuv; go for it.

[lenti_75]

don't you realize that compressing into an MPEG file, in real time, you lose quality ? it's imposible for any sistem to render the image "real time" and keeeping the same quality.
to compress an AVI file into an MPEG file takes hours ( a good compresion) and you want to do this instantly....
it's simple as that.....

[MrMungus]

don't you realize that compressing into an MPEG file, in real time, you lose quality ? it's imposible for any sistem to render the image "real time" and keeeping the same quality.
to compress an AVI file into an MPEG file takes hours ( a good compresion) and you want to do this instantly....
it's simple as that.....

How do DVD Recorders and Tivo do it in real time?

[Kevin abq]

it's imposible for any sistem to render the image "real time" and keeeping the same quality.

My Panny standalone caps and burns DVD-R real-time off DSS and you can't tell it from a commercial DVD. What quality is it lacking that I am not aware of

[lenti_75]

i'm not saying it's not possible, it is, but not with the same quality.
and if you don't trust me, just do some research and you will see. maybe you can't tell by "eyes" but there is difference.

[MrMungus]

i'm not saying it's not possible, it is, but not with the same quality.
and if you don't trust me, just do some research and you will see. maybe you can't tell by "eyes" but there is difference.

So a DVD recorder gives worse quality than AVI > MPEG-2?

[curritch]

Here is my 2 cents.

I think it is a function of the quality of the mpg2 encoder since that is the weak link in the whole process. And some are quite bad if you don't have control of bit rate, motion detection, etc.

If your analog-to-mpg encoder is satisfactory then that's obviously the most efficient way to go.

But software encoders like TMPGEnc can provide most any quality mpg that you have time to produce.

I don't believe that there is any inherent reason that the DV step should degrade anything since frame compression has to occur somewhere. And there are definitely potential quality advantages to using a software encoder if your analog-to-mpg encoder is subpar.

[Thorn]

I think it is a function of the quality of the mpg2 encoder since

Yes! I am wary of direct-to-mpeg capture because many of the encoders are shoddy. More than that, the slower your CPU is, the more quality is thrown away so the CPU can keep up.

I prefer capping high res analog in huffyuv, then using TMPGenc to encode. This minimizes the sources of loss, given that analog source is already typically noisy.

That said, everyone has their own threshold for quality. If you can't see enough of a difference on a 60" tv to make you want to do things the hard way, why not go with what's easier.

No one here can tell you what will look good enough for you. Your source material, CPU speed, software, and even capture card all affect the quality. Try short clips typical of your source, then play them on the target machine. Or maybe burn to DVD then go to best buy and say you want to test out their big screen... because you might buy it . Then you can compare side-by-side and decide how much of a perfectionist you need to be.

[MrMungus]

I think it is a function of the quality of the mpg2 encoder since

Yes! I am wary of direct-to-mpeg capture because many of the encoders are shoddy.

This ADS Instant DVD unit I have supposedly has the same chipset as many DVD recorders.

[indolikaa]

don't you realize that compressing into an MPEG file, in real time, you lose quality ? it's imposible for any sistem to render the image "real time" and keeeping the same quality.
to compress an AVI file into an MPEG file takes hours ( a good compresion) and you want to do this instantly....
it's simple as that.....

I think I disagree.

MPEG is a mathematical process. So long as the hardware has the resources, horsepower and clockspeed to handle the number of calculations necessary between frame captures, how would there be an issue?

I mean, I can barely tell the difference between my TMPGEnc encodes and my ATI encodes. And we're talking a miniscule difference between a process that happens in realtime and a process that takes 12 hours per 1 hour of rendered video.

Do you have any links to some sources? I'd love to read that!

[bobbin]

Just some input from a hardware engineer.....

Our high speed PC processors are designed to be very broad in their scope of applications. They play games, pay bills, browse the net and collect our email.

Dedicated processors on the other hand are optimized for one purpose only; in this case processing video. The processor on your video card can't open a spreadsheet. On the other hand, it is an order of magnitude faster at video conversion, which is its only purpose.

So, what takes 10 times as long on a PC, can be accoplished in real time on a decicated video processor.

[proxyx99]

Amen. Until not long ago hardware encoding was the only efficient method of compressing to MPEG2. Since CPU's are getting faster software method is catching up (speedwise). Good HW encoder will do an excellent job, at least as good as software. Quality output is relative to amout of money you put in your solution.

[indolikaa]

Keep it simple. Eliminate as many post-processing steps as you can, because they add opportunities for sync problems.

Funny you should say that.

I'll be online with a Canopus DVStorm2 soon. I've been told this is the ultimate solution for MPEG A/V and one of the most stable platforms ever designed.

We shall see...

[Kevin abq]

Keep it simple. Eliminate as many post-processing steps as you can, because they add opportunities for sync problems.

Funny you should say that.

I'll be online with a Canopus DVStorm2 soon. I've been told this is the ultimate solution for MPEG A/V and one of the most stable platforms ever designed.

We shall see...

I have a lot of respect for Canopus. It seems all the vendors have come a long way in the last 2 years in the A/V sync arena. I hope your card is as stable as it's advertised to be.

[OmegaSupreme]

I use a storm 2 to capture direct to MPEG2 with highly impressive results. I have never had a dropped frame or snyc problems. Results will vary depending on the capture card.

[racerxnet]

I'll vouch for the Canopus products excellent capture and output to DVD. I get great results with 0 hassle using the DV REX.

[lenti_75]

don't you realize that compressing into an MPEG file, in real time, you lose quality ? it's imposible for any sistem to render the image "real time" and keeeping the same quality.
to compress an AVI file into an MPEG file takes hours ( a good compresion) and you want to do this instantly....
it's simple as that.....

How do DVD Recorders and Tivo do it in real time?

And...what is the quality ??? that's the question. You sure it's the best ?

as long as you don't realize...then it's the masterpiece....but see what other people with testings have to say....
and stop bugging me with that tivo and standalones, baby toys....

[MrMungus]

don't you realize that compressing into an MPEG file, in real time, you lose quality ? it's imposible for any sistem to render the image "real time" and keeeping the same quality.
to compress an AVI file into an MPEG file takes hours ( a good compresion) and you want to do this instantly....
it's simple as that.....

How do DVD Recorders and Tivo do it in real time?

And...what is the quality ??? that's the question. You sure it's the best ?

as long as you don't realize...then it's the masterpiece....but see what other people with testings have to say....
and stop bugging me with that tivo and standalones, baby toys....

I'm just playing devil's advocate. I actually am doing a lot of testing and I'll have the results of some tests on my site (eventually).

[Thorn]

Good hardware and software realtime encoders are hard to beat. The problem is, that most entry-level hardware and software for realtime cap is just plain mediocre.

Of course, it's getting better, and I probably wouldn't be the one to know when we finally pass the threshold. I do know that the software for realtime cap still wasn't good enough when I passed the 1GHz mark.

Sure, with a fast enough CPU and good software you should be able to cap great video in realtime. Is the software there yet though?

It annoys me it when people assume that just because software should *theoretically* do something that automatically every piece of software that comes out does it. Software is just like the rest of the commercial world - just because the technology exists to do it right, doesn't mean that the manufacturer won't cut corners and do it as cheap as possible.

Keep us updated - user feedback is the best way to know if a piece of software/hardware is up to snuff.