How many minutes per DVD without quality loss?

[cinclodes]

I have been archiving video to DVD since 2003. Back then, I was told that the rule of thumb (using iMovie and iDVD) was that about 80 minutes could be put on a DVD without any noticeable loss in quality (based on my experiences, that seemed to be about right). Going beyond 80 minutes, you would still have high quality that might be perfectly acceptable for some applications, but you would start to get imperfections that might not be acceptable for archiving purposes. I'm sure that encoders have improved over the years, but by how much? I'm curious what are the current rules of thumb for the current generation of encoders. My work involves sports events in which the camera often pans horizontally -- this type of movement is the first to show signs of degradation when you get greedy. I know several individuals who think that you should never go over about 60 minutes. I want to preserve quality, but 60 minutes seems to be over-kill.

[hech54]

30 to 70 minutes
https://www.videohelp.com/dvd#comp

"Archiving" means 100% quality....IE...you should never even consider changing or lessening the quality
by worrying about movie/footage duration. A 1 to 1 copy is the only acceptable answer. A movie that
spans across two layers of a dual layer DVD gets "archived" to either another dual layer DVD or it is split
to fit two single layer DVDs.

[guns1inger]

Unless your source was DVD compliant MPG-2 to begin with, and all you did was author it to a DVD structure, the answer is none. Every time you encode using a lossy compression, you lose quality. Your DVDs are not archival. They may be good quality backups, but they are not the original video if you had to encode to get there.

[cinclodes]

Thanks for the suggestions, but I'm not talking about archiving the Dead Sea Scrolls or the Mona Lisa. I'm talking about projects that are routine but important enough for quality to be an issue. I'm interested opinions on the maximum number of minutes that can be put onto a DVD without a significant loss in quality. I realize that such opinions may be subjective and may depend on the software as well as the encoding time one is willing to invest. But it would be interesting to hear any opinions or thoughts on this issue. I would be particularly interested in how encoders have improved over the past several years and how different encoders compare.

[jagabo]

It really depends on the nature of the video. You can fit many hours of a noiseless slideshow with no transition effects. Only an hour of noisy, shaky handheld video. MPEG 2 encoders haven't changed much in the last 5 years.

If you're just talking about using DVDs as storage -- you can double or triple the amount using h.264 encoding.

[lordsmurf]

You're limited with iMovie/iDVD too, so the goal of ~60-70 minutes is best in this case.
I would not go more than 70 minutes, personally.

[PuzZLeR]

I'm interested opinions on the maximum number of minutes that can be put onto a DVD without a significant loss in quality.

I was pretty much going to say something similar to what Jagabo said - it really depends on the Source. This can vary from, like maybe, 50 minutes of high quality video to 3-4 hours.

To add, this is why I consider an encoder useless without a Constant Quality mode. Using Constant Quality, and the quality level you give it, will indeed reveal your answer as to just how much your Source would need, whether it's 45 minutes, 1 hour, 1.5 hours, or more, or whatever.

I realize that such opinions may be subjective and may depend on the software as well as the encoding time one is willing to invest. But it would be interesting to hear any opinions or thoughts on this issue.

Yes, this is opinion, but few would disagree with me. Two of the best MPEG-2 encoders are CCE and H Enc. Other than if you're using filters, the difference between any faster or slower DvD compliant settings is quite immaterial IMO with MPEG-2 encoders at this point.

I would be particularly interested in how encoders have improved over the past several years and how different encoders compare.

Very little other than maybe speed, and much of that speed increase is due to optimization across multiple threads. MPEG-2 is incredibly mature and has reached the Law of Diminishing Returns at this point in its timeline.

[cinclodes]

Thanks for the inputs!

[ronnylov]

There is also a maximum bitrate of DVD-Video that cannot be exceeded if you want it compatible with the DVD-Video specifications. A single layer DVD at maximum bitrate 10.08 Mbit/s will give the absolute minimum duration of time that you can fit at the highest possible DVD-Video compatible quality. My calculation gives 62 minutes at 10,08 Mbit/s on a single layer DVD but I guess there are some muxing overhead and other information that may need to be stored so something like 60 minutes is the limit where you don't have to lower the quality from the maximum possible DVD-Video quality.

But VBR encoding can remove bits from parts where there is no gain so practically you can get a little bit more than 60 minutes with top quality.

[PuzZLeR]

There is also a maximum bitrate of DVD-Video that cannot be exceeded if you want it compatible with the DVD-Video specifications. A single layer DVD at maximum bitrate 10.08 Mbit/s will give the absolute minimum duration of time that you can fit at the highest possible DVD-Video compatible quality. My calculation gives 62 minutes at 10,08 Mbit/s on a single layer DVD but I guess there are some muxing overhead and other information that may need to be stored so something like 60 minutes is the limit where you don't have to lower the quality from the maximum possible DVD-Video quality.

But VBR encoding can remove bits from parts where there is no gain so practically you can get a little bit more than 60 minutes with top quality.

You do raise an interesting point. Although this is highly subjective, and really depends on the Source, I would say a CBR encode of ~10mbps = a VBR encode of ~8.2mbps for similar quality.

Again, there are many variables that can play a role. But you can indeed get more out of a VBR encode, or even a CQ encode, at the same bitrate than you could from a CBR encode.