Best MPEG2 to DVD encoding strategy...what losses occur at each step?

I am transfering VHS to DVD. I capture using PowerDirector 5. I want about 100 min per DVD-5. My goal is to find the encoding/re-encoding/authoring/shrinking sequence that results in the best quality possible in the final DVD. My ignorance of what is going on during encoding and re-encoding leads me to ask this question.

First I capture the VHS to MPEG2 using an absurdly high bitrate (10,000kbps) at the highest “quality” setting…7…that PowerDirector allows, to be sure to not introduce any MPEG compression artifacts at this stage. But when PowerDirector authors the files in the final step, there is no longer a quality setting, only a choice of HQ, SQ, etc. Maybe I am losing the advantage of the quality setting 7 at this step?

I can think of two strategies. Either (1) I reduce the capture bitrate (but keep the high “quality” setting 7) so that the resulting files will be small enough to fit on DVD-5 without further compression, then author with PowerDirector; or (2) I keep the capturing at 10,000kpbs/quality 7, have PowerDirector author files at HQ which will be too big for a DVD-5, then use DVD Shrink to reduce them to DVD-5 size. My question is: Which strategy is better?

I have a bias (which I can’t support or defend) that DVD Shrink will do a better job of compression than PowerDirector, so my bias is to follow strategy (2). But I have a nagging doubt: Maybe there is something valuable about this quality 7 encoding which would have been kept in strategy (1) (no need to re-encode?), but would have been thrown out with re-encoding in strategy (2). Or does PowerDirector have to re-encode to go to HQ in either case?

First of all, the DVD-Video specification only allows specific bitrates which is why your authoring programs will only output HQ, SP, LP etc. These correspond to the actual basic modes:

HQ - High Quality (1 hour per side)
720 x 576 pixels - 9.72 Mbit/sec VBR

SP - Standard Play (2 hours per side)
720 x 576 pixels - 5.07 Mbit/sec VBR

LP - Long Play (3 hours per side)
360 x 576 pixels - 3.38 Mbit/sec VBR

EP - Extended Play (4 hours per side)
360 x 576 pixels - 2.54 Mbit/sec VBR

It doesn’t matter what your initial capture bitrate is; if it doesn’t match the above then a further re-encoding will take place during the authoring to DVD.

Since you are much better off just doing one encoding step, and it’s VHS quality to start with, why not capture at SP and then author without further loss (it depends on your perception whether you can further drop the capture one more notch down to LP).

Your capture bitrate is way over the top considering that you are coming from a vhs source.

imkidd is right—drop it down to 3.5 to 5mb and see if your results match the vhs input well enough. You don’t want to rework the material to fit your dvds if at all possible.

Thanks for the suggestions. Believe it or not, I can already see artifacts at SP, maybe because the source is grainy and has a constant VHS “warbling” which doesn’t go away even when using the smoothing and noise removal features of PowerDirector, and this must eat up bitrate.

My question also has to do with the quality of different encoders, whether anyone has an opinion about whether PowerDirector or DVD Shrink would be better, and whether a program like PowerDirector re-encodes the MPEG during final authoring if it doesn’t really need to (like when no further compression is needed) or whether it just passes the encoded MPEG “as is”.

I’m not an expert but there are claims scattered all over forums and documents that encoders can be of different quality even if they are the same bitrate and VBR.

1-capture with virtualdub or virtualvcr to mjpeg.avi
2-encode video with “hc”(freeware) or “cce basic”(50$ i believe…quality rather simillar to hc, but faster) and audio with (for example) tmpgenc if you need mp2, or ac3(if you’re making ntsc dvds) with aften(gui)
3-author with freeware too; DVDAuthorgui or dvdstyler
4-burn what authoring tools output with nero…

links for those progs that perhaps can’t be found by google alone;
http://kurtnoise.free.fr/index.php?dir=Aften/ (get last gui and last .exe file…currently it’s "aften-0.05_rev214.zip " )
http://www.bitburners.com/HC_Encoder/
if you wanna use hc, then avisynth too
http://www.avisynth.org/

i suggest doing it this way(this is the right way to do it) or doing it simple way(just buy a standalone dvd-recorder); capping to mpeg2 and then re-encoding later is probably worst way; it is not simple enough(dvd-recorder beats it) and quality is bad(mjpeg.avi process beats it).
so pick simple , or pick hard, but don’t pick mpeg2 cap which you must process again.

It’s pretty much accepted that different encoders can give widely variable results at the same bitrate. That’s really why Canopus and Mainconcept cost a lot.

That you are getting audio ‘warbling’ at SP capture is bad news indeed, but it depends on whether it is present at source (ie on the original VHS recording) or not. Video artefacts can be compensated for but if the encoder can’t faithfully record audio then it’s hopeless, and since you imply it’s present on the original material then that’s not good. Recording VHS at SP should be problem-free so the simpler DVD recorder route as outlined by [I][B]i4004[/B][/I] may be cheaper for you.

Personally I have had absolutely no problem with VHS -> DVD recorder via SCART (EU) or S-Video (US).

That’s really why Canopus and Mainconcept cost a lot.

hc encoder is free, yet beats both of these when it comes to quality.
(probably beats them on speed too.)

actually i could challenge anybody to find a better mpeg2 encoder than hc(when it comes to quality), and they would all fail.
as said, cce is a bit faster…but also a bit worse when ti comes to quality.
days when prices were a measure of things evaporated in the pc age.
freeware can get you rather far.
in fact, one might argue, even much further than payware.

Sorry I have been absent from my own thread for so long…

Thanks for your all of your comments. I played around with both ways I mentioned (as I probably should have done before I posted). The results were remarkable. Using PowerDirector to author files for a 9 GB DVD and then DVD Shrink to reduce them to DVD-5 gave a MUCH better result than authoring files for a DVD-5 directly. Which convinces me that PowerDirector’s MPEG2 codec is much inferior to DVD Shrink’s.

I do appreciate the suggestions for using other tools, but I have (foolishly?) invested a lot of time in hacking PowerDirectors menu creation to do exactly what I want it to do. If I can get PowerDirector to use a better (free!) codec, that would be ideal. Of course the PowerDir people don’t tell you how to do this. I know VirtualDub for example has a simple GUI menu page which allows you to choose different encoder, but PowerDirector does not have any such interface. Does anyone have experience doing this kind of hacking in general, even if not specifically for PowerDirector, and can you give me any pointers?

Sorry, not a programmer, so I can’t tell you how to hack PowerDirector. It’s mpeg encoding engine must be very poor if DVDShrink can outclass it so easily. DVDShrink doesn’t reencode, it is a vbr transcoder. I hope you are doing the Deep Analysis with Shrink before you reduce the size of your videos.

Personally, I’d look through a few other tools, including some of those free ones that I4004 mentioned a few weeks back. There is a free version of DVDRebuilder that can use several different encoders, including HC, Procoder and CCE. The pro version of Rebuilder is my favorite tool for reducing the size of dvd video.

Thanks for the suggestion, I have started to look at DVD Rebuilder. Meanwhile I have another question, which is more about the structure of an authored DVD with menus. Could I author with PowerDirector, and then delete just the VOB that it creates, and swap in one that I create with VirtualDub and the codec of my choice? Or would this hopelessly screw up the navigation of the DVD?

I don’t see any way that could work.

You probably need to ditch PowerDirector and find better authoring tools. There were a couple mentioned earlier that are free, dvdstyler and DVDAuthorgui. And there are some reasonably good commercial programs, like the ones from Pegasys, Ulead and the one I use, called DVDLab Pro (not cheap). Not all of these have mpeg encoding capability built in though.

If you want to learn the nuts and bolts of capturing, encoding and authoring look through I4004’s post for some of the tools you can get for free. Go to doom9.org and videohelp.com and read through their guides.

If you’d rather have a nice gui and some guidance within the program, some of the commercial packages might fit you better. Ulead Movie Factory 5 for example.
I just think you’re on the wrong track with this Cyberlink program.