How does bitrate affect Q. Factor and D2S options?

[joesloppy]

Thanks for all your help.

I have been searching around this forum for several weeks now and still and puzzled by something. I have learned about Q. Factor and how the D2SRoBa plug-in for D2S makes for the best quality and fastest encodes, even better than multi-pass VBR (so I read), using the Q. Factor. I also read that to have a Q of 30, 32, or 35 gives great results, the lower the better. So here are my questions,

One, is the D2SRoBa plug-in primarily for DVD 2 SVCD conversion and not DVD 2 DVD? With a SVCD conversion, you could say you don’t want a Q. Factor above 35. With a DVD conversion, it doesn’t seem that you can set a max Q. Factor, but rather the plug-in bases it on what will fit on the given DVD size. Then I just assume that a DVD is so big that you will always get a pretty good Q. Factor so who cares about max. Am I right?

This leads me to my second question, since the plug-in is just bases Q. Factor on size, can I just forget the plug-in and use the Q. Factor option in the Encoder tab of D2S to pick it and let it come out to be whatever size it wants based on the given Q. Factor? Say I choose DVD for output and then choose 1 pass VBR and then a Q. Factor of 35, is this the same thing as the plug-in but I will just have a file size that will be whatever it wants to be.

Oh, one last question, how does bitrate affect Q. Factor, I assumed at first that it didn’t matter what the bitrate was but I noticed file size differences in two samples I made with different bitrates, but the same Q. Factor. So will bitrate determine the Q. Factors quality? What should I set bitrate at for a Q. Factor of 35 on a DVD, what are some guidelines for all those bitrate options in D2S?

Thanks for your help again guys!

[Crackhead]

Originally posted by joesloppy
One, is the D2SRoBa plug-in primarily for DVD 2 SVCD conversion and not DVD 2 DVD? With a SVCD conversion, you could say you don’t want a Q. Factor above 35. With a DVD conversion, it doesn’t seem that you can set a max Q. Factor, but rather the plug-in bases it on what will fit on the given DVD size. Then I just assume that a DVD is so big that you will always get a pretty good Q. Factor so who cares about max. Am I right?I don't do any DVD2DVD conversions, so I'm just guessing, too, right now, but if the method of the encoding itself remains the same (i think it doesn't change), you are absolutely right!

Originally posted by joesloppy
This leads me to my second question, since the plug-in is just bases Q. Factor on size, can I just forget the plug-in and use the Q. Factor option in the Encoder tab of D2S to pick it and let it come out to be whatever size it wants based on the given Q. Factor? Say I choose DVD for output and then choose 1 pass VBR and then a Q. Factor of 35, is this the same thing as the plug-in but I will just have a file size that will be whatever it wants to be.Yes, but if you choose any static Q, you will get maybe 1 1/2 filled CDs or a 1/2 filled DVD, so if you don't mind this feel free to avoid D2SRoba!

Originally posted by joesloppy
Oh, one last question, how does bitrate affect Q. Factor, I assumed at first that it didn’t matter what the bitrate was but I noticed file size differences in two samples I made with different bitrates, but the same Q. Factor. So will bitrate determine the Q. Factors quality? What should I set bitrate at for a Q. Factor of 35 on a DVD, what are some guidelines for all those bitrate options in D2S?it's the other way round, the average bitrate depends on the q.factor, the lower the q.factor the higher the average bitrate. for example, with q=35 you get an average bitrate of let's say 1500kbps! with q=20, however, you get an avb of 2200, maybe! This values depend on the movie you encode, its compressibility, its length, the filters you apply, etc.!

If I'm saying some BS here, please correct me @the more advanced users!

if something's still unclear, just ask!

[DDogg]

One, is the D2SRoBa plug-in primarily for DVD 2 SVCD conversion and not DVD 2 DVD? It is also for DVD. You will notice that when D2SRoba detects that you have DVD 2 DVD selected it actually changes the interface. It shifts to a size based method. No auto or "?". Just 1 disk target.

With DVDs you may select multiple audio tracks which may be very large AC3 streams. This eats up a lot of space. Since the Resolution and the Max bitrate is so much higher, the video portion uses up much, much more room. So fitting the encode within the allocated space is just as important as with a SVCD.

True, normally even with the AC3 audio and subs you still have a bunch of room left and so the Q tends to be lower. This may not always be the case though. Especially with a long movie. In some ways I think D2SRoba could be called more important to DVD users than SVCD'ers. One aspect of this is that D2s and D2sRoBa lends itself to allowing multiple movies on one DVD because you can set the target space to, say 2200 kb instead of the full DVD.

Oh, one last question, how does bitrate affect Q Forget bitrate! :) Well that is not exactly true as you do want to pick your min and max to remain with the particular standard you are using. Other than that, just put it out of your head. CCE OPV looks at the source and constantly changes bitrate, within the constants of MIN and MAX, so that the Q is what you specified when you chose the Q value you desired. The allocated bitrate CCE chooses is just a byproduct of this process and should not concern you. It constantly adapts itself to the complexity of the source and allocates X amount of the bitrate it "has in the bank" to achieve the goal given it. This is an over simplification, but it should do to get a general understanding.

That all sounds good, but if we did not have a sampling process to project the final size of the encode we would then be in big trouble as a complex source of 2 hours may take twice the amount of bitrate as a simple source of the same length (exaggeration for making the point)to achieve the same Q level. There is just no way to know where we will end up size wise without a pre-sampling test before the encode is done. With the test (like that done with D2SRoBa), we then know if a certain specified Q can be achieved within a certain specified size. That is one of the main beauties of 1pass Bach. We know in advance of the actual encode what quality we will end up with. Kinda magic if you think about it.

So, the only slight problem with all of this was that the sampling process could be off just a tad...just enough to make you a little crazy if you were a perfectionist (trainwreck if it was too big). tylo then added a conditional second pass that is only activated if the final size of the encode is not within the size tolerance the user specified. That's why I have always called it a "hybrid 1pass Bach process".

With a DVD as the target these small errors are not nearly as big a deal as with SVCD so the tolerance can be set much broader, thus I suspect the second sizing pass will be called only infrequently for DVD'ers making D2SRoBa even better for them.

btw, in the encoder tab when CCE has been chosen, always make sure "calc bitrate as max" is unchecked.

[ArchStanton]

Originally posted by DDogg

Oh, before I forget it, note that B3 has a very irritating bug that can bite you when you are not using D2sRoBa. The #$%# "calc bitrate as max" check-box stays checked unless you manually uncheck it. With it checked *everything* to do with OPV is screwed up so make sure to uncheck it. It was put in for a legacy issue (that I caused I am told) that was later proved false. *Always* have "calc bitrate as max" unchecked.

Sorry, I don´t get that. Do you refer to the Big 3, the bitrate options in D2S, or an option within CCE?

But I also have another question. First let me state, that I use 3 pass CCE for years now, it never bothered me, but when I read that one could have equal or better quality with 1 pass, of course I was curious. However at all my tests with D2sRoBa there always was a 2nd pass necessary, either undersized or oversized. Well, I´m not bitching around, since 2 passes are better than 4, but I wonder if the *.vaf files differ? IF they are the same, the sizing pass just would be "ordinary" VBR-encoding?

Arch Stanton

P.S. Btw, yes i do unsterstand the difference between VBR ans OPV, at least basically ;)

[r6d2]

Originally posted by ArchStanton
Well, I´m not bitching around, since 2 passes are better than 4, but I wonder if the *.vaf files differ? IF they are the same, the sizing pass just would be "ordinary" VBR-encoding?
I think that this (http://forum.doom9.org/showthread.php?s=&threadid=64886) thread may give you an answer, particularly Part I, Q4 (and the guide pointed there).

[DDogg]

Arch, thanks for reminding me of that. Disregard it. I have edited it out of my post so as not to cause further confusion. It was a bug in a particular test build I had been using for a long time. Long enough that I forgot it was a non released test build. Since it never made it to the public, what I posted was incorrect information. Sorry for that.