digitalFAQ.com Forums [Archives]

digitalFAQ.com Forums [Archives] (http://www.digitalfaq.com/archives/)
-   Video Encoding and Conversion (http://www.digitalfaq.com/archives/encode/)
-   -   DVD-RB and audio stuttering? (http://www.digitalfaq.com/archives/encode/12572-dvd-rb-audio.html)

Dialhot 10-14-2004 11:13 AM

Quote:

Originally Posted by J-Wo
Right now I'm doing min=300, max=9000?

Two things :
1/ DVD-RB reports the min/avg/max bitrate of the original source. There is no real need to put your own max at 9000 if the original was 5000 (that means that the movie doesn't need a lot of bitrate peaks).

2/ with KVCD-Notch matrix, you have really excellent result with MAX=5000. I did all my KDVD with max=6000. This way the Q values is a couple of points better than with a max=9000.

J-Wo 10-14-2004 12:18 PM

Quote:

Originally Posted by Dialhot
1/ DVD-RB reports the min/avg/max bitrate of the original source. There is no real need to put your own max at 9000 if the original was 5000 (that means that the movie doesn't need a lot of bitrate peaks).

Rb by default sets a min of 0 and a max of 9000 for its encodes, I only set it manually down to 5000 per Boulder's request.

Quote:

Originally Posted by Dialhot
2/ with KVCD-Notch matrix, you have really excellent result with MAX=5000. I did all my KDVD with max=6000. This way the Q values is a couple of points better than with a max=9000.

I just did an OPV encode with min=300, max=5000 in Rb 0.59, and there was no change in audio dropouts compared to min=0, max=9000 (default Rb settings). This is using standard CCE matrix. When I change to notch matrix, I generally set min=64, max=5500, and still there are dropouts. So right now it really doesn't look like min/max bitrates has an effect (at least not for me).

I'm running a one click mode with v0.59 right now since I've always used three click in conjunction with Rb-Opt. So I'll see how this one turns out...

Dialhot 10-14-2004 12:22 PM

Quote:

Originally Posted by J-Wo
Rb by default sets a min of 0 and a max of 9000 for its encodes, I only set it manually down to 5000 per Boulder's request.

#1 I know what Rb does and #2 I use 5000 as an instance ;-)
But the day where all that RB does will be the best to do is not for tomorrow. So trust me, even if the default is 9000, do not use so much with a source that is 5328 (for instanbce ;-))

Quote:

So right now it really doesn't look like min/max bitrates has an effect (at least not for me).
Let say that this is the most common source for that kind of problem but in your case it seems to not be related.

Quote:

I'm running a one click mode with v0.59 right now since I've always used three click in conjunction with Rb-Opt. So I'll see how this one turns out...
You should try first the multi pass, just to see...

J-Wo 10-14-2004 12:36 PM

Yes I'm running a one-click encode so EVERYTHING is at DVD-Rb default settings (i.e. 2-pass VBR). Rb reports High/Low/Avg bitrates at 3758/1852/2759. Looking at my CCE window, it says "VBR: 2690 (0-9000) kbps". So I guess I'll just leave it at default settings, even though max=9000 may not be needed.

J-Wo 10-15-2004 07:50 PM

well hallelujah!! OPV seems to have been the culprit all along!!! Which might be a reason why I've been having so many problems because practically EVERYONE in this forum does one-pass encoding. The default 2 Pass VBR worked like a charm -- video even looked much better than OPV, but of course took twice as long to encode.

So if I'm going to stick with VBR, should I change the min/max bitrate? Again, Rb defaults to 0-9000 but for most KDVDs I do 64-5500. Any suggestions?

Dialhot 10-15-2004 09:47 PM

Quote:

Originally Posted by J-Wo
video even looked much better than OPV

We definitely not speaking the same language. Sorry... :-)

Quote:

So if I'm going to stick with VBR, should I change the min/max bitrate? Again, Rb defaults to 0-9000 but for most KDVDs I do 64-5500. Any suggestions?
I think I gave you mine already :-)

J-Wo 10-15-2004 10:45 PM

Quote:

Originally Posted by Dialhot
Quote:

Originally Posted by J-Wo
video even looked much better than OPV

We definitely not speaking the same language. Sorry... :-)

I dunno, but maybe that was true back in the days of kvcd and tmpgenc. But now with cce and kdvds perhaps multi-pass might be better! But I don't much care, whichever will solve my audio problems is good enough for me.

Quote:

Quote:

So if I'm going to stick with VBR, should I change the min/max bitrate? Again, Rb defaults to 0-9000 but for most KDVDs I do 64-5500. Any suggestions?
I think I gave you mine already :-)
I just wasn't certain if min/max bitrates should be set different for OPV vs. multi-pass

Boulder 10-16-2004 01:26 AM

Some movies may look better with multipass VBR, Saving Private Ryan is generally considered one of them. Maybe it's the ultra-shaky camera movements which actually benefit from using VBR..who knows.

In general, OPV will give you better results.

J-Wo 10-16-2004 01:30 AM

that's interesting... it just seems like OPV is a kvcd "thing" and everyone else using CCE does multipass vbr... or is it just me? :)

Boulder 10-16-2004 02:10 AM

Quote:

Originally Posted by J-Wo
that's interesting... it just seems like OPV is a kvcd "thing" and everyone else using CCE does multipass vbr... or is it just me? :)

The RoBa method was developed for getting the best quality out of CCE, and this is basically OPV encoding. I guess most of the DVD2SVCD users use Tylo's D2SRoBa plugin to do OPV encodes.

Quality-based encoding will result in a better quality in most cases simply because it doesn't have to consider any bitrate reservoirs etc. It just uses as many bits as are needed in a frame. With VBR the encoder must consider whether it can use more bits or does it need them elsewhere.

jorel 10-16-2004 03:17 AM

@ Boulder
as i tryed to explain long time ago, D2S-RoBa can do VBR multipass mode.
(and sorry my poor english now and before a long time in old thread) :oops:

tylo did a version with VBR multipass for me after i ask for it: http://forum.doom9.org/showthread.ph...0&pagenumber=9

and now is a"public version":
v3.75 (5th October 2004)
fixed: NB! Bug in AutoIt3 caused wrong calculations in v3.61 - v3.74. Please update.
improved: Computes 'Adjust %' value into the sample estimation directly.
:arrow: changed: Starts DVD2SVCD in VBR multipass mode - forces D2S to calculate accurate BR.
added: Keeps old D2SRoBa_log files.

http://home.tiscali.no/tylohome/history.php

Dialhot 10-18-2004 03:52 AM

Quote:

Originally Posted by J-Wo
that's interesting... it just seems like OPV is a kvcd "thing" and everyone else using CCE does multipass vbr... or is it just me? :)

It's not just you (see message of Jorel telling he asked to Tylo to do a vbr version of D2SRoba - that means that Jorel find multi pass better too) but as boulder says : this depend of the source and :
Quote:

Originally Posted by Boulder
In general, OPV will give you better results.

And I underline that. I do anything in quality based mode, whatever the encoder used (OPV for CCE, CQ for Tmpgenc).

Boulder 10-18-2004 06:54 AM

jorel,

I'm not going to argue about that with you again. If you read the threads at D9 carefully, you'll see that D2SRoBa is originally meant for OPV encodings. Then Tylo added the possibility to do the resizing pass if the OPV encode is a certain percentage off the target. Then he added the ability to do a multi-pass VBR.

If you don't believe my words, this is what Tylo answered to your request:

Quote:

Ok, so I could have made an option that D2SRoBa does not set OPV mode, but you would probably be the only one to use that , or?
D2SRoBa is meant for OPV, all the rest is just added from user requests. In fact, why do you even use D2SRoBa if you do a multipass? You can do exactly the same in DVD2SVCD anyway.

Dialhot 10-18-2004 07:04 AM

Quote:

Originally Posted by Boulder
D2SRoBa is meant for OPV, all the rest is just added from user requests. In fact, why do you even use D2SRoBa if you do a multipass? You can do exactly the same in DVD2SVCD anyway.

I also wonder what is the added value of D2SRoba if you use vbr (as DVD2SVCD already does that very well). But perhaps Jorel found one and can tell us more.

Boulder 10-18-2004 07:49 AM

The only thing that comes to my mind is that the bitrate distribution graph (aka the vaf file) is better when you have a close-to-target OPV run first. But in this case it would be rather useless doing more than one 'sizing' pass since the bitrate distribution should already be optimal or at least very close to it.

Then again, I suppose that multipass-RoBa doesn't do any OPV first or does it?


All times are GMT -5. The time now is 02:56 AM  —  vBulletin © Jelsoft Enterprises Ltd

Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.