More bitrate matrix selections?

[blutach]

We already have the ability to set a favourite matrix for low (<3000) and very low (<2000) bitrates (as well as extras). But many segments fall into "Medium" (say 3000-3500) and "Medim-High" (3500-4000), as well as High (4000+).

I wonder if it's possible (or worthwhile) to add 3 more choices to the options to cater for these segments?

Regards

[Wombler]

We already have the ability to set a favourite matrix for low (<3000) and very low (<2000) bitrates (as well as extras). But many segments fall into "Medium" (say 3000-3500) and "Medim-High" (3500-4000), as well as High (4000+).

I wonder if it's possible (or worthwhile) to add 3 more choices to the options to cater for these segments?

Regards

Sounds good to me.

There's probably a decreasing need to optimise the higher the bitrate but it wouldn't do any harm to have the choice.


Wombler

[tom942]

It's a good idea, yes :).

However will you use it just in the classical method or even with the bitrate distribution method?.

I don´t have clear if with the bitrate distribution method, we have to tick different matrices or just use one (the ticked in Main feature) for the whole process.

Regards.

[auldyin]

Bloody Hell chaps!!!....I don't even know what you are talking about!

Can someone point me in the direction of "hidden options", matrices and such like cos' I don't have a clue.

Cheers

[Sharc]

However will you use it just in the classical method or even with the bitrate distribution method?.

I don´t have clear if with the bitrate distribution method, we have to tick different matrices or just use one (the ticked in Main feature) for the whole process.


It's up to you.
You can tick different matrices with redistribution.
DVD-RB will use the "main feature matrix" for the Q-analysis and the redistribution profile, and then use the actual matrix according to the redistributed bitrate for the encode.

[tom942]

@Sharc

I thought like we are doing a CQ or OPV pass in BD, it was better to use only one matrix in the encoding phase too, because the new bitrates was somehow related or tied to the matrix used, like when we are doing "One Pass VBR w/analysis".

I thought that although the original bitrate was, for instance, 3200 and goes down to 1700 kbps, I can still use the medium-high bitrate matrix that was used for redistribution in the encoding process, because that matrix gave me that 1700 kbps., it will not introduce artifacts.

Also, I saw your workflow for "Saving Private Ryan" and you wrote that you used the same matrix for the whole process (redistribution and encodig) and that your habitual workflow is to use one matrix only.
Something similar Robot1 and you said in other thread, that it could be safer to use only one matrix.

BTW, I ask this in this in this thread:

http://forum.doom9.org/showthread.php?t=126796&page=15

Regards :).

[Sharc]

If you tick the same matrix for "Main Feature Matrix", "Low bitrate Matrix", and "Very Low Bitrate Matrix", then the same matrix will be used for redistribution and the encode - independent on the actual redistributed bitrate. I think using one single matrix throughout the process preserves the original idea of redistribution based on a fixed Q or CQ best.

I really can't tell if selecting different matrices for the redistribution and the encode - depending on the actual redistributed bitrate - would improve the result further. DVD-RB leaves the choice to you and to your eyes.

[blutach]

@auldyin

These are not hidden options, but on the main RB GUI.

Options tab - Advanced (Expert) Options - Quant matrices



@Sharc - normally, I'd agree that a single matrix preserves the idea of CQ, but segment bitrates can vary wildly, especially after redist, where some will be cranked up and others taken down accordingly.

Regards

[tom942]

So the redistribution just try to equal the quality between all the segments, independently the matrix used.

It is like the encoder says: "in this segment you don´t need more bitrate than this or on the contrary, in this segment you need more bitrate than this".

I mean that, for instance, if you use FHE, CCE's standard or AVAMAT6 for redistribution, with the calculated Final_Q (not the Base_Q) for each one, it will give you more or less the same redistributed bitrates, won´t it?.

I thought that if you use FHE, it will give you a very different redistribution than, for instance, AVAMAT6, so consequently I thought that it was a must to use FHE.

If yes, then I was mixing apples with pears :).

Regards.

[Sharc]

@auldyin

@Sharc - normally, I'd agree that a single matrix preserves the idea of CQ, but segment bitrates can vary wildly, especially after redist, where some will be cranked up and others taken down accordingly.


Yes, but the "wild" swing of the redistribution is actually produced by the given Q/matrix combo. If we change the matrix afterwards we are in a chicken and egg situation.
Anyway, I don't think that adapting the matrix afterwards does any harm to the quality, I just doubt if it has a real benefit.
Maybe there are different experiences around.

[Sharc]

I mean that, for instance, if you use FHE, CCE's standard or AVAMAT6 for redistribution, with the calculated Final_Q (not the Base_Q) for each one, it will give you more or less the same redistributed bitrates, won´t it?.


Exactly, my experience. That's why I suggested to go for the Final_Q rather than Base_Q.

[tom942]

Thank you Sharc for help me to clarify my doubts, and I also second blutach idea :).

Regards :)

[jdobbs]

We already have the ability to set a favourite matrix for low (<3000) and very low (<2000) bitrates (as well as extras). But many segments fall into "Medium" (say 3000-3500) and "Medim-High" (3500-4000), as well as High (4000+).

I wonder if it's possible (or worthwhile) to add 3 more choices to the options to cater for these segments?

RegardsShouldn't be too hard. I'll look into it.

[blutach]

Many thanks!

Regards