Questions about AGK behaviour (dealing with codec settings and AC3 decoding)

[numaios]

Hello!

Since Auto GK is an automated tool I was wondering where I can find some guide that describes its modus operandi (if it does exist, of course). If it doesn't, that's what I'd like to know if possible (I have already read the official AGK FAQ):

-Which command line is set in AZID when encoding from AC3 to MP3? Which output mode? Is it always the same or it changes depending on the number of channels the source has?

-In XviD encoding. Which level of motion search precision is used? Is chroma motion enabled? Which level of VHQ mode? How many B-frames? What B-frame configuration? VHQ also for B-frames? Adaptive quantization? Which custom matrices are included? Chroma optimizer? Trellis quantization? Any quantizer restriction?
Are all this settings always constant?

Thank you very much.

[BigDid]

Hello!

Since Auto GK is an automated tool I was wondering where I can find some guide that describes its modus operandi...
Hi,

Try the guides or tutorial from manono: http://forum.doom9.org/showthread.php?t=92089
or dig in the main AGK thread, tons of infos.

-Which command line is set in AZID when encoding from AC3 to MP3?
Start an encode and open the dos window when the audio part is running, you will find the command lines info
Which output mode?
If input is multichannel, I believe output is DplII

Is it always the same or it changes depending on the number of channels the source has?
For stereo input stereo or dual channels as output

-In XviD encoding. Which level of motion search precision is used? Is chroma motion enabled? Which level of VHQ mode? How many B-frames? What B-frame configuration? VHQ also for B-frames? Adaptive quantization? Which custom matrices are included? Chroma optimizer? Trellis quantization? Any quantizer restriction?...

Sorry no more time ATM, any takers?

Did

Edit: for Xvid, just launch the Xvid encoder Gui once in the second pass or after, it will retain the informations you are looking for.

[numaios]

Thank you for your answers :) Some people must think that my questions are too obvious or even "inappropriate". But I think someone who is used to AutoGK (I'm not) can perfectly answer them. And you have answered, so thanks again.
for Xvid, just launch the Xvid encoder Gui once in the second pass or after, it will retain the informations you are looking for.
Do you mean Start-->Programs-->XviD-->Configure encoder?

I did take a look on the guides you said and I didn't find that information, they just say how to use AGK, not how it works.

[BigDid]

...
Do you mean Start-->Programs-->XviD-->Configure encoder?...
Yes.

Thank you very much
You are welcome. Bye.

Did

[numaios]

OK, I guess I will do some tests in different conditions and try to understand its algorithms, opening the XviD encoder GUI.

Bye!

[BigDid]

OK, I guess I will do some tests in different conditions and try to understand its algorithms, opening the XviD encoder GUI.

Bye!
Hi again,

Maybe if you can be more specific in your questions? ... Just a guess.

Did

[unskinnyboy]

Which level of motion search precision is used?6.Is chroma motion enabled?Yes.Which level of VHQ mode?1.How many B-frames? When used, 1.What B-frame configuration? If you mean ratio and offset, 1.50 & 1.00VHQ also for B-frames? Yes.Adaptive quantization? No.Which custom matrices are included?EQM V2 (specially designed for AutoGK) and Jawor's 1CD, IIRC. There used to be HVS matrices too, but I don't think they are being used anymore.Chroma optimizer? Yes.Trellis quantization? Yes.Any quantizer restriction?Yes, but it depends on the compressibility. Lower the compressibility, higher the max. quantizer for all the frame-types. Min. quantizer always remains 2.Are all this settings always constant?No, because of the reasons mentioned above.

[numaios]

Wooooooooow!!!!!! :D :D :D Thank you very much, unskinnyboy, that was exactly what I wanted to know (and yes, I meant ratio and offset). Some options look like if they were always constant (like MSP: 6 or Chroma ME: on), and some others look variable depending on the compressibility (B-frames or quantizer restrictions). But I'm not sure with one of them: is VHQ always "1" or it changes depending on the compressibility?


Maybe if you can be more specific in your questions? ... Just a guess.What can be more specific than questions like "Which level of motion search precision is used?" or "What B-frame configuration?" or "Is chroma motion enabled?". And they are easy to answer too: just "yes", "no" or a number (or "it is not constant") ;)

Thank you guys for your answers :)

[unskinnyboy]

is VHQ always "1" or it changes depending on the compressibility?Always 1, else I would have mentioned otherwise. VHQ has got nothing to do with compressibility per se.

[BigDid]

...What can be more specific than questions like ... ;)

Ok, it was a shortcut after my first answer. The long version would have been: Once you will have looked in the encoder gui, if you cannot find somes answers settings meanings ... than maybe you can come back, give some feedback and ask for specific questions.

General thoughts: if you surf the forum quite often you will find, from time to time, some posts from veterans/experts ... complaining about what is called "mouth feeding" which can translate by if the poster make some efforts, do some searches, tries by himself and asks specifics questions ( I have done that I have this problem or interrogation...) then it is likely he will get answers; on the contrary somebody asking: hey! how do I do this or that? show me all from point A to point Z, it is likely that poster will get no answers or acid remarks. ;)

The time I give here cannot be given for others and I don't have enough ATM, so I will stop and not answer back. Good luck in your encoding projects, glad you try AutoGK and Xvid for them. Bye.

Did

[numaios]

Always 1, else I would have mentioned otherwise. VHQ has got nothing to do with compressibility per se.Please tell me if I'm wrong, but if some source has a bad compressibility, a higher VHQ would help to achieve a better quality, wouldn't it?

Ok, it was a shortcut after my first answer. The long version would have been: Once you will have looked in the encoder gui, if you cannot find somes answers settings meanings ... than maybe you can come back, give some feedback and ask for specific questions.OK, I definitely missunderstood you :). And thank you for encouraging me to come back later with new questions.

if the poster make some efforts, do some searches, tries by himself and asks specifics questions [...] then it is likely he will get answersYou are totally right. But try to understand that if someone is thinking about starting to use a new encoding tool, and he wants to know some small details before deciding, there are 3 options:

1) He can do hundreds of tests just to know if he needs that tool.
2) He can do a good search. I did, and it was pointless, since I didn't get the answers.
3) He can ask to a veteran that knows perfectly how it works, and already knows the answers.

I asked, and you and unskinnyboy answered, and I'm very grateful for that. Both of you have wasted a couple of minutes, but I would've wasted several days.

Thank you.

[unskinnyboy]

Please tell me if I'm wrong, but if some source has a bad compressibility, a higher VHQ would help to achieve a better quality, wouldn't it?A higher VHQ would give better quality, but it doesn't have to be tied to compressibility. For every source, that is true (at least theoretically) - better ME, higher PSNR and all that. In practice though, a lower VHQ sometimes might help with a bad source from a quality viewpoint, not from a compressibility viewpoint. Used to be a time when a higher VHQ would *lower* compressibility, but the VHQ code has been rewritten since then and thus not tied to compressibility anymore. As a general rule of thumb, VHQ=4 would do just fine.

[numaios]

Thanks again for your explanatory answer. So:
-For a good source: VHQ 4?
-For a bad source: VHQ 1?

As a general rule of thumb, VHQ=4 would do just fine.I see... But VHQ=4 isn't worth the time it wastes and that's why it's not used in AutoGK, right?

[unskinnyboy]

I see... But VHQ=4 isn't worth the time it wastes and that's why it's not used in AutoGK, right?It's worth the time for many, but the audience for AutoGK are generally novices who want to do encodes without much hassle and the 4x encoding time increase with VHQ=4 is a hassle for them. Hence VHQ=1 which isn't all that bad anyway.

[numaios]

It's worth the time for manyEven with bad sources? (I ask this because of your previous post, where you said that a lower VHQ mode could increase quality in some bad sources).

By the way, can I ask if "many" includes you :)? I mean, do you use VHQ=4 in general?

[unskinnyboy]

In general, yes. But depending on the situation (some bad sources, some noisy sources, just for fun, I am drunk etc), I lower it too. There is only one way to know for sure whether it helps or not -> test.

[numaios]

There is only one way to know for sure whether it helps or not -> test.You mean doing two compressibility tests with GK, one with VHQ=1 and the other with VHQ=4 and see if given values are different?

Or encoding the movie twice with different values and deciding subjectively which looks better?

[unskinnyboy]

You mean doing two compressibility tests with GK, one with VHQ=1 and the other with VHQ=4 and see if given values are different?If you want to, but I'd be surprised if it shows a lot of difference or even any. Like I said before, VHQ ain't got a lot to do with compressibility. But sure wouldn't hurt to do some tests. Last tests on this front are more than a couple of years old.Or encoding the movie twice with different values and deciding subjectively which looks better?Yes.

[numaios]

OK! I'll give it a try.

Thank you, unskinnyboy.