HC 0.18 vs HC 0.17

[Kayaker]

I've seen a rahzel post similar to this in other forums.
So here are my 2 cents.

I also noticed a lot of macroblocks with HC 0.18, specially in the fade from black initial sequences.

I've tried with BIAS 20 ( the 0.17 default ) and BIAS = 0 ( the .18 default ) and BIAS=50 and whatever you like and it was the same.
So no matter what BIAS I use with 0.18 there is a lot of macroblocks in the beggining scenes.

Maybe Hank could explain a little bit this behaviour ?

The overall quality of 0.18 seems better, but those starting macroblocks are really annoying.

[writersblock29]

At any particular bitrate, though? If spread thin enough, pretty much any encoder's going to pixelate the image, regardless of the Bias used. Not saying you're wrong -- we all notice what we notice, and quality's one heck of a subjective beast! But most of my experiments with HC18 yeild very similiar results to CCE (except for the time involved encoding). Of course, that's only offering proof about the subjective thing...

[Kayaker]

Sorry I forgot to mention that was at very low bitrates.
I forgot sometimes that the rest of the world can't read my mind ;)

Below 2000 for sure ( I don't remember exactly. )
It was in an extra I steal bitrate from.

The macroblocks were big. Believe me.
But the point is, HC 0.17 didn't do that huge macroblock pixelation in the beginning.

And I've tried bias 0 and bias 20 with BOTH versions.
As DVD-RB isn't passing bias in the command line (I've cheched with Process Explorer ) HC take it from the ini file.

I've checked compression all combinations of 0.18 and 0.17 with various BIAS settings from a bat file repeating DVD-RD command line.

Don't get me wrong, overall quality seems better with 0.18.
But it has a huge macroblocks in the fade from black beggining scenes that HC 0.17 didn't have.
At least for low bitrates.

[jdobbs]

DVD-RB Pro passes parameters to HC in the HC.INI file.

[bennynihon]

do you use Best quality, or do you use a Faster encode that compromises quality?

[linx05]

Best quality always!

[jdobbs]

Rebuilder uses the quality setting that you've selected under the "HC Settings" menu. You can choose "Fast, Normal, or Best".

[Kayaker]

Yep.

Always best quality.
I'm about to test now Autoqmatenc now.

The theory that it's good for low bitrates (below 4000 o 3500 for standard dvd resolution) cause it automatically tunes the mpeg matrix seems interesting.

The encoding code its the same of QueEnc ( libavcodec ) but this automatic matrix thing seems interesting.

Gotta research about it... but the idea is promising, isn't it sapstar ? :)

[DocDragon]

Yep.

Always best quality.
I'm about to test now Autoqmatenc now.

The theory that it's good for low bitrates (below 4000 o 3500 for standard dvd resolution) cause it automatically tunes the mpeg matrix seems interesting.

The encoding code its the same of QueEnc ( libavcodec ) but this automatic matrix thing seems interesting.

Gotta research about it... but the idea is promising, isn't it sapstar ? :)
it has been already mentioned in different threads that using a customized matrix could improve the encoding result quite a bit. i did try out the AVAMAT7 matrix in combo with HC018 and was *extremely* surprised how good the result was.

i used "farscape peacekeeper wars," different "star trek voyager," and "stargate SG-1" discs, and HC018/AVAMAT7 *visibly* beat CCE basic and AQM! i even fed AQM with the AVAMAT7 matrix (don't forget to turn OFF AQMopt for that), and the result was still worse than HC, which was kind of surprising for me b/c i expected to have the same picture quality when using the same matrix :confused: .

from now on, HC018/AVAMAT7 is on the top of my list. i only wish HC was multithreaded and could make use of multiple CPU cores...

DD

[^JazzMan^]

i used "farscape peacekeeper wars," different "star trek voyager," and "stargate SG-1" discs, and HC018/AVAMAT7 *visibly* beat CCE basic and AQM! i even fed AQM with the AVAMAT7 matrix (don't forget to turn OFF AQMopt for that), and the result was still worse than HC, which was kind of surprising for me b/c i expected to have the same picture quality when using the same matrix :confused: .

from now on, HC018/AVAMAT7 is on the top of my list. i only wish HC was multithreaded and could make use of multiple CPU cores...

DD

On what bitrate limits you used AVAMAT7 and what kind of sources(high/low motion, rate of scene changes, amount of details)

If you enable Multiple encoder processes you'll use both cores since DVD-RB would start 2 threads of HC, but I think you already know that :) since you don't encode a single video file(at least not with DVD-RB) you won't be needing HC working on 2 cores when you run it as a single thread

cheers

[DocDragon]

On what bitrate limits you used AVAMAT7 and what kind of sources(high/low motion, rate of scene changes, amount of details)

If you enable Multiple encoder processes you'll use both cores since DVD-RB would start 2 threads of HC, but I think you already know that :) since you don't encode a single video file(at least not with DVD-RB) you won't be needing HC working on 2 cores when you run it as a single thread

cheers
bitrate was around 2600-3000. on what kind of sources? well, that's why i mentioned those movies for comparison. AVAMAT7 was obviously used throughout the entire film -- if that's what you mean. i'm not sure if changing matrices during a movie is DVD compliant at all, and i wouldn't know how to do it anyway :o :

yes, i'm aware of using HC with 2 instances of DVD-RB to fully make use of my dual core, but i think it would be advisable to have 2 copies of DVD-RB in *different* directories b/c REBUILDER.INI gets modified during the encoding process, and one instance of DVD-RB might confuse the other one. that's just too much organizing for me... ;).

CKA

[dragongodz]

The encoding code its the same of QueEnc ( libavcodec )
just a small correction here. while AQE and QuEnc both use libavcodec they are not the same thing. that is as far as i know we dont use all of the same settings and QuEnc in 2 pass encoding uses Xvid ratecontrol. i do not know exactly what AQE is set for its ratecontrol, being closed source, but i have to assume its not Xvid since AQE can do more passes.
since rate control is actually a rather important thing that makes a rather large impact on the output you should not consider the results from QuEnc will be the same as from AQE.

[Kayaker]

@dragongodz

I know they are not the same ( as a final product ).
Changing the plethora of settigns for libavcodec can make a great difference, as you already pointed out.
So, the encoding library/code IS the same, what is not the same are the settings. And of course that can make a great difference.

@DocDragon

Very interesting you info about custom matrix AVAMAT7.
I've never tried customed matrices because never dared to it.
And I thought I was getting on risky lands.

But I'll try that.

And of course with DVDRB and doing whole DVD without stripping extras and stuff you are almost always working on low bitrates.

And if AVAMAT7 is the matrix to go for such low bitrates, let's try it.

Also it's very interesting that using a fixed matrix like AVAMAT7 it's better than using AQMEnc with its automatic and tuned matrix generation (at lest for low bitrates).

I suppose Amnon82 did really know what he was doing.
Gotta read about matrices.

[dragongodz]

I know they are not the same ( as a final product ).
yes and i thought it should be made clear to anyone else reading this thread that testing with 1 is not the same as testing with both because of their differences.