Different bitrate in splitted segments of same movie HELP ?

[nightrhyme]

Just encoded Godfather III. used Gknot 2-Pass. XviD: Quantizer MPEG,VHQ:4, Qpel and CM. Chose Size 3 CD's.

According to Gspot the Bitrate is as follows:

Joined Movie 2100MB: 1543 kb/
Segment 1 (CD1): 1540 kb/
Segment 2 (CD2): 1787 kb/
Segment 3 (CD3): 1478 kb/

Did use a XviD Zone Weight 0,20 for Credits. But that should only affect CD3 right ?

Is all as should be with the different bitrates ?

Thanx in advance

[gircobain]

What's there to be surprised about?

Remember, that's VBR - Variable Bitrate - the final average bitrate is just that - average - it doesn't mean it is kept thorough the encode

Think like this - if you drive 100km in 1 hour, your average speed is 100km/h - but that doesn't mean you kept driving at 100km/h through all the way

[nightrhyme]

Thanx so much for clarifying :-)

[manono]

I think you'll get a better idea of the relative quality of the different CDs by comparing the average quants with DRF Analyzer (http://www.geocities.com/analyzerdrf/index.html). They should be more similar than the bitrates, except, if using VBR MP3, then that may affect the outcome a little bit.

The fact that the 2nd CD has a higher bitrate doesn't really tell you very much. It probably has more complex scenes (fighting, crowd scenes, action, etc.), and just needed more bits to maintain the same quality. As gircobain said, it comes out with the average you set in the beginning, but can be all over the place during the different parts of the movie.

[nightrhyme]

Used DRF Analyzer

It says:
Average Frame quality is LOW [Average DRF/quantizer is 5.13]
Standard Deviation: Quality is LOW
How come. What could I do to better this ? And is this analysis to be trusted ? Does the Report tell you guys anything about what I could do better in the encoding process ?

Filesize: 2.203.594.752 bytes, Length: 02:43:10
DivX DRF Analyzer v0.9.5 Report!
File Name: D:\Godfather 3\godfatherIII.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 692 Height: 388 ]
Frame Rate: 25.000 frames per second
The Video has 244775 frames [ 02:43:10 ]

Average Frame quality is LOW [Average DRF/quantizer is 5.13]
Standard Deviation: Quality is LOW [Std. Deviation is 3.61]
Image Resolution is HIGH

The file has Packeted Frames!

There are NO frame drops ( NO drops is better )
This video may have some frames with VERY POOR Quality!

Recomended Resolution: [496x272] (Target DRF/quantizer=2.8)
The filesize should be larger!

Performance Caracteristics:
Macroblocks per frame: 1032 ( Poor Playback in Slow Computers, PIII450 or better required )
The Width is not multiple of 32. May degrade performance in some systems.

Kilobits per Second: 1552.84
Kilobits per Frame: 62.11
Kilobits per Macroblock: 0.060
Bits per Pixel: 0.24

Frame Type Statistics :
I Frames: 0.74%
P Frames: 50.21%
B Frames: 49.05%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 42493 17.5%
DRF=3: 40191 16.5%
DRF=4: 47677 19.6%
DRF=5: 31033 12.8%
DRF=6: 6959 2.9%
DRF=7: 33159 13.6%
DRF=8: 4288 1.8%
DRF=9: 4418 1.8%
DRF>9: 32751 13.5%
KeyF/DeltaF: 0.74%
KeyDRF<4: 1659
KeyDRF=4: 4
KeyDRF>4: 143

AverageKeyDRF: 2.01
MAXDRF: 31
AverageDRF: 5.13
Deviation: 3.61

[nightrhyme]

I just encoded a movie used Gknot 2-Pass. XviD: Quantizer MPEG,VHQ:4, Qpel and CM. Chose Size 3 CD's.

Used DRF Analyzer to analyze the result. But I having some trouble interpreting the output. Was hoping some of you guy's could help :-)

It says:
Average Frame quality is LOW [Average DRF/quantizer is 5.13]
Standard Deviation: Quality is LOW
How come ? What could I do to better this ? And is this analysis to be trusted ? Does the Report tell you guys anything about what I could do better in the encoding process ?

Thanx in advance

DRFAnalyzer Report:
Filesize: 2.203.594.752 bytes, Length: 02:43:10
DivX DRF Analyzer v0.9.5 Report!
File Name: D:\Godfather 3\godfatherIII.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 692 Height: 388 ]
Frame Rate: 25.000 frames per second
The Video has 244775 frames [ 02:43:10 ]

Average Frame quality is LOW [Average DRF/quantizer is 5.13]
Standard Deviation: Quality is LOW [Std. Deviation is 3.61]
Image Resolution is HIGH

The file has Packeted Frames!

There are NO frame drops ( NO drops is better )
This video may have some frames with VERY POOR Quality!

Recomended Resolution: [496x272] (Target DRF/quantizer=2.8)
The filesize should be larger!

Performance Caracteristics:
Macroblocks per frame: 1032 ( Poor Playback in Slow Computers, PIII450 or better required )
The Width is not multiple of 32. May degrade performance in some systems.

Kilobits per Second: 1552.84
Kilobits per Frame: 62.11
Kilobits per Macroblock: 0.060
Bits per Pixel: 0.24

Frame Type Statistics :
I Frames: 0.74%
P Frames: 50.21%
B Frames: 49.05%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 42493 17.5%
DRF=3: 40191 16.5%
DRF=4: 47677 19.6%
DRF=5: 31033 12.8%
DRF=6: 6959 2.9%
DRF=7: 33159 13.6%
DRF=8: 4288 1.8%
DRF=9: 4418 1.8%
DRF>9: 32751 13.5%
KeyF/DeltaF: 0.74%
KeyDRF<4: 1659
KeyDRF=4: 4
KeyDRF>4: 143

AverageKeyDRF: 2.01
MAXDRF: 31
AverageDRF: 5.13
Deviation: 3.61

[killingspree]

no need to start a new thread...

[niamh]

bitrate is high, bits/pix is high...........hhhmmm, Average quantizer shouldn't be that high(>5).... could be a screw up..
2 things :
-your width isn't multiple of 32. This is defaulted in GK so if i were you I wouldn't touch it.(might have confused the analyzer, who knows)

-You have a high amount of B-frames that I wouldn't think you'd need for such a high bitrate..Please tell us your B-frames settings, I have a feeling they come into the equation.You have a lot of frames with enormous quantizers, that you shouldn't even get encoding at half that bitrate.

edit:
ATTENTION: The screen color depth must be 32bit. RightClickOnDesktop->Properties->Settings.This Program was tested in Windows2000, XP and 2003RC2 only. In Win95/98/ME, it may not work properly. The Frame Size graphic isn't draw in Win98se.
It just could be your problem :)

[nightrhyme]

Thanx for reply
I didn't touch the B-Frame settings. Everything is on default except for XviD: Quantizer MPEG,VHQ:4, Qpel and CM.

I did mess with the With Modul since i don't care about compatability. I just wanted perfect AR.

About the posted ATTENTION. I don't know what you mean. I'm running Windows XP SP1 screen color depth is 32bit.
Anyone have a link that explains the different frames ?

http://img1.uploadimages.net/391982doom9.jpg

[niamh]

the Attention! was taken from the drf analyzer readme . You could try to cap the quantizers to 2-31 instead of default 1-31
http://forum.doom9.org/showthread.php?threadid=75527
http://forum.doom9.org/showthread.php?s=&threadid=80044
You can get a few explanations of why in those 2 links, even though the subject was different.

To disable B-frames, put 0 in the top box.And for all the answers you need, try Crusty's FAQ (http://www.geocities.com/atreya2011/index.htm) until you're sick of it.

Finally check your encode visually, I'm sure you can see the quality for yourself and decide if DRF analyzer screwed up or not ;) Ultimately, it's what the rip looks like, not any numbers that is a proof of quality. (Though if that analysis is anything correct, you will certainly see the disaster for yourself)

edit: Heh,I just checked the width you used(better late than never), it's not even a multiple of 16!

from the faq:-Using non-mod16 resolutions: encoding clips to resolutions of which both width and height are not a multiple of 16 is considered a no-no and is not compliant. It is quite possible to do mod-4 resolutions with many builds (don't ask me how) but is not vey good for compressibility. The smallest MPEG-4 building block is not the 8x8 blocks but the 16x16 macroblocks, and using lower-than-mod16 resolutions will make the codec use macroblocks 'outside' the frame border....which is 'A Bad Thing'(tm).
There, you're warned ;)

[nightrhyme]

Thanx so much for your answer. Very informative :)
Read the two threads you linked. As far as I understand what was wrote there. Setting b-frame 1->32 leads less often to oversizing than 2->32 leads to undersizing and that setting b-frame 1->32 gennerally leads to best quality. Is that right undetstood ?

In the threads I stumbled on Overflow Treatment Settings. Where do you adjust those ? Couldn't find them in the encoder options.

I will try and make a new encode using mod16 resolutions and see if thet helps

Again Thanx

[Tommy B.]

Q1 simply eats up bitrate and does not make the quality of the picture that much better than Q2.

It is generally suggested to encode with 2-31 instead of 1-31.

Overflow Threatment can be adjusted by clicking the MORE Button in the 2nd Pass.

[nightrhyme]

Thanx for reply.

When you say Q1. Is this what you mean ?
http://img3.exs.cx/img3/7994/default-xvid.jpg

And when you say Q2. Is this what you mean ?
http://img3.exs.cx/img3/909/Q2-xvid.jpg

Didn't know whether or not I was supposed to change all the Quantizers or just B-Frame.

I'm also having a hard time figuring out whether to leave Trellis on or not ?

[Tommy B.]

That's exactly what I mean. Q stands for quantizer.

Trellis: I think you should leave this checked for best possible quality.

[nightrhyme]

Thanx again :D
Will conduct some tests :)

[manono]

Hi-

It looks to me as if you're trying to learn advanced encoding techniques, but have neglected to master the basics. For example, as has already been mentioned by niamh, your resolution is no good at all. It's Mod 4. There's no guide that recommends that. You changed the W and H-Moduls in GKnot from their default 32/16 to 4/4. You should set them back to default, or at least to 16/16. There are good reasons why the default values are set as they are.

The reason why your ave quant is so high is because you selected a resolution too high for the movie. I would guess that you didn't run a compression test first, to help you determine a suitable resolution. Maybe try 640x352 for that movie. But that's probably still too high.

It says:
Average Frame quality is LOW [Average DRF/quantizer is 5.13]
Standard Deviation: Quality is LOW
How come? What could I do to better this?

To improve the frame quality, a lower resolution (as mentioned) will help, and perhaps some judicious filtering may help. But because you use B-Frames at standard settings, about the best you can hope for is to improve that figure to MEDIUM. HIGH is almost certainly out of the question unless you drop the B-Frames entirely, or at least drastically alter the settings (to something like, for example, 1/0.9/1). I am not necessarily recommending that you do that. The B-Frame quants are a lot higher than their corresponding P-Frame quants. So I wouldn't place much importance in getting a HIGH figure. MEDIUM is good enough. In addition, you set the end credits encoding for .20 in the Weight Zone. So you can expect all of them to be at very high quants (look at the graph for the end of the movie-all red?), and that will, of course, skew the results for the whole movie. But it looks to me like you have many other parts of the movie using high quants as well (red parts in the graph).

I happen to believe that a good Standard Deviation result (at least MEDIUM, and preferably HIGH) is important. But the only way to achieve that is to heavily restrict your quant spread. As rightly recommended by niamh and Tommy B., change the minimum I, P, and B-Frame quants to 2. There isn't really any need to use quant 1, and when it is used, it forces other quants higher to balance out. If you want to experiment with restricting the quant range, then maybe try 2/4 2/6 2/6 for starters. But if you don't know what you're doing with that, it's easy to get oversized.

Don't use Packed Bitstream (called Packeted Frames in DRF Analyzer).
Do use Trellis quantization.

DRF Analyzer is a real eye-opener when you use it for the first time, isn't it?

Disclaimer: These represent only the opinions of the poster, and not necessarily the official views of Doom9, the XviD developers, or the XviD encoding "Pros".

[niamh]

Bits per Pixel: 0.24 In addition, you set the end credits encoding for .20 in the Weight Zone.
Ha, missed that. that would explain some of the big Quants. Still, DRF>9 32751 13.5%!!! That's a lot...
Here are some results with an encode I made, and where I know I "damaged" the credits :D. I know it can't really be compared, but it gives an idea, and that's with xvid quants defaults, nothing is restricted, AND it has B-frames.(The output bitrate is significantly lower too).I seem to remember that filtering was very light too, nothing more than a little temporal soften.
Codec: XviD0034
Resolution: [ Width: 640 Height: 272 ]
Average Frame quality is HIGH [Average DRF/quantizer is 3.15]
Standard Deviation: Quality is MEDIUM [Std. Deviation is 1.01]
Image Resolution is HIGH
Performance Caracteristics:
Macroblocks per frame: 680
The Width is multiple of 32

Kilobits per Second: 931.35
Kilobits per Frame: 37.24
Kilobits per Macroblock: 0.055
Bits per Pixel: 0.22

Frame Type Statistics :
I Frames: 1.03%
P Frames: 31.87%
B Frames: 34.97%
S Frames: 32.13%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 35763 26.8%
DRF=3: 52608 39.4%
DRF=4: 26041 19.5%
DRF=5: 12769 9.6%
DRF=6: 1699 1.3%
DRF=7: 254 0.2%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 4370 3.3%
KeyF/DeltaF: 1.04%
KeyDRF<4: 1374
KeyDRF=4: 0
KeyDRF>4: 20

AverageKeyDRF: 2.61
MAXDRF: 31
AverageDRF: 3.15
Deviation: 1.01



Kilobits per Second: 1552.84
Kilobits per Frame: 62.11
Kilobits per Macroblock: 0.060
Bits per Pixel: 0.24
That's high, and doesn't suggest the resolution would really need to be lowered(it's why I didn't talk of it, and think the problem lies somewhere else)...what do you think manono? (I know...compressibility test ;) but all the same, even if the movie isn't very compressible, it doesn't explain that horrible result... I've the feeling the fancy resolution is responsible. But unfortunately we will never know unless nightrhyme reencodes the movie with the very same setting but with a decent resolution.

[manono]

Hi-

Cool, you got HIGH frame quality with normal settings. I stand corrected. However, the use of GMC seemed to knock down the percentage of B-Frames, and helped to keep the average quant low. Plus, your movie seems to be pretty compressible. But I don't know much about GMC (the S-Frames in DRF Analyzer), since I don't use it, as my standalone doesn't support it.

Although nightrhyme's BPP isn't much different from yours, evidently his movie is much less compressible. I've never encoded the Godfather movies, but I've read that's the case. In addition, because he used quant 1, XviD's rate control went beserk, and gave him frame quants all over the place. Just using a minimum of quant 2 should improve the look of it by quite a bit.

Here's a typical one of mine:

DivX DRF Analyzer v0.9.5 Report!
File Name: E:\Movies\The Hypnotist.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 640 Height: 352 ]
Frame Rate: 23.976 frames per second
The Video has 157224 frames [ 01:49:17 ]

Average Frame quality is MEDIUM [Average DRF/quantizer is 3.21]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.74]
Image Resolution is HIGH

There are NO frame drops ( NO drops is better )

Recomended Resolution: [592x320] (Target DRF/quantizer=2.8)

Performance Caracteristics:
Macroblocks per frame: 880
The Width is multiple of 32

Kilobits per Second: 650.93
Kilobits per Frame: 27.14
Kilobits per Macroblock: 0.031
Bits per Pixel: 0.12

Frame Type Statistics :
I Frames: 1.23%
P Frames: 50.71%
B Frames: 48.05%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 27140 17.5%
DRF=3: 63622 41.0%
DRF=4: 60925 39.2%
DRF=5: 1847 1.2%
DRF=6: 0 0.0%
DRF=7: 1750 1.1%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 1.25%
KeyDRF<4: 1902
KeyDRF=4: 0
KeyDRF>4: 38

AverageKeyDRF: 2.06
MAXDRF: 7
AverageDRF: 3.21
Deviation: 0.74

Is that too small? I usually wind up with 47-48% B-Frames, which throws the frame quality to Medium (although this average quant is only slightly higher than yours). Except for the end credits which were quant 5 (all the quant 5s and 7s in the figures), the settings were 2/2 2/3 2/3, with 1/1.25/1, using the EQM v2 matrix. This kept the graph entirely green, with yellow for the end credits. Note the low BPP of 0.12. AC3 2.0 audio also. 109 minutes long, all on 1 CD. Real compressible movie.

[niamh]

I've analyzed several of my movies and the result seems to point towards the fact that compressibility test is indeed very important and bits/pixel mean nothing. My turn to stand corrected :) (you get much better results than me with a particular 2 CD rip @ 1550kbps and 0.24bits/pixel)
Your movie is indeed very compressible :eek: But time for me to tighten the codec too and see what happens.
But I don't know much about GMC (the S-Frames in DRF Analyzer), since I don't use it, as my standalone doesn't support it.
You might wish it did now :D
I will keep using GMC for sure, it seems extremely effective.If, for the same output size I can get lower quants, why then, it's tempting :)(I'm going to test that out, I can't really figure how it happens)


HHhhhhhhhmmmm, I ran an analysis of that same movie without credits, and this time I get an average DRF of 3.16 :confused: , instead of 3.15 :/. Anyway the other frames percentage looks correct(all DRF 9 gone).It looks like the credit correction in the analyzer works, and that high quant credits do not affect the general DRF value.

[nightrhyme]

Again Thanx for your reply's.

Made 2 tests:
Mod 16:W and H-Moduls = 16,16, trellis, quant 2-31,mpeg. vhq 4, chroma motion.One with GMC and one without.
They became quite undersized. Aimed for 350MB for the 20 min. clip but both was about 220MB.
Perhaps you would care to comment on the reports ?

Thanx in advance

DRF ANALYSER REPORTS:

TEST 1

Encode Settings: Mod 16, trellis, quant 2-31,mpeg. vhq 4, chroma motion, GMC


DivX DRF Analyzer v0.9.5 Report!
File Name: D:\TMP\test.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 624 Height: 352 ]
Frame Rate: 25.000 frames per second
The Video has 32842 frames [ 00:21:53 ]

Average Frame quality is MEDIUM [Average DRF/quantizer is 3.25]
Standard Deviation: Quality is MEDIUM [Std. Deviation is 0.97]
Image Resolution is MEDIUM

There are NO frame drops ( NO drops is better )

Recomended Resolution: [576x320] (Target DRF/quantizer=2.8)

Performance Caracteristics:
Macroblocks per frame: 858
The Width is not multiple of 32. May degrade performance in some systems.

Kilobits per Second: 1102.18
Kilobits per Frame: 44.06
Kilobits per Macroblock: 0.051
Bits per Pixel: 0.21

Frame Type Statistics :
I Frames: 0.80%
P Frames: 15.28%
B Frames: 62.42%
S Frames: 21.50%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 12080 37.1%
DRF=3: 0 0.0%
DRF=4: 20500 62.9%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.80%
KeyDRF<4: 262
KeyDRF=4: 0
KeyDRF>4: 0

AverageKeyDRF: 2.00
MAXDRF: 4
AverageDRF: 3.25
Deviation: 0.97


TEST 2

Encode settings: mod 16, trellis, quant 2-31,mpeg. vhq 4, chroma motion

DivX DRF Analyzer v0.9.5 Report!
File Name: D:\TMP\test2.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 624 Height: 352 ]
Frame Rate: 25.000 frames per second
The Video has 32842 frames [ 00:21:53 ]

Average Frame quality is MEDIUM [Average DRF/quantizer is 3.25]
Standard Deviation: Quality is MEDIUM [Std. Deviation is 0.97]
Image Resolution is MEDIUM

There are NO frame drops ( NO drops is better )

Recomended Resolution: [576x320] (Target DRF/quantizer=2.8)

Performance Caracteristics:
Macroblocks per frame: 858
The Width is not multiple of 32. May degrade performance in some systems.

Kilobits per Second: 1120.00
Kilobits per Frame: 44.77
Kilobits per Macroblock: 0.052
Bits per Pixel: 0.21

Frame Type Statistics :
I Frames: 0.80%
P Frames: 36.78%
B Frames: 62.42%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 12080 37.1%
DRF=3: 0 0.0%
DRF=4: 20500 62.9%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.80%
KeyDRF<4: 262
KeyDRF=4: 0
KeyDRF>4: 0

AverageKeyDRF: 2.00
MAXDRF: 4
AverageDRF: 3.25
Deviation: 0.97

[manono]

You saturated the codec. Actually, this is what minimum quant 1 was made the default for, to keep from getting undersized files. But if you had set up the encode properly to begin with (and had run a compression test in advance to alert you to the problem), you wouldn't have come out undersized.

All your P-Frames are quant 2. It's impossible for you to reach the desired file size with those settings. Set your size for, say, 175 MB, and run your 2nd pass again.

[nightrhyme]

I see what you are saying. The Comp test is 154% at my desired target file size = 320MB. At file Size = 200MB it's 90%.
The bitrate of the rip using Quant 2-31 is only 1200 kb/s.

So now my questions is. How do I get a High Bitrate (about 1700-1800 Kb/s) without saturating the codec when using Quant 2-31 ?
I would like to make a 3 CD (2800MB)rip of this movie.

BTW: Did try setting the Overflow Settings to 20%. That seems to have no effect.

Again Thanx for your insight

[Tommy B.]

If you want a higher bitrate, lower "Motion Search Precision" and "VHQ Mode". Also don't use B-Frames and increase resolution of the movie...

[nightrhyme]

Thanx for reply :)

How exactly do I disable B-frames ?
And what about the I frames ?
Whould these be my Xvid settings ? (See Images):

http://img13.exs.cx/img13/5135/xvid1.jpg

I guess Unticking B-VOPs disable B-Frames ?
http://img64.exs.cx/img64/9848/xvid3.jpg

http://img64.exs.cx/img64/8881/xvid2.jpg


Is the only solution to disable B frames ?
Granted I don't know much about XviD I seem to recall having read some posts about not to do so.

Thanx in advance

[killingspree]

from niamh's post - on the previous page!
To disable B-frames, put 0 in the top box.And for all the answers you need, try Crusty's FAQ until you're sick of it.


Another way to increase the max bitrate, is to use sharp avisynth filters and no noise reduction!

for example use lanzcos resize for resizing, remove undot() from the script (gknot uses it automatically) and also remove any noise reduction! if you want to go even further you can apply sharpening filters like msharp (by d. graft) or unfilter (by t. barry).

perhaps manono has some more hints on that!

hth
steVe

[manono]

Hi-

My hint to boost the file size is to use a high bitrate matrix. There are some included with the XviD codec. You'll have to unzip them, and change from H.263 or MPEG to Custom Matrix, hit the Edit Matrix button, and then browse to the matrix you want to use. One high bitrate matrix you might try is the andreas_78er Matrix. Personally, for easy to compress sources, I'm partial to Selur's Matrix.

You can get a bunch more matrices including the Selur Matrix here (http://www.ligh.de/software/qmatrix.zip).

For matrix comparisons, you might try here (http://forum.doom9.org/showthread.php?s=&threadid=76021&highlight=Selur+matrix), and especially here (http://forum.doom9.org/showthread.php?s=&threadid=73666&perpage=20&highlight=Selur%20matrix&pagenumber=3) where, about a third of the way down, you'll see Soulhunter's matrix comparison by file size.

If you don't care much about the final file size, or if your source is pretty compressible, you might be surprised at the difference a good high bitrate matrix can make in the amount of detail that's preserved.

[nightrhyme]

Thanx so much for all you surgestions. Will try them.

Did in fact try with the Matrix called HVS_BEST_PICTURE because I read good things about it. But it had no effect on the filesize. Still undersized.

I guess I'll try Selur's Matrix. Should I still keep all Quant's at 2-31 and disable B-Frames ?

[nightrhyme]

Just made 2 tests. Bumped up the Resolution. No more undersizing. Now I just have to figure out which of these two encodes is the better ? They look almost identical so they are really hard to distinguish from each other. But looking at the reports for you wise people what is the better rip ?

First Test:
No B-Frames. All Quants 2-31. Matrix = HVS_BEST_PICTURE
VHQ = 1 (for speed), Chroma motion, Trellis. The rest DEFAULT

DivX DRF Analyzer v0.9.5 Report!
File Name: D:\TMP\no b frames.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 736 Height: 416 ]
Frame Rate: 25.000 frames per second
The Video has 32843 frames [ 00:21:53 ]

Average Frame quality is HIGH [Average DRF/quantizer is 2.55]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.50]
Image Resolution is HIGH

There are NO frame drops ( NO drops is better )

Recommended Resolution: [768x432] (Target DRF/quantizer=2.8)
The filesize can be smaller!

Performance Caracteristics:
Macroblocks per frame: 1196 ( Poor Playback in Slow Computers, PIII450 or better required )
The Width is multiple of 32

Kilobits per Second: 1849.30
Kilobits per Frame: 73.93
Kilobits per Macroblock: 0.062
Bits per Pixel: 0.25

Frame Type Statistics :
I Frames: 0.82%
P Frames: 99.18%
B Frames: 0.00%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 14867 45.6%
DRF=3: 17707 54.4%
DRF=4: 1 0.0%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.82%
KeyDRF<4: 268
KeyDRF=4: 0
KeyDRF>4: 0

AverageKeyDRF: 2.75
MAXDRF: 4
AverageDRF: 2.55
Deviation: 0.50



Second Test: All Quants 2-31. Matrix = HVS_BEST_PICTURE
VHQ = 1 (for speed), Chroma motion, Trellis.
Max BVOP's : 1 <-- Suggested by Manono (What is the relationship between these numbers ?)
Quantizer Ratio: 0,90 <-- Suggested by Manono (What is the relationship between these numbers ?)
Quantizer Offset: 1 <-- Suggested by Manono (What is the relationship between these numbers ?)
The rest DEFAULT.
NOTE: I inadvertanly made this rip 10MB smaller than the other.


DivX DRF Analyzer v0.9.5 Report!
File Name: D:\TMP\altered b frame setings high res.avi
FourCC: XVID
Codec: XviD0035
Resolution: [ Width: 736 Height: 416 ]
Frame Rate: 25.000 frames per second
The Video has 32843 frames [ 00:21:53 ]

Average Frame quality is HIGH [Average DRF/quantizer is 2.49]
Standard Deviation: Quality is HIGH [Std. Deviation is 0.50]
Image Resolution is HIGH

The file has Packeted Frames!

There are NO frame drops ( NO drops is better )

Recomended Resolution: [768x432] (Target DRF/quantizer=2.8)
The filesize can be smaller!

Performance Caracteristics:
Macroblocks per frame: 1196 ( Poor Playback in Slow Computers, PIII450 or better required )
The Width is multiple of 32

Kilobits per Second: 1786.82
Kilobits per Frame: 71.43
Kilobits per Macroblock: 0.060
Bits per Pixel: 0.24

Frame Type Statistics :
I Frames: 0.81%
P Frames: 50.27%
B Frames: 48.92%
S Frames: 0.00%
N Frames: 0.00%
(More Advanced Codecs use B and S frames)
Frame Quality Statistics :

DRF=1&2: 16639 51.1%
DRF=3: 15937 48.9%
DRF=4: 0 0.0%
DRF=5: 0 0.0%
DRF=6: 0 0.0%
DRF=7: 0 0.0%
DRF=8: 0 0.0%
DRF=9: 0 0.0%
DRF>9: 0 0.0%
KeyF/DeltaF: 0.82%
KeyDRF<4: 267
KeyDRF=4: 0
KeyDRF>4: 0

AverageKeyDRF: 2.56
MAXDRF: 3
AverageDRF: 2.49
Deviation: 0.50


On a side note. Reading this thread http://forum.doom9.org/showthread.php?s=&threadid=63492&highlight=undersizing about Undersizing. I can't figure out what the 4 numbers beeing discussed are and where they are located in the XviD Codec ?
I guess they are Max BVOP's, Quantizer Ratio and Quantizer Offset but that only accounts for 3 of them.
My current config:
6 - Ultra High
MPEG
XVID
1 - Mode decision (or 4)
300
1
Use chroma motion
(sometimes quarterpel)

2 <-- these numbers
150 <-- these numbers
100 <-- these numbers
0 <-- these numbers

DX-50 B-VOP


Thanx in advance

[manono]

That 4th number is the B-Frame sensitivity. You can find it in the Zone Options.

[nightrhyme]

Originally posted by manono
That 4th number is the B-Frame sensitivity. You can find it in the Zone Options.

Ok Thanx.
Any info on my two new tests ?

[manono]

The DRF Analyzer reports look pretty much the same to me. One .avi uses no B-Frames and no Packed Bitstream. The other uses B-Frames at very conservative settings and also uses Packed Bitstream. Both .avis should look very good.

[nightrhyme]

Originally posted by manono
The DRF Analyzer reports look pretty much the same to me. One .avi uses no B-Frames and no Packed Bitstream. The other uses B-Frames at very conservative settings and also uses Packed Bitstream. Both .avis should look very good.

Yeah. They both look fine.
But since I'm very new to this I'm having trouble figuring out whether using some B-frames or None at all would be the better choice :confused:

Packet bitstream was used inadvertanly and will not be used in the final rip