collection of available AVC custom quant matrices

[nimitz]

Hi Sharktooth,

Thanks for the reply. I have access to the IEEExplorer(well...my school has its subscription.) Do you have the title of the paper? Or it's a bunch of papers from which ReferenceDivX derived his HVS Good matrix? I have read Nill's 1985 paper and some other papers such as Ngan's 1989 paper and K.R.Rao's 1990 paper, but seems none would lead to the following weighting matrix as ReferenceDivX used in the other thread.

double hvsfwm[64]={ 1,1,1,1,1,1,1,1,
1,1,.9599,1,.9571,1,.9599,
.8746,.9283,.8898,.8898,.9283,
.8746,.7684,.8404,.8192,.7617,
.8192,.8404,.7684,.6571,.7371,
.7371,.6669,.6669,.7371,.7371,
.6571,.6306,.6471,.5912,.5419,
.5912,.6471,.6306,.5558,.5196,
.4564,.4564,.5196,.5558,.4495,
.3930,.3598,.3930,.4495,.3393,
.2948,.2948,.3393,.2480,.2278,
.2480,.1828,.1828,.1391};


Thank you.

Originally Posted by Sharktooth

the documents are available here: http://ieeexplore.ieee.org but they want $$$...

[Sharktooth]

Sorry i dont remember the paper title

[NeD tHe OnE]

Originally Posted by Oline 61

BTW, here is an archive with all the matrices listed in this thread. If you don't want your matrix in this archive, tell me so and I will remove it.
http://home.nc.rr.com/oline61/x264_custom_matrices.7z

I am currently planning to generate several clips designed to test for specific behaviors, and then encode with different quant matrices, and test PSNR and SSIM. What settings should I use to do this?

I don't think --qp or --crf will work because they produce different sizes with different matrices which is no good for SSIM and PSNR testing. Should I do 2-pass or 1-pass ABR? For 2-pass do I need to do a new first pass for each matrix? Should I use slow high quality settings, or faster settings? What bitrates would you like to see represented?

http://home.nc.rr.com/oline61/x264_custom_matrices.7z

I need this but its not there in the server

[Oline 61]

Originally Posted by NeD tHe OnE

http://home.nc.rr.com/oline61/x264_custom_matrices.7z

I need this but its not there in the server

Fixed.

[Oline 61]

Does anyone want to recommend deblocking settings for any of the matrices, or should I continue with -3:-3?

I am thinking of doing some clips of the Nebuchadnezzar, for high detail and motion, maybe some of the green text, although the source is blocky there already, and probably a few fighting scenes.

[Sharktooth]

not me... it really depends on the source and the quant/bitrate you use (i know deblocking is adaptive but it also influences compressibility, so...)

[Oline 61]

How about The Matrix Reloaded at 1600kbit/s. No processing, just IVTC crop and Lanczos4Resize(848,352).

[NeD tHe OnE]

Hey Oline! Wich matrix is better 4 movies like The Patriot .. n what bitrate should I go 4?

[Sharktooth]

Originally Posted by Oline 61

How about The Matrix Reloaded at 1600kbit/s. No processing, just IVTC crop and Lanczos4Resize(848,352).

i would go anamorphic with no resize at all...

[*.mp4 guy]

-3:0 is a good setting most of the time, thats what I recommend for my cqm.

[Sharktooth]

Originally Posted by NeD tHe OnE

Hey Oline! Wich matrix is better 4 movies like The Patriot .. n what bitrate should I go 4?

Stop asking what's best and what bitrate you should use.
and .
And FFS

[Desmodeus]

I've been busy writing some custom matrices of my own , and I decided I shouldn't keep them all to myself ...plus I'm kinda interested on an Expert opinion of them.

So without further ado, I give you:

#Superior Custom Matrix

INTRA4X4_LUMA =
8,12,16,24,
12,16,24,32,
16,24,32,40,
24,32,40,48

INTRA4X4_CHROMAU =
8,12,16,24,
12,16,24,32,
16,24,32,40,
24,32,40,48

INTRA4X4_CHROMAV =
8,12,16,24,
12,16,24,32,
16,24,32,40,
24,32,40,48

INTER4X4_LUMA =
8,12,16,20,
12,16,20,24,
16,20,24,32,
20,24,32,40

INTER4X4_CHROMAU =
8,12,16,20,
12,16,20,24,
16,20,24,32,
20,24,32,40

INTER4X4_CHROMAV =
8,12,16,20,
12,16,20,24,
16,20,24,32,
20,24,32,40

INTRA8X8_LUMA =
8,12,14,16,20,22,24,28,
12,14,16,20,22,24,28,32,
14,16,20,22,24,28,32,34,
16,20,22,24,28,32,34,36,
20,22,24,28,32,34,36,40,
22,24,28,32,34,36,40,42,
24,28,32,34,36,40,42,44,
28,32,34,36,40,42,44,48

INTER8X8_LUMA =
8,10,12,14,16,18,20,22,
10,12,14,16,18,20,22,24,
12,14,16,18,20,22,24,26,
14,16,18,20,22,24,26,28,
16,18,20,22,24,26,28,32,
18,20,22,24,26,28,32,34,
20,22,24,26,28,32,34,36,
22,24,26,28,32,34,36,40

And it's corresponding x264 commandline:

--cqm4i 8,12,16,24,12,16,24,32,16,24,32,40,24,32,40,48 --cqm4p 8,12,16,20,12,16,20,24,16,20,24,32,20,24,32,40 --cqm8i 8,12,14,16,20,22,24,28,12,14,16,20,22,24,28,32,14,16,20,22,24,8,32,34,16,20,22,24,28,32,34,36,20,22,24,28,32,34,36,40,22,24,28,32,34,36,40,42,24,28,32,34,36,40,42,44,28,32,34,36,40,42,44,48 --cqm8p 8,10,12,14,16,18,20,22,10,12,14,16,18,20,22,24,12,14,16,18,20,22,24,26,14,16,18,20,22,24,26,28,16,18,20,22,24,26,28,32,18,20,22,24,26,28,32,34,20,22,24,26,28,32,34,36,22,24,26,28,32,34,36,40

Composing a custom matrix was not very difficult for me, a complete newbie, but not completely uncomplicated either.
I started with a base of eight for every string so I could be sure the resulting bitrate was properly divisible, them I set an end-quantitizer which was a direct multiple of eight (40 and 48 respectively)
Then I started subtracting binary values for each stair, doubling the subtracted value where i felt detail could suffer further compression (middle and end values), making sure that I included all other multiples-of-eight along the way (40, 32, 24, and 16 respectively)
The result... every value in those matrices gives an Integer when divided by eight, which greatly aids Compression IMO, hence the presumptuous "Superior" title.

Here's how that compression looks on the 8x8 matrices for I and P frames respectively:





As far as i can tell, these matrices offer:

* slightly better compression than their flat counterparts (up to x3 on it's highest value 48)

* more stable streams than the default JVT due to better-divisible binary values (JVT uses multiples of three mostly)

* Medium-High signal-to-noise ratios which translate into better retention of detail.

But then again, these are rather presumptuous assumptions on my part, so i await further feedback, advice and constructive criticism from more experienced members of this community

[Dark Shikari]

Originally Posted by Desmodeus

* slightly better compression than their flat counterparts (up to x3 on it's highest value 48)

How is this a good thing? You can get x16 compression by setting the HFs to 256, but that's not a good idea either.

Originally Posted by Desmodeus

* more stable streams than the default JVT due to better-divisible binary values (JVT uses multiples of three mostly)

Why would even values be "more stable"?

[Desmodeus]

Originally Posted by Dark Shikari

How is this a good thing? You can get x16 compression by setting the HFs to 256, but that's not a good idea either.

I've actually come across some matrices with high-frequency values of 128+ which nonetheless pertained to be "sharp" on this very Thread.
Also... I believe the maximal hexadecimal value is 255 (or FF), so wouldn't 256 be breaking that boundary O.o?

Why would even values be "more stable"?

I simply assumed even values which could be integrally divided by an octet would be easier to compress and/or quantify ...was I wrong?

[Dark Shikari]

Originally Posted by Desmodeus

I've actually come across some matrices with high-frequency values of 128+ which nonetheless pertained to be "sharp" on this very Thread.

That doesn't have any bearing on whether high values are good or bad. "X happens to have a good Y" does not mean that X was related to the cause of a good Y.

Originally Posted by Desmodeus

I simply assumed even values which could be integrally divided by an octet would be easier to compress and/or quantify ...was I wrong?

"Easier to compress"? No, it's pretty much irrelevant...

[Lyris]

Originally Posted by Sharktooth

AVC-HR at 1600kbps...

grain is perfectly preserved...

I would struggle to call the coarse quantization inbetween the two actors "perfect".
Or is that what the already compressed DVD source looked like?

[gioppatta]

somebody knows these matrices? And what they do?

Custom intra4X4 luma:
6 7 10 16
7 7 11 17
10 11 12 20
12 13 20 16

Custom intra4X4 chromau:
16 16 16 16
16 16 16 16
16 16 16 16
16 16 16 16

Custom inter4X4 luma:
10 13 28 41
13 14 32 84
28 32 41 111
41 46 111 16

Custom inter4X4 chromau:
16 16 16 16
16 16 16 16
16 16 16 16
16 16 16 16

Custom intra8X8 luma:
9 9 10 10 11 16 21 29
9 9 10 10 11 16 21 29
10 10 10 10 11 16 22 31
10 10 10 11 11 17 23 33
11 11 11 11 11 19 25 36
12 12 13 13 14 20 27 40
15 15 16 16 21 27 31 45
20 20 21 22 36 40 45 16

Custom inter8X8 luma:
12 13 15 18 20 70 163 255
13 13 16 18 20 72 170 255
15 16 17 19 21 81 190 255
18 18 19 21 23 96 228 255
20 20 21 23 25 120 255 255
33 34 36 39 45 32 255 255
64 66 71 80 164 255 255 255
151 155 169 192 255 255 255 10

[mp3dom]

They comes from a Sony Blu-Code encode. It's one of those various matrices that the encoder may use (based on the type of footage).

[mandarinka]

Originally Posted by gioppatta

somebody knows these matrices? And what they do?

Custom intra4X4 luma:
6 7 10 16
7 7 11 17
10 11 12 20
12 13 20 16

Custom intra4X4 chromau:
16 16 16 16
16 16 16 16
16 16 16 16
16 16 16 16

Custom inter4X4 luma:
10 13 28 41
13 14 32 84
28 32 41 111
41 46 111 16

Custom inter4X4 chromau:
16 16 16 16
16 16 16 16
16 16 16 16
16 16 16 16

Custom intra8X8 luma:
9 9 10 10 11 16 21 29
9 9 10 10 11 16 21 29
10 10 10 10 11 16 22 31
10 10 10 11 11 17 23 33
11 11 11 11 11 19 25 36
12 12 13 13 14 20 27 40
15 15 16 16 21 27 31 45
20 20 21 22 36 40 45 16

Custom inter8X8 luma:
12 13 15 18 20 70 163 255
13 13 16 18 20 72 170 255
15 16 17 19 21 81 190 255
18 18 19 21 23 96 228 255
20 20 21 23 25 120 255 255
33 34 36 39 45 32 255 255
64 66 71 80 164 255 255 255
151 155 169 192 255 255 255 10

That looks like the matrix called unofficialy "Prestige". Search the forum for that.

It has been discussed for example here (original thread IIRC).

[gioppatta]

Thank you!