Brute Anti Blocking Matrix

[apfraats]

Hi everybody, nice to see yu again !

(I have done my suspended vacation :) )

Before and during my vacation I tried to do anything in my power to prevent CCE from blocking. With the standard matrices CCE blocks a lot but I still consider it a good encoder as my Sony can handle this one without problems whatever buttons I use and whatever trick I'll do....)

(anybody wants it ???, please do an offer, youre garantueed to get a piece of equipment that will faill if there's just 1 bit wrong.......) :)

But back to blocking.

I have developped (see me, I even don't know a thing about it!) a matrix that will as strong as possible force CCE not to block.

At every bitrate used.

So don't laugh, try it on youre blocky scenes and enjoy !
The matrix is called BAB as for Brute Anti Blocking..........
You'll see why.......
It even procduces a picture ! (whow, I didn't expect that when doing my as usual black/white analysis)

So here they are, and they REALLY HELP in CCE's anti-blocking:


08 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127

16 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127
127 127 127 127 127 127 127 127

I agree, they look absolutely stupid........

But however they work at least in the range 3000-4000 average material with high motion high action scenes and are specially tried on DIE ANOTHER DAY from James Bond keeping DTS,5.1,2,2 Audio (over 2 hour movie), chapter 4 including high motion high action (explosions, hovercrafts, shooting, blowing up things e.d.)

Try them, they work... even better as Very Low Bitrate or Angels best Very Low matrices.....

If you are like me, hating blocky scenes, despite how ridiculous they look, try them !!!!!

Call them the BAB-matrix...... Brute (force) Anti Blocking.

Compare youre high motion/high action scenes from CCE (default) with the result done with this matrix.....

[SAPSTAR]

Of course it wont block, you remove every detail !!!!!! :) Sorry to laugh but with such a matrix you can as well backup your DVDs with with a filter at its maximum smoothness settings !!!!

[apfraats]

Nope, CCE tends to worry too much about details.....

It's not comparible in any way to UNDOT().DEEN(), and as tested this performs better keeping much more detail then using filters.

A noisy source will stay noisy......

And even then, using a filter takes time, a different matrix not....

And even using anti-noise, smoothing filters result in blocky pictures, this matrix just tells CCE to hold on with it's useless waste of bitrate in details we never see. The blockiness however is much more disturbing.

Just try and see..... Even I was suprised too.......

[SAPSTAR]

Nope, CCE tends to worry too much about details.....

It's not comparible in any way to UNDOT().DEEN(), and as tested this performs better keeping much more detail then using filters.

A noisy source will stay noisy......

And even then, using a filter takes time, a different matrix not....

And even using anti-noise, smoothing filters result in blocky pictures, this matrix just tells CCE to hold on with it's useless waste of bitrate in details we never see. The blockiness however is much more disturbing.

Just try and see..... Even I was suprised too.......
Just a question, are you using the Quantization Adaptive mode in CCE SP ?!? If yes, it overrides a big part of your given matrix...anyway, read carefully how works quantization matrices(http://forum.doom9.org/showthread.php?p=318172#post318172), your matrix removes every detail, it may be better for example on LCD screens which tends to enhances "defects" but on a regular TV or a DLP, it's better to keep as much detail as possible....A simple test is to check the PSNR, you will see that you left much details....Why do think people are working at finding the matrix which will give the best compromise quality/detail ?! If it was that simple it would have been hardcoded in the encoders themselves !

[apfraats]

Brute Anti Blocking is already stating what it does.....

so BAB.

Nope, not adaptive quantization, as I use 2.67.0.10, not 2.70...

It started all as a joke......

Untill I saw the results, on a tube-based TV that has internal 20 MHZ bandwidth and 1680 * 833 internal resoluion..... So see very little detail on this particular 34" inch set.

Not that much detail is lost as you pretend, it's not introducing 'artificial image properties' as all filters seem to do. And sharpness is still way up.

This is because of CCE. Every encoder with the same matrix used will be behave differently.

This matrix seems to be a complete overdose, but CCE is so trigger-happy on details it hardly touches image details at all.

Then again, this is just a anti-blocking highest priority matrix, that's especially usefull in CCE as it doesn't really supresses any detail noticable to me when genreally looking.

Everybody who hates blockiness but likes CCE has to try this one and even see what they want.

CCE isn't going to give up on details, even not with this matrix...

Why ? ask the designers.......

People can give it a try and decide to use this option or not.....

I use it al lot when blockiness pictures appear in high motion high action scenes as CCE tends from nature to give good results at higher average bitrates only....

This matrix just puts down the border at which real blocking is occuring, without going to offer details too much...

Blockiness is the most annoying 'artifact' to be seen IMO.

So let people try and decide for themselves.

I now use this matrix more and more depending on average bitrate and source material.

It's really not like an out of focus or heavy smoothened image that you'll get.

If people get blockiness pictures they can use this with CCE, and decide for themselve what's best.

Do not say anything about matrices in general, even some are causing more blockky pictures than MPEG standard using with CCE.

I tried all the matrices that promised to give less blocky pictures in GENERAL. But this absurd matrix really calms CCE down.......

Just try and see, decide to use it in specific situations or not.

There is freedom of choice.....

As I notice artifacts in originally mastered movies, like blockiness in commercial releases, I think my TV set gives a bit too much detail.

First try then complain.....

[manono]

Hi-

Of course it wont block, you remove every detail !!!!!!

Hehe. He said it didn't block. He didn't say too much about the detail loss. A question of priorities, I guess. He hates blocking.

this matrix just tells CCE to hold on with it's useless waste of bitrate in details we never see.

But that's not true, as you're not only removing high frequency detail, but low frequency detail, and a lot of it. I've been using this matrix on my low bitrate encodes recently. It comes from a commercial DVD. It compresses about the same as the Standard matrix, but holds more detail, and compresses the heck out of the high frequencies. You might try it out, or something between this one and yours and see if you like it at all. Don't you have the Philips TV set? I can't remember, but if it's an LCD, then those are about the worst for showing blocking. A good calibration would help, but even DLPs are better about blocking than are LCDs. Maybe we've been through this before:

08 12 13 14 15 16 19 22
12 13 14 15 16 19 22 26
13 14 15 16 19 22 26 32
14 15 16 19 22 26 32 41
15 16 19 22 26 32 41 53
16 19 22 26 32 41 53 70
19 22 26 32 41 53 70 94
22 26 32 41 53 70 94 127

12 12 13 14 15 16 19 22
12 13 14 15 16 19 22 26
13 14 15 16 19 22 26 32
14 15 16 19 22 26 32 41
15 16 19 22 26 32 41 53
16 19 22 26 32 41 53 70
19 22 26 32 41 53 70 94
22 26 32 41 53 70 94 127

[Boulder]

Doesn't that matrix introduce noticable ringing due to the excessive high-frequency compression?

[SAPSTAR]

Hi-

Of course it wont block, you remove every detail !!!!!!

Hehe. He said it didn't block. He didn't say too much about the detail loss. A question of priorities, I guess. He hates blocking.

this matrix just tells CCE to hold on with it's useless waste of bitrate in details we never see.

But that's not true, as you're not only removing high frequency detail, but low frequency detail, and a lot of it. I've been using this matrix on my low bitrate encodes recently. It comes from a commercial DVD. It compresses about the same as the Standard matrix, but holds more detail, and compresses the heck out of the high frequencies. You might try it out, or something between this one and yours and see if you like it at all. Don't you have the Philips TV set? I can't remember, but if it's an LCD, then those are about the worst for showing blocking. A good calibration would help, but even DLPs are better about blocking than are LCDs. Maybe we've been through this before:

08 12 13 14 15 16 19 22
12 13 14 15 16 19 22 26
13 14 15 16 19 22 26 32
14 15 16 19 22 26 32 41
15 16 19 22 26 32 41 53
16 19 22 26 32 41 53 70
19 22 26 32 41 53 70 94
22 26 32 41 53 70 94 127

12 12 13 14 15 16 19 22
12 13 14 15 16 19 22 26
13 14 15 16 19 22 26 32
14 15 16 19 22 26 32 41
15 16 19 22 26 32 41 53
16 19 22 26 32 41 53 70
19 22 26 32 41 53 70 94
22 26 32 41 53 70 94 127

:goodpost: :thanks: Exactly what I meant !!!!!

[roux]

Hm his account is already deleted, gopper apfraats dvdbackup all the same :P

[manono]

Hi Boulder-

You talking to me? Introducing ringing? No, I don't think so. At the time I checked the e-mail about your post, I had just finished encoding an extra using that matrix. So I had a good close look at it to check. Are you using ringing in the halos/edge enhancement meaning, or in the mosquito noise/Gibbs artifacts meaning? I've always taken ringing to mean edge enhancement, but I've seen it used the other way also. Anyway, I didn't see either.

I probably should have phrased it differently in the earlier post. Instead of saying, "compresses the heck out of the high frequencies", I should have said something like "it allows for very good compression by removing a lot of high frequency detail". I kind of like that matrix, not for a main movie, unless I have to give iit a pretty low bitrate, but for lower bitrate extras in the 2500-3000 range.

Also, I was just checking my notes, and I was incorrect when I said it compresses similarly to the Standard Matrix. I meant to say that it compresses similarly to the MPEG Standard Matrix of CCE. It doesn't compress nearly well as the Standard Matrix, which I don't like at all.

EDIT: I just saw roux's post. You know, I was wondering if they were all one and the same. I had pretty much decided that apfraats and dvdbackup were the same, and then GOPPER showed up recently with the same confrontational style.

Anyway, apfraats, if you're still reading, your blocking problem sounds real serious if you have to use that matrix to get it under control. I'd suggest a quick 15 minute calibration with the THX Optimizer. It's on all the Star Wars DVDs, and many others as well. You may wind up lowering the brightness a lot, and maybe boosting the contrast a bit, and that should help you with your problem.

[Boulder]

Hi Boulder-

You talking to me? Introducing ringing? No, I don't think so. At the time I checked the e-mail about your post, I had just finished encoding an extra using that matrix. So I had a good close look at it to check. Are you using ringing in the halos/edge enhancement meaning, or in the mosquito noise/Gibbs artifacts meaning? I've always taken ringing to mean edge enhancement, but I've seen it used the other way also. Anyway, I didn't see either.

I meant Gibbs which is quite irritating at times;)

I probably should have phrased it differently in the earlier post. Instead of saying, "compresses the heck out of the high frequencies", I should have said something like "it allows for very good compression by removing a lot of high frequency detail". I kind of like that matrix, not for a main movie, unless I have to give iit a pretty low bitrate, but for lower bitrate extras in the 2500-3000 range.

I'll have to give that matrix a go then. I often end up with avg bitrates of 2000-3000kbps as I mostly put two movies per disc. And yes, they look good to my eyes;) Then again, I don't have a huge TV to reveal the horrible truth.

By the way, do you know any decent matrices for interlaced stuff? mb1's interlaced matrix requires quite a high bitrate IMO, I've been looking for something that would settle for a bit less.

[manono]

Nah, I don't know anything about interlaced matrices. That's FredThompson's specialty. I hardly encode interlaced material, except for talking heads on extras, and I IVTC them. :)

[Boulder]

That's FredThompson's specialty.
He only does wrestling. There's no matrix that can make it look appealing :D

[chrisjim]

@manono,

I'm intrigued. Which particular (commercial) dvd did you get the matrix from?

[manono]

Hi-

I'm sorry, chrisjim, but I don't remember. I got it some time ago, and didn't note where I got it. It may have been from an anime DVD, but I wouldn't swear to it. Rockas had included it in his Matrix Editor at my suggestion, as the #6 Medium_Low Matrix, listed as compressing better than the Standard Matrix. I finally got around to running some tests and discovered that it's about the same in compressibility as the #4, Medium_Hgh MPEG Standard Matrix. If you're implying with your question that it's strange that a matrix like that would be used for a main movie, then I agree. How about this one:

08 05 07 09 11 13 14 17
05 06 08 11 13 14 17 18
07 09 10 11 14 17 17 19
09 10 13 13 14 17 18 127
10 13 13 14 16 17 127 127
13 13 14 16 17 20 127 127
13 13 14 17 19 23 127 127
13 14 17 19 127 127 127 127

08 05 07 09 10 10 11 11
05 06 09 10 10 11 11 12
07 09 10 10 11 11 12 12
09 10 10 11 11 12 13 127
10 10 11 11 12 13 127 127
10 11 11 12 13 13 127 127
11 11 12 13 13 14 127 127
11 12 12 13 127 127 127 127

I just discovered this one yesterday on an R1 DVD of a Japanese film. It's the only matrix used:

http://www.imdb.com/title/tt0142998/

I also find this one iinteresting, and plan to run some tests to see how it compares to others. It goes along from upper left to lower right as a very fine sharp and detailed matrix, and then just falls over the edge (where it doesn't matter, I think).