Should i change matrix on high compression ratio?

[roux]

I'm trying to make a backup from Saving private ryan. But the overall bitrate is a bit low:

-- Processed 242.791 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 58,2%
- Overall Bitrate : 2.882Kbs
- Space for Video : 3.417.088KB
- HIGH/LOW/TYPICAL Bitrates: 3.770/1.168/2.882 Kbs

I allready did a test encode with the standard dvdrb settings and the segment looked a bit distorted around persons. I've read about filters in a similar topic but there's a lot of grain in the source. So i was thinking maybe a different matrix would give better results. Does anyone have some advice on this one?

Thanks :)

[OvERaCiD23]

Give them a try and see if there's an improvement. I did SPR a long time ago (with the Big3) with the default CCE matrix and it looks fairly good. I'm sure another matrix would yield better results as it has on other films.

[FredThompson]

Try HC with the highest quality setting.

In my testing, HC with the normal quality setting beats 5-pass CCE. HC gives sharper output and less blockiness.

[elizerrojas]

i'm just here to say that: CCE=$58, HC= a little slower but free and better. oh yeah baby. thanks MR. hank315.
i have increased the # of backups i do with RB. i do main movie only with RB-pro/HC. in my opinion, nothing better out there.

[dvdbackup]

-- Processed 242.791 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 58,2%
- Overall Bitrate : 2.882Kbs
- Space for Video : 3.417.088KB
- HIGH/LOW/TYPICAL Bitrates: 3.770/1.168/2.882 Kbs

Thanks :)

First: Average below 3000 I consider 'mission impossible'.

So I think youre doing the whole DVD and not just the movie.

The first important steps:

1) Do a movie-only backup, get rit of things you don't need.
2) If the film is noisy, this noise eats some bitrate, try UNDOT().DEEN() in the AVS filter section of DVD-RB. And noise causes artifacts at lower bitrates.
3) Try CCE, or HC. If you have CCE SP try:

VBR_BIAS=0
Precision= 0 (due to low bitrate, if it's higher go the a higher value).
Passes: 3

4) Select the Very Low Bitrate matrix. Even if you have somewhat higher bitrate.

Try and see the result....


Good luck...

[hank315]

I tried HC-encoder and it even generates bugs in the MPEG stream. Bugs in the MPEG stream? Could you be more specific what kind of bugs except not playing reverse correct.

CCE SP has a long development history and costs about 1995,-.
But HC is still Beta, and using Beta has risks.....Both are absolutely true...
Using beta freeware is a risky business, but it is free ;)

Well, seems you just deleted my quote :D

[dvdbackup]

The bugs I mention was about not playing reverse correctly.

No other bugs were found.

Playing reverse deed seem to do steps to non I frames or something like that.

It can be the player, the medium (dvd-rw for testing) or indeed the HC MPEG stream generated.

No other bugs I found.

I did already edited the post, as I have to do more testing with HC.

I take the same matrices and same source and going to compare.

I didn't say HC is bad or so, it's still under development.

For a free encoder you always have to be glad...

But as it seems, CCE SP is free too, for some if not most people at least. They complain about costs, but sure use all CCE SP ??????
Just look at the posts......


Looks a bit strange to me............. :p

[FredThompson]

First: Average below 3000 I consider 'mission impossible'.Run the math, don't perpetuate myths. 2 hours with AC3 audio on a single-layer disc is less than 3000 for the video. I just did Tora! Tora! Tora! without filtering using HC's highest quality mode. It's virtually indistinguishable from the original and that movie is 3 1/2 hours. You must be using an incredibly poor encoding method.

[dvdbackup]

Nope, I use an incredibly good (and BIG) TV set concerning picture quality.

That's were the difference is.

Even lot's of commercial DVD's I consider bad !!!

They block al lot.... Look at the BATMAN dvd's in the past........

I just walked trough my collection to redo 'bad copies'. But when I get my hands on the original again I just see GIBBS-artifacts and blocking already in the original more than once (here it is allowed to copy what you want).

3000 Kbps/sec video is 3000 Kbps/sec video, audio is of no infuence on that, it's already calculated off the total bitrate with DVD-RB, so even DTS doesn't matter.

3000 Kbps for video will NOT do (for me) on average movies with aktion scenes.

Bad encoder ???

Yep, sure CCE SP 2.67 3 passes...... payed 1995,- is really bad ;)

TV set:

It's resolution is 1680 pixles * 833 lines. Any signal not complying to this resolution is processed and adapted to it.
Even the smallest GIBBS-artifacts are visible. Even the slightest blocking is visible.

And when I look at the smaller TV set im my bedroom, I would say every DVD is ok, or it has to be really bad.

But on the TV-set in my living room, every little detail is shown...

That's why optimum picture quality is only reached on high bitrate CBR dvd's at 8000 - 9000 Kbps level.

VBR already causes GIBBS-artifact to show up.

Below 3000 Kbps average my pictiure quality on my bedroom TV seems impressive.

Don't I try it in the living room, it's going to be a verry big dissapointment....

When you can't compare, you can't tell, as simply as that....

If I reduce sharpness to it's lowest level, the picture get's less sharp, and everything looks just fine, but not sharp. This is the effect seen on many 'digitally remastered' old recordings.

When youre TV set hasn't the capability to show details , you won't see them.

3000 Kbps is mission imposibble......... or you'll have to have a movie that shows only static scenes.... And even then the GIBBS-artifacts are seen.

So I think you have to improve youre TV set first...........

[FredThompson]

Settle down, Francis, and open your mind.

Given discs have a finite storage ability, available video bitrate is related to audio bitrate.

You have pure, native 1680x833 TV resolution and you're able to feed it such a signal from a DVD? BS. Look at the actual specs of your equipment. You're playback device probably has scaling/noise reduction/bandwidth squishing/image enhancement of some sort feeding a modified signal to your TV where it is rescaled/cleaned/interpolated for display.

Use a good encoder, not CCE. Use ProCoder or HC. CCE has a horrible tendency to create blocks.

No wonder you see crap on your large TV, you're making blocky encodes and rescaling multiple times.

For that matter, judging encoder quality based on a TV display is...how shall I say this...less than intelligent. Use a PC and proper editing software.

You are confusing a lot of issues which have nothing to do with bitrate.

[dvdbackup]

Yep, the signal is processed, but is processed with certain chips especially developped for this kind of image processing. You can't even switch it off !!

Looks reaaly bad, huh ???

But it isn't.....

How can you explain that original encoded DVD's CBR at 8000-9000 only gives perfect results ?????

The picture is clean, sharp, and has verry little GIBBS atrifacts, and no blocking at all.

If you say CCE is a bad encoder, there are a lot of people thinking the opposite.... Even in all tests I've seen CCE is on number 1.

HC is not an option at the moment. I want to be able to use reverse viewing without have picture disruptioon. HC causes this at the moment.
I don't gonna use encoders that generates streams with bugs.........


Finally, whatever you try, it's all aboute bitrate.

Nothing can replace a lack of bitrate......

Not even the best encoder can......

If this was true, all comercial DVD's should be just ok, but they are not.

I will try HC however on DVD-RW just to test it's results.

But it's no option to use it to burn the real backups with at this moment.


It's just like JPEG. The less space, the more imperfections. This is just a golden rule for every 'lossy' compression method.

When I see improvements however, I'll go for them.....

So far it has been a true mission impossible below 3000 average.

CCE BAD ???? Nope.
HC better ???? Nope, picture quality isn't a consideration when you get corrupted mpeg streams.........
ProCoder better ???: I will give it a try and see, not hoping for any magic at all.


It's all about bitrate. A high bitrate doen't mean you get high quality, but a low bitrate is a promise for lack of quality, whatever way you wanna see it...


And judging quality on the TV is not intelligent ??????

Judging on a PC is ?????

Yep, sure. When I want to look at a DVD movie, I'm going to setup my pc-configuration in the middle of my living-room, let everyone take a seat, and watch it on a 19 Inch LCD monitor...............

Nope, I use my TV for this...... Sure it's not intelligent in youre opinion to do so.... Let me be one of those Billions of people who uses their TV-set to watch DVD's.............

[FredThompson]

You're in the classic position of not knowing what you don't know and being too convinced of yourself to learn.

[dvdbackup]

Probably, as you say so, Master of the Universe.

Let people read and decide for themselves. A lot of people consider 3000 Kbps a certain rule of thumb....

If you give me ANY way, to backup LORD OF THE RINGS, THE TWO TOWERS, with good quality results on just a 4.35 GB dvd just keeping [5.1] audio track, be my guest......

I will even truly TRY it...........

Can't promise I will be satisfied with the result however........

If the result is objectively better than what I already have, I will truly thank you for it in public if you want......

My basic driver is: High Quality backups.

My ego is not that important.

But sure I'm one of those that first wanna have seen it him/her self before making any decisions. Words won't do the job. Seeing it however, just does do the job.......

[jdobbs]

IMHO... magic threshold numbers of any type never work. The demands of the source stream determines the bitrate required... and the eye of the viewer determines what is acceptable.

You can't use absolutes when dealing with what ultimately becomes a subjective conclusion.

[dvdbackup]

Again you are right.......

It's all subjective and absolute numbers will not be appropriate for all. As I said, it's kind of a rule of thumb.

However, pressing pause in the middle of an action scene, you can get two things:

1) Seeing nice square shaped blocks in it that are artificial and have no relation to the scene played.
2) Don't see the nice square shaped blocks at all.

If 1 happens, I reject it as a 'good quality' backup.

Of course I presume a good quality player and a good quality TV-set (with sufficient diagonal/viewing distance ratio) and even a good quality source.....

[feedback]

I'm trying to make a backup from Saving private ryan. But the overall bitrate is a bit low:

-- Processed 242.791 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 58,2%
- Overall Bitrate : 2.882Kbs
- Space for Video : 3.417.088KB
- HIGH/LOW/TYPICAL Bitrates: 3.770/1.168/2.882 Kbs

I allready did a test encode with the standard dvdrb settings and the segment looked a bit distorted around persons. I've read about filters in a similar topic but there's a lot of grain in the source. So i was thinking maybe a different matrix would give better results. Does anyone have some advice on this one?

Thanks :)

Try the BDVD Matrix or the AVAMAT6 Matrix they may give you a better result.
Also, you may want to get the Rebuilder Matrix Editor Here (http://dvd-rb.dvd2go.org/modules.php?name=Downloads&d_op=getit&lid=103) to test short segments with a variety of Matrices, such as the QLB Matrix.

Anyway, I used the BDVD Matrix on my 'Band of Brothers' DVD set with the HC encoder and actually prefered the backup copy to my original ( it got rid of the grain effect Steven Spielberg seems to love...I don't like the grain effect).

I have about 40 Matrices that I like to experiment with...some are mpeg2 compliant and some are not.

BTW the guys like Soulhunter and Sharktooth over in the Xvid forum are always coming up with different Matrices to try. Many are not mpeg2 compliant but some are.


Regards,:)

[FredThompson]

Probably, as you say so, Master of the Universe.

Let people read and decide for themselves. A lot of people consider 3000 Kbps a certain rule of thumb....

If you give me ANY way, to backup LORD OF THE RINGS, THE TWO TOWERS, with good quality results on just a 4.35 GB dvd just keeping [5.1] audio track, be my guest.......Drop your attitude and read what I've told you. Your system isn't scaling well at all. Additionally, your "testing method" is highly flawed.

You're pumping modified data through a modifying process then making judgements about the source data based on those multiple transformations. Again, you are confusing many issues. What you describe is most likely a poor MPEG2 decoding/scaling problem combined with multiple resizings. Your comments attribute to the source material corruption from your playback method.

I said I ran Tora! Tora! Tora! through HC to yield a nearly-identical 1/2 size replica. I never said 1/4 reduction of a more visually complex source would work as well. Don't jumble a bunch of issues together.

[Boulder]

SPR is a real Pain in the A**. It's very good testing material though, the ultra-shaky camera causes problems for any encoder. Motion-compensated denoising might be useful, Fizick's DePan plugin can be used to do that.

As has been said, the avg bitrate means very little unless you know how complex the movie is. My record is encoding "2010" at considerably less than 2000kbps and it looked just like the original. The movie just compresses incredibly well.

I nearly always do a sample encode with CCE. I create a script that I would use for the final encode, then add SelectRangeEvery(300,15) at the end to create a 5% sample clip. Then I encode it at Q30 with CCE with settings that I would use for the final encode. From that encode I get the lowest average bitrate I will tolerate. The nice thing is that it works for every movie:)

[dvdbackup]

Drop your attitude and read what I've told you.

You're pumping modified data through a modifying process then making judgements about the source data based on those multiple transformations. Again, you are confusing many issues. What you describe is most likely a poor MPEG2 decoding/scaling problem combined with multiple resizings. Your comments attribute to the source material corruption from your playback method.

I said I ran Tora! Tora! Tora! through HC to yield a nearly-identical 1/2 size replica. I never said 1/4 reduction of a more visually complex source would work as well. Don't jumble a bunch of issues together.

Just waiting for you to tell me the best way to backup LORD OF THE RINGS, THE TWO TOWERS, that's all.........

Probably you know a lot, say a lot, but just can't answer a straight easy question..........

Proves a lot, doesn't it ??!!!! :D

[Boulder]

There is no best and if you play by the rules, you don't ask for it.

Try my method and see what the lowest acceptable average bitrate is for the encode.

[dvdbackup]

I nearly always do a sample encode with CCE. I create a script that I would use for the final encode, then add SelectRangeEvery(300,15) at the end to create a 5% sample clip. Then I encode it at Q30 with CCE with settings that I would use for the final encode. From that encode I get the lowest average bitrate I will tolerate. The nice thing is that it works for every movie:)


Q30 seems pretty high, condering from what I've seen with bitrate-viewer.

But more importantly, how do you set the Q-factor in a DVD-RB environment ?

And as far as I know, the Q-factor isn't a constant.....

[Boulder]

What I meant is that if DVD-RB would give you an average bitrate that is lower than the one you get with Q30 in CCE (or whatever the lowest point for your eyes is), you might need some filtering etc. I would try DePan for motion compensated denoising.

The same Q gives different average bitrates for different sources. The ones that compress well (such as 2010) give a low avg bitrate. SPR will give you a high one because it's a very difficult source for the encoder.

[dvdbackup]

There is no best and if you play by the rules, you don't ask for it.

Try my method and see what the lowest acceptable average bitrate is for the encode.

To be honest, I thought you were playing with the Q-factor, which I can't and don't understand (yes I know is a value saying how much compression is done) . How should I set a constant Q-factor ???????

All the graphics I've seen, have a fluctuating Q-factor, well below 30.

So to be honest, I don't completely understand what you are doing.

I think finally you have to use DVD-RB and can't set a Q-factor.


I tried a lot with LORD OF THE RINGS movies, all about 3 hour and average bitrates around 3000. I didn't find any way to get a good quality backup. Finally splitting it up into 2 discs was the best solution......

I just took VOB's of the compressed disc and used bitrate-viewer.

VOB Bitrates (P/A) Q-level (P/A) P-PEAK A=AVG.
====================================================
VTS_01_1 6283/2787 9.56/6.17
VTS_01_2 6026/2794 8.86/6.02
VTS_01_3 7783/3148 8.65/4.89
VTS_01_4 7687/3529 8.27/5.30
VTS_01_5 5852/2774 7.77/3.68

And believe me, if anyone has any improvement tips, I will all try them.

Attached is a BMP that shows what I mean with blocking.....
(It's another LORD OF THE RING movie, but perfectly similar to the values above)...

This BMP is just a 1:1 frame capture. It's just a piece of the total frame. It's unprocessed. Enlarge it to 400% and you can clearly see the blocking....
( Nope, the blocks arent 2*2 at 400%, a lot bigger !!)

[Boulder]

Q is not the same as q which means quantizer. Q is the value CCE uses for determining constant quality level in OPV encoding. Lower Q means better quality (and higher average bitrate).

In extreme cases, I would personally edit the Avisynth script file so that the encode would result in 704x576 with 2 blocks overscan. FitCD can do the math for you.

[dvdbackup]

Run the math, don't perpetuate myths. 2 hours with AC3 audio on a single-layer disc is less than 3000 for the video. I just did Tora! Tora! Tora! without filtering using HC's highest quality mode. It's virtually indistinguishable from the original and that movie is 3 1/2 hours. You must be using an incredibly poor encoding method.


Yep, you know it all......

If it's all depending on the source the statement that a movie of 2 hours with AC3 audio (of what kind 1 channel or 5.1 chanells ???) is less then 3000 just can't be made.............

2 hours displaying "THIS IS A MOVIE" as white on black text even has 500 Kbps or less!!! (without audio).

2 hours high aktion, high motion, high detail has averages >4000 Kbps....

Ánd having said this I would have to take youre other reasoning for granted ???

Such as CCE is a bad encoder........

Well I'm on planet Earth, don't know were you are......... :D

[m1ckran]

I've followed this thread because, from its title, I thought I might learn something about matrices. I've not really learned much but I certainly find this thread amazing and a little disappointing.

I really don't think there is any need for sarcasm and unpleasantness in a forum dedicated to helping others.

I must admit that if my television made most dvd's look poor, I would wonder what was the matter with my TV. I would have thought that a set that uses heavy data manipulation to present an image must be suspected of causing distortions of its own. Any TV set can develop a fault.

On the other hand, if the tv and dvd player are perfect, it could be that the viewer's eyes are particularly good at finding faults in original dvd's that are missed by people with average eyesight. If the viewer considers an original dvd to be of poor quality, how could he possibly expect any encoder or transcoder to improve on the original - especially when implementing compression techniques. A lost cause, I think.

I wonder how broadcast TV appears?

I used to know a bloke who was a hi-fi fanatic. He spent hours listening to his system, trying to find sonic deficiencies and thinking of ways to upgrade his already superb and incredibly expensive system. He listened to his equipment and the sounds that came out of it. But he rarely listened to the music! I felt sorry for him because he seemed to have forgotten why he bought a hi-fi in the first place. There must be a moral (and pehaps a parallel) there.

m1ckran

[johnhamler1]

I have encoded 50-60 DVD with HC, my last one was 21grams PAL version (2003?) , with a grainy picture and a contrast/color which it not usual .
on my beamer, it looks like the original, 3000-4000 bits/s encoded.

the few artifacts come from the conversion of numerique to analog(mainly ghost on straingt lines), analog back to numerique and back to analogue.
so i think , before to judge a movie from his point of view, we should look at our own hardware first.

HC did an excellent job even for low bitsetting, just slow but it is free.
except my dvd-+R are not working on some old dvd players, QLB matrix by exemple is not working on some player!

conclusion: HC, is very good!