Removing of 500 minimum bitrates improves results

[apfraats]

I have done some experimenting, as usual, and I used RB-OPT to set the minimum bitrate to 0 instead of 500 and the maximum flat to 8000.

The original movie was of PERFECT quality and I observed two things:

1) Bitrates have higher peak-values than DVD-RB would allow them to be. The encoded material has somewhat higher peaks than the original. This was observed with perfect wuality high action source material "SAHARA", so it would make sense to remove bitrate limitation based on original. You can better allocate MAX_BITRATE to theoritical maximums as calculated by subtracting all stuff from the 10.05 Mbps limit. leaving enough room for error.

Because every encoder has it own charasteristics, you can't simply state that the highest encoded bitrate should always be equal or less than the original video bitrate. If you use filters, you can even change sharpness and cause higher bitrate to occur for that reason only.

This should be an option in DVD-RB-PRO, not a rule. It simply is not true to make max_bitrate re-encoded =< max_bitarate original.


2) The minimum bitrate of 500 dhould also be removed or being an option to set as user in DVD-RB-PPO.
Studying the bitrate graphics from bitrate viewer clearly stated with min_bitrate=0:

a) Lower bitrates in the lower bitrate parts, in regard to using 500 as minimum. So using the 500 minimum wastes bitrate in less demanding parts.

b) Higher peaks and steeper bitrate changes at parts were high peak bitrates are needed to cover high movement high action scenes. So using to 500 minimum causes demanding scenes to have less bitrate and earlier blocking of that particular scene.

So it's bettter to:

a) Not use 500 minimum, but 0 as minimum or optional setting.
b) Not use or make optional setting for restricting maximum bitrate to original max_bitrate found for a segment.

The best should be:

a) Make a variable min_bitrate option in DVD-RB-PRO (range restricted if you wish)

b) Make a enable/disable setting for restricting max_bitrate to original max_bitrate. If disabled it should be set mathematical as 10.05-stuff needed - room for error. Of course on a per segment base as this is the core way of handling stuff in DVD-RB-PRO.


Now I have to use the 'tweak tool' RB-OPT to get desired results.

It should be done from within DVD-RB-PRO as it improves quality.

[raquete]

a) Not use 500 minimum, but 0 as minimum or optional setting.
some players don't like low bitrates and give heavy issues!!!!

[apfraats]

I know, that why I stated 'optional' and not FIXED !!

IIRC JDOBBS included the 500 minimum because of CCE problems, but my CCE has no problems at all with it....

Also there is someting like a minimum_bitrate required for the demultiplexer in official DVD-specs, but this is related to the existence of at least 1 audio track (I don't remember the details exactly) besides video in the stream.

However, if you look at some orininal DVD's with total BLACK parts, the bitrate goed down to about 0.08 or something Kkps according to POWER-DVD and that's video bitrate only.... If this is true for originals, I don't see a problem here....

Also as it improves the quality of the 'difficult stuff' it should be at least an option.....

Using other 'tweaking' tools because of the fact the original program doens't perform optimal even in SOME cases, is always the lesser way......

So it might be considered by JDOBBS.........

He knows best...

[raquete]

optional? (all my players can use min 300) take a look:
http://forum.doom9.org/showthread.php?p=748993#post748993
adjusting like this works? read the whole thread :)

[apfraats]

So what ?????

I never have problems with low bitrates.........

The problems are when there is demanding stuff and I get macroblocking....

Just making it POSSIBLE to set the min_bitrate to 0 if you wish would improves all my cases....

MIN_BITRATE as parameter in the INI has no effect if it is lower than 500...

DVD-RB-PRO overwrites it with it's default value 500 in that case.

I tweaked with RB-OPT to 0, and it gives better results......

It's not a big job for JDOBBS to make that 500 bottom setting a variable one.

Then everybody is happy, even not having to edit the INI files manually when they need a higher MIN_BITRATE and is capable of using only DVD-RB-PRO if ones wants to run the minimum at 0.

Don't see any problems there, it's a usefull improvement in functionality....

If you can set things such as GOP-SIZE, VBR_BIAS e.d. for CCE , it sure isn't asked too much to also be able to set MIN_BITRATE to 0 or any (logical !) value you wish......

It sure has significant impact on the result.....

Especially using borderline compression and stuff.

[urvieh]

@apfraats: ... higher peaks than the original one ...

What is it good for?

[apfraats]

Ask the people that build cce........ Maybe they can explain it to levels of assembly language.....

In fact, I just test, evaluate and observe....

I noticed even with GOOD QUALITY sources, the bitrate of CCE can be higher than original. I suppose it's better to allow CCE this room as it is there.

I don't see low-bitrate scenes duffer, only problems with high bitrate scenes.

This is caused by the minimum 500 bitrate by which the encoders is forced to use 'other calculations' which tend to spend bitrate on less demanding scenes.

Removing this 500 minimum and setting it to 0, gives more room for the high motion, high action scenes.

Some people complain about CCE's blocking..... no wonder....

The other encoders don't even allow a MIN_BITRATE setting as far as I know (maybe one of them, i'm not 1005 sure).

To give you an example:

You use an AVI filter that enlarges contrast. This contrast is also enlarged in High Motion High Action scenes.

Resulting in higher bitrate needs for CCE in comparission to the original.

Another reason can be the use of matrices that use more detail bij lowered treshold values for diferences.....

Then finally, each encoder has own charasteristics. This can cause fluctuation in bitrate peaks even over original....

If you gonna limit these peaks, the encoders cannot do it's job in the most optimal way.

To make it black/white: Use 4000 as max_birate and do a movie with 3600 average and high motion high aktion scenes. And count the blocks......

So why LIMIT bitrate ???

It can be usefull for bad sources where there is already blocking causing high contrasted lines that have to be encoded again, stealing bitrate that's useless trying to properly recreated the already disorted picture.

But with propper sources you keep the encoder from doing it's work.

Given the same source, using different encoders, gives compleet 'different' graphs in bitrate viewer.... This is the evidence....

That's why....

So a user setting KEEP MAX BITRATE EQUAL TO SOURCE (Y/N) or something like that would be very usefull.

One can choose. Of course MAX_BITRATE must NEVER be out of specs, but that's a general rule......

The movie SAHARA I did came out wonderfully and much much better with MIN_BITRATE=0 and MAX_BITRATE=8000 which was significantly higher than source limits.

But I needed RB-OPT to do so, and it would be nicer to see such basic settings in DVD-RB itself......

[blutach]

I still wonder why max BR should be higher than original ... IMHO, you can't get better than the original - what you do with filters (which might raise BR) is put your interpretation of how you want your encode to turn out. For me, I'll stick to how the studio wanted it.

Plus, raising the max BR can have unintended consequences (e.g. stutter - see Rebuilder.txt file).

Regards

[apfraats]

Aha, so you want to copy the studio ????

Well youre right if:

1) You use to exact same encoder the studio uses.......
2) You use the same matrix the studio uses......
3) You use all the same paramters for the encoder the studio uses.

And even then you have to deal with an extra DECODER step, that introduces artififacts that the original movie in the Sudio didn't have.

So youre wish is a bit wishfull thinking, making an exact copy of the materail the studio used, isn't simply possible. That's why it's called lossy compression.


However, it won't be a problem if it is a setting in advanced options.

You will probably like to set: Restrict to source limits....

I'll probably set it to: Restrict to maximum limits..... (10.05- bitrate needed for stuff audio/subtitling) - room for error.

Getting rit of the 500 minmum already gives room for extra bitrate for demanding parts, why limit the possibilities the encoder offers ????


So DVD-RB-PRO can easily keep both us happy........

There is a program RB-OPT, that should be needed NOT to be there.....

I rather like only one author and 1 programm to prevent incompatabilities and such...

Would be nice if RB-OPT functionality is put into DVD-RB-PRO, so there is no need for it anymore.

And raising bitrate CANNOT give you problems as long as it below the MAXIMUM bitrate thst's possible for a certain video stream. You cannot go over a total bitrate of 10.05 Mbps. As long as you keep that in mind, there is no problem. Even an EXTRA reason to implement it in DVD-RB-PRO, because that's the program able to do MAX_BITRATE calculations. If you use RB-OPT you can make basic mistakes by setting a too high value....

Just an extra reason to make it an advanced setting in DVD-RB-PRO !!!

[Video Dude]

You don't need RB-OPT to set the min bitrate. In DVD-RB's ini file in the [OPTIONS] section just put the line min_bitrate=0.

[blutach]

@Video Dude

I read somewhere that this setting may not work with HC as it doesn't accept that argument being parsed. Not sure.

And, apfraats, I think RBOpt is a great little addon - indeed it is very useful for tweaking your BR. Found it perfectly compatible with dvdrb.

Regards

[Video Dude]

He posted he is using CCE, thats why I suggested using the min_bitrate in the ini.

[jdobbs]

@Video Dude

I read somewhere that this setting may not work with HC as it doesn't accept that argument being parsed. Not sure.

And, apfraats, I think RBOpt is a great little addon - indeed it is very useful for tweaking your BR. Found it perfectly compatible with dvdrb.

Regards Actually I don't think HC has a minimum bitrate setting. My assumption is that Hank315 has decided to control that within the encoder.

[apfraats]

The most important fact is lowering the MIN_BITRATE to 0.

As seen on copmparissions this gives better results, becuase it not only influences the minmum birtate but the WHOLE BITRATE DISTRIBUTION in every parts of the video stream.

1) You get lower birtrates on lower parts. So if average = 4200 Kbps/sec and minimum is 500, at exactlty the same spots in the movie you'll get things like:

3500 3800 3900

2) If you do the same with a 0 minmum bitrate you get:

3400, 3650, 3825 for example.

So the lower bitare parts are lowered in bitrate more.
This saved bitarte will be shifted towards the often MUCH SHORTER high demnading high action scenes that graetly improve by this.

So saving 100 Kbps for 10 minutes is 1000 extra for 1 minute action....

But if the aktion last just 10 seconds it is the difference between HEAVY BLOCKING and NO BLOCKING AT ALL !

Then that 100 kbps saved during 10 minutes is an extra 6000 Kbps for the 10 seconds action, that's what I stated. You'll see this is a SIGNIFICANT difference.... and this is just an example, but as far as I know, there are no movies containing full action from begin to end....
This such an approach will yield better results...

@VIDEO_DUDE & JDOBBS: Are you sure the MIN_BITRATE=0 in the INI WILL put the minimum to 0 ???
I thought you could only lower the MIN_BITRATE to values above 500, just like you cannot use MAX_BITRATE to go over the value DVD_RB has determined ???

Or are both true ??

Just have to try, I suppose.

It would make live a lot easier if the settings could be done in the CONTROLLED ENVIRONMENT of DVD-RB, not editing INI files and not using BETA TOOLS.....

The tools are there because there is probably NEED for them....

I wish to see it integrated in DVD-RB-PRO.....

That's a lot better because DVD-RB-PRO can apply the contraints against wrong settings, that RB-OPT cannot.

Of course in the advanced CCE settings, also build in filters and their parameters should be able to set.....

As even matrices are selectable, I don't see why more basic settings of the encoder used, like CCE's filters and parameters are NOT implemented in DVD-RB-PRO.......

It would enchange the great DVD-RB-PRO even more....

The 'SKIP PREPARE IN BATCH' is a nice way of 'let i't be' but not the most user-freindly setting.

If for some reason JDOBBS decides to use other intermediate file structures the RB-OPT tool won't work any longer.......

So it's better to implement this RB-OPT functionality in DVD-RB-PRO.

[jdobbs]

@VIDEO_DUDE & JDOBBS: Are you sure the MIN_BITRATE=0 in the INI WILL put the minimum to 0 ???
I thought you could only lower the MIN_BITRATE to values above 500, just like you cannot use MAX_BITRATE to go over the value DVD_RB has determined ???If you set the minimum bitrate to zero, that's what will be stored in the ECL file. 500 is just the default value.

[jdobbs]

By the way... I'm not absolutely positive, but I don't think there is a minimum bitrate that you need to set for encoding for DVD... there might be, though, a resultant minimum caused by padding that might be done for buffering, multiplexing, etc.

The default of 500 was set because of problems that appeared in early versions of CCE 2.70... it was crashing on some sources with lower values.

[apfraats]

@JDOBBS: Nice you bring this up.

I have a trial version of 2.7.2.X or something (the latest I suppose), but I cannot get it to work. I set DVD-RB-PRO to TRIAL, and use NO ECLCCE, but if I run DVD-RB-PRO CCE 2.7<latest> is activated and all segments get skipped very fast (would be nice to have that speed for real :D )

What is the BASIC setup for a 2.7<latest> 'trial' ?

ELCCE is not used, I have seen on the forum, but simply setting CCE TRIAL won't work either ??

What do I have missed ?

(This is because there is a possibilitie of blocking in fast fade-in/fade-out scenes, as this is given as a 'special 2.7' improvement).

I saw a lot of changes in the readme.txt, but I don't know if they affect DVD-RB-PRO or the RB-OPT settings and paramter passing.

However RB-OPT seems to wrongly report PRECISION as 16 while it is set to 32 in DVD-RB-PRO ??

Is all parametric control correctly supported using 2.7<latest> ???

Thanks..

[jptheripper]

they are all correct. cce changes scaling on every version it seems.

[apfraats]

I know, but RB-OPT uses a scaling from 0-64 and DVD-RB does also use it.

A setting of 32 in DVD-RB results in a setting of 16 displayed in RB-OPT.

I know CCE is really a messy tools concerning it's parameters, it changes with every version as it seems.

I know use MIN_BITRATE=0 which indeed works as JDOBBS already stated and I'm really glad it works , because it grealty improves CCE's behaviour on stuff that goes BLOCKING at critical scenes. I never expected it to have SO MUCH impact. But it seems the overal bitrate distribution is changed by lowering MIN_BITRATE.

Some people use excessive settings for number of passes. But lowering MIN_BITRATE to 0, always gives far better results at 3 passes total inclusing the VAF pass.

@JDOBBS: That minimum bitrate that is required on DVD doens't worry me too much, because you always have 1 or more audio tracks. So always there is a minimum>0 bitrate for the demultiplexer, so it can do it's job. I think that minimum bitrate contraint also was 'catched' by the official DVD spec, to have at least one audio track of some kind besides VIDEO. So a video-only track is NOT complying to official specs. This garantuees a minimum bitrate above 0. I can imagine that a total bitrate of 0 or very very low, can get the demultiplexer in trouble. I don't know if there is always some information for the demultiplexer to proces. But theoretically a rate of 0 bits if hard to demultiplex.

[rendez2k]

Can anyone tell me if theres any drawbacks to using MIN_BITRATE=0? I tried it on a disc in the Doctor Who box set (which has been really difficult to back-up well) and the result was a lot better - much less blocking! So, is it OK to use this on every disc?

[jdobbs]

I don't think there is any disadvantage. As I said, the 500 is only there to counteract a problem with one version of CCE. The former default was 100 -- which I think is probably a good number to use.

I'm modifying the next version of DVD-RB so the min_bitrate will again default to 100 -- I'm then adding code that will automatically raise it to 500 if a segment fails during encoding with CCE.

[dragongodz]

as i remember it was mentioned long ago that when the minimum video bitrate was below about 300(from memory) there was some playback problems on a couple of dvd players aswell, and yes this also had an audio track. ok 99.9% of players will play fine with a min bitrate of say 0 and if you are in that 99.9% then you are happy. however when something bites you on the backside that 99.9% of others dont experience please remember this post. :)

I think that minimum bitrate contraint also was 'catched' by the official DVD spec, to have at least one audio track of some kind besides VIDEO. So a video-only track is NOT complying to official specs.
hmm and where do you draw these incredible conclusions from exactly ? ok not many people on here would have the actual specs but i have never ever seen it said ANYWHER there MUST be an audio track and i would lay odds it doesnt say it in the official specs either. so ye i would be interested to know exactly where you have read this or what testing you have done to come to these conclusions.

[jdobbs]

@dragongodz

Does QuEnc have a minimum bitrate value it uses during encoding (or is that something the library does)?

Maybe I'll change the default to 300 just to be safe. People can change it if they want.

[influenza]

I always use min bitrate 0 on my encodes (manual encodes, no DVD-RB). Never had any problems. Not that I ever check, but I don't think there are actually a lot of parts that will get encoded with such a low bitrate.

[dragongodz]

Does QuEnc have a minimum bitrate value it uses during encoding (or is that something the library does)?
when it was strictly avcodecs rate control no because setting a min bitrate totally messed RC up. ye avcodecs great RC. ;)
however Nic was playing with it when Xvid RC was added. i cant remember offhand how he left it but i do recall him saying something about having a small min bitrate hardcoded in to it. will have to ask him if he remembers. :)

I always use min bitrate 0 on my encodes (manual encodes, no DVD-RB). Never had any problems. Not that I ever check, but I don't think there are actually a lot of parts that will get encoded with such a low bitrate.
yep as i said it was only on a couple of players(models) so not something that 99.9% of people will probably ever see and yes it wouldnt happen all the way through an average movie for sure. the same can be said of other problems that have cropped up though where they only effect specific players. doesnt mean they should be ignored though just because its a rare thing. having a safe default with the ability to change it for those knowing what they are doing and willing to accept responsability for any problems caused by fiddling seems a good idea to me. just my humble opinion anyway.

a post where LigH mentions it aswell
http://forum.doom9.org/showthread.php?p=676858#post676858

without looking i also seem to remember hank315 hardcoded a small min bitrate in to HC for dvd resolution for the same reason.
ok just spent half an hour finding this
http://forum.doom9.org/showthread.php?p=604634#post604634

oh and jdobbs, that first part of my previous post wasnt aimed at you. looking back at it it may appear so because its right after your post. ;)

[dragongodz]

ok Nic said he cant remember how he left it either. he would have to wade through the source again to have any idea.

Nic and i have to be the worst for remembering all the changes we have made in the past. :D