Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > (HD) DVD, Blu-ray & (S)VCD > DVD Rebuilder

Reply
 
Thread Tools Search this Thread Display Modes
Old 18th October 2005, 15:56   #1  |  Link
cali310
Registered User
 
Join Date: May 2004
Posts: 2
CBR movie ???

Is there any point in using DVD-RB as opposed to CloneDVD, DVDShrink etc to backup a CBR DVD? Overnight I backed up a Bubblegum Crisis DVD and noticed the high/low/avg bitrates were all the same.
cali310 is offline   Reply With Quote
Old 18th October 2005, 16:20   #2  |  Link
robot1
Registered User
 
Join Date: Apr 2003
Posts: 651
DVD-RB uses VBR for every segment.
You will have all segment with the same average bitrate, but you will have a VBR encode.

Anyway, for a CBR movie, I'd use OPV in DVD-RB.
robot1 is offline   Reply With Quote
Old 18th October 2005, 16:52   #3  |  Link
jdobbs
Moderator
 
Join Date: Oct 2001
Posts: 19,802
@cali310

You have to understand that the "High/Low/Average" you see are talking about average bitrates for segments -- not individual bitrate settings. With DVD-RB and an encoder you will get a truly 2 pass VBR result... with transcoders you can only get a requantized modification to the original stream -- so they would result in CBR.
__________________
Help with development of new apps: Donations.
Website: www.jdobbs.net
jdobbs is offline   Reply With Quote
Old 22nd October 2005, 18:25   #4  |  Link
dvdbackup
Guest
 
Posts: n/a
True.......

I have had dirfferent CBR originals, 7000-8500 CBR in most cases.

I use CCE. If you set VBR_BIAS to a value low as 0, you get true VBR, any value higher goes more towards CBR. Using CCE with CBR source and CCE (SP) and VBR_BIAS=100 will result in a true CBR results. You should not be happy with this however when using any significant compression ratio.

It's more usefull to let the encoder distributes the bitrate compared to the needs of the specific scene.

So using VBR_BIAS of 0 or something higher as 25 (what is the standard setting if I rember it correctly) let's the encoder 'decide' what kind of bitrate is needed for a certain scene.

If you have a static scene with less detail and less motion , you automatically get low-bitrates. On scenes with higher demands regarding detail and motion, the encoder will automatically use higher bitrates.

This is true to a certain compression level however. If the compression level is so high that regarding the original bitrate you get averages bitrates of 2500 or lower, you get problems such as blocking in fast action, high motion scenes. You'll see the bitrates going above average in these scenes, but they cannot reach a high enough value to generate a picture without blocking in high demanding scenes. This happens even when using the very-low-bitrate matrix and full VBR (Bias set to 0).

For example: the 3 hours lasting Lord of the Rings movies.

They have 3 hours of video with 5.1 sound.

It's has shown to me that it's impossible to get a good quality compressed movie out of this. You'll always see 'blocking' ocurring in the high action fast movement scenes.

Some peolple are happy wth the result, but I'm not.

So in these cases there is just one solution: Split the movie to two disks.

CBR in source at high levels have some advantages:

1) There are no or little GIBBS-artifacts in the non-aktion scenes.
2) They bitrate is always high enough to perfectly display high action scenes.


When going to VBR you can get hold of point 2, but you'll get more GIBSS-artifacts in the result as of the lower bitrates used.

Gibbs artifiacts are noticed as a kind of blurry noise around high contrast objects which have sharp edges in a movie. Even in static scenes.

To prevent this from happening and get the most optimal viewing result, some DVD's use high bitrate CBR.

There is space enough, so why not ????

On a VBR source you have already a bitrate distribution. DVD-REBUILDER will try to improve this by determining the average bitrate on a cell by cell basis.

But this is NOT grantueed to be the best solution. Why not ? See example bellow:

--low action--------|+++++High action++++++++++| ----- low action ----

--------------cell 1--------------|---------------cell 2-----------------|

DVD-RB will determine average bitrates for cells 1 and 2.

But the action scenes are at the end of cell1 and the beginning of cell 2.

When this happens, the method that DVD-RB uses fails.....

This is because the bitrate is based on a cell by cell basis.

Not on some kind of automatic scene selection.

But doing it on a cell based average bitrate is in most cases always better than using 1 averga bitrate for the whole DVD.
  Reply With Quote
Old 22nd October 2005, 18:41   #5  |  Link
jdobbs
Moderator
 
Join Date: Oct 2001
Posts: 19,802
Quote:
When this happens, the method that DVD-RB uses fails.....

This is because the bitrate is based on a cell by cell basis.
You're wrong.

It has been proven over-and-over and is based upon mathmatics and the fact that DVD-RB considers the original encoding as a basis for bitrate distribution... but it's your opinion, so let it be, no sense arguing the same points over and again. Try "search" next time.
__________________
Help with development of new apps: Donations.
Website: www.jdobbs.net
jdobbs is offline   Reply With Quote
Old 22nd October 2005, 23:22   #6  |  Link
dvdbackup
Guest
 
Posts: n/a
That doens't explain why cell based distribution should be better than distribution based upon scene detection....

In fact for most movies is cell is no more than the video between two chapter points...... Why should this be the best solution ? Sure it's more easy to implement....

But if I would reauthor the movie and place chapter points depending on where the action scenes starts and ends, DVD-RB would give a different cell-assignment and probably a better one.....

So it would by better to prescan the source and evaulate it in the way bitrate changes in the source material are found...... Of course that's a much more complex proces, but a better one.

Now it's 'just' cell based, totally unrelated to the general MPEG-2 stream behaviour.

However, it's most of the time if not always better to have some way of splitting the Mpeg stream into parts and doing processing based on these parts than doing it for the whole at once.

The cell based assigment is probably choosen because it's the easiest way to implement futher splitting up the total MPGEG-2 stream, not because it's the best way.
  Reply With Quote
Old 22nd October 2005, 23:36   #7  |  Link
jptheripper
Registered User
 
Join Date: Aug 2003
Posts: 913
wrong wrong wrong

only the average bitrate is set on the cell, but within the cell the vbr is done by scene detection in the encoder. Hence high action will still recieve high bitrate

it has been shown many times that the difference in bitrate done this way versus using the whole movie at once is insignificant.
jptheripper is offline   Reply With Quote
Old 23rd October 2005, 00:51   #8  |  Link
dvdbackup
Guest
 
Posts: n/a
Quote:
Originally Posted by jptheripper
wrong wrong wrong

only the average bitrate is set on the cell, but within the cell the vbr is done by scene detection in the encoder. Hence high action will still recieve high bitrate

it has been shown many times that the difference in bitrate done this way versus using the whole movie at once is insignificant.

Nope Nope Nope, the encoder uses VBR but the DEFAULT average is set cell by cell. So if you have a cell with 90% average of 2000 and 10% avergae 6000 in one cell, the avergae would be 9*2000+1*6000/10= 2400 average.

Having this value as a target value, the encoder will do encoding in VBR but it will be totally at the expense of the 6000 part, which will degrade in quality and probably have blocks when played back......

Now make it a distribution of 95% to 5% and it's even worse. The encoder has to reach the 6000 from 2400 average, which won't be possible in this way.

I have seen this happening in the LORD OF THE RINGS movies.
The bitrate is enough for the more static scenes, but too little for high action/high movement scenes and these have a blocky appearance as a result. That's because the encoder has not enough 'headroom' to encode them at the lowest possible bitrate without going the get the blocky picture.

I would like to steal some bits form the more static scenes and apply them to the scenes that get blocky. As the movie is 95% unblocked and just 5% of the movie has blocky pictures this should be possible. But with the current way DVD-RB assigns segments and average bitrates it's impossible to get this done. Yep manually change bitrate assignments.............


That's why cell based distribution is in essence not correct. It should be divided into bitrate-alike segments in stead of cell-based segments....

If you can set a kind of treshhold and and other parameter values to prevent the formation of too many segments, this would be an ideal solution.......

The spitting is then really 'lot alike same bitrate' based and not cell based.

In this case the following would happen:

----------- cell 1 -------------|------------ cell 2 -----------------------|
--------- low action--------|+++++++ high action +++++|----- low action-

--- SEGMENT 1 AVERAGE X | ------- SEGMENT 2 AVR. Y | - SEGMENT 3--

and not:

-------- SEGMENT 1 AVR Z1----|-------- SEGMENT 2 AVR Z2-------------|

as now happens in DVD-RB.

Last edited by dvdbackup; 23rd October 2005 at 01:01.
  Reply With Quote
Old 23rd October 2005, 01:16   #9  |  Link
jdobbs
Moderator
 
Join Date: Oct 2001
Posts: 19,802
Nonsense.

If the original stream was encoded at a set average bitrate, that would take into consideration the distribution of available bits as required. That allocation is used when reallocating in DVD-RB. Even if that weren't the case you'd find that the number of frames existing in a typical cell are already high enough to statistically provide the variability needed for VBR.

You also don't seem to understand how VBR works... if the high demand part of the cell needed more bitrate -- it would be allocated as necessary to create some degree of consistent quality... it might even match the same bitrate level as the original if that is what is required... but will lower itself as necessary to reach some level of consistency.

If you were encoding the entire film all together you would get virtually exactly the same results because the first pass of the encoder would recognize the requirements of the scenes and allocate the same amount of bandwidth to that section of the film (just as the original encoder did). DVD-RB uses the original encode as a "first pass".

As I already said, this has all been discussed and settled before... let's end it.
__________________
Help with development of new apps: Donations.
Website: www.jdobbs.net

Last edited by jdobbs; 23rd October 2005 at 01:24.
jdobbs is offline   Reply With Quote
Old 23rd October 2005, 03:35   #10  |  Link
jptheripper
Registered User
 
Join Date: Aug 2003
Posts: 913
@dvdbackup.

this point has been brought up many times , and disproven every time. you havent come up with anything new, you are just refusing to listen to the correct explaination.

while your reasoning is passioned, it is off track.

it has been mathematically shown (use the search, way back when rebuilder began dynamic bitrate assignments) that the cell approach works extremely well with no precievably different result.

you example is right, however, you are assuming that
--- SEGMENT 1 AVERAGE X | ------- SEGMENT 2 AVR. Y
segment y will not have a significant bitrate to satisfy the need in high action

two flaws in that reasoning
1. the source will have had a higher bitrate on the segment y b/c it is high action
- therefore y is much greater than x as dynamically assigned by rb according to the source
2. average bitrate = quality
- this is incorrect. segment y might average at a particular bitrate, but if the scene needs a spike the encoder will supply it via VBR.

again, i doubt jdobbs will comment again. this has been discused ad nausem, and each time the current method used by dvd-rb is shown to have excellent results (i.e. it works perfectly).

if you feel strongly about this, please post encoded frames of a blockly frame and a "fixed" frame encoded by the big3 with same settings to show that it is dvd-rb methodology that is the difference
jptheripper is offline   Reply With Quote
Old 23rd October 2005, 05:32   #11  |  Link
jptheripper
Registered User
 
Join Date: Aug 2003
Posts: 913
oh and another poiint, a cell with 2400 average can easily have 6000, 7000, even 8000k spikes.

thats why its called vbr
jptheripper is offline   Reply With Quote
Old 23rd October 2005, 08:13   #12  |  Link
steptoe
Registered User
 
steptoe's Avatar
 
Join Date: Mar 2003
Location: UK
Posts: 360
I've always wondered about the VBR Bias setting for CCE, is it better to leave it at the default or simply set it to 0

Not that I can tell the difference on my TV .....

I currently use 20 for movie only, and 10 for episode DVD's


If I set it to 0 what will this effect compared to setting it to 20 or 15. Does the 0 setting allow the encoder to have more free choice over the bitrates as I also use 2-pass for movie only or 3-pass for episode disks
steptoe is offline   Reply With Quote
Old 23rd October 2005, 14:34   #13  |  Link
jdobbs
Moderator
 
Join Date: Oct 2001
Posts: 19,802
Lots of violent opinions about this, so I'm walking on shaky ground making a comment...

Setting it to 0 is probably too much. While it does push it more into the VBR direction... it also tends toward large fluctuations in bitrate. One of the side effects that I've seen noted in very low settings, for example, is more blocking during fade-from-black scenes.
__________________
Help with development of new apps: Donations.
Website: www.jdobbs.net
jdobbs is offline   Reply With Quote
Old 23rd October 2005, 15:20   #14  |  Link
Carpo
Registered User
 
Carpo's Avatar
 
Join Date: Dec 2002
Location: /dev/null
Posts: 1,365
i used to have mine at 0 but never saw an issues with it - mind u i never looked that close

i have put it back to default (25 - with cce sp 2.70 trial) and qual prec at 36 - is that ok or is that a bit much also ?
__________________
The Internet: where men are men, women are men, and children are FBI Agents
Carpo is offline   Reply With Quote
Old 23rd October 2005, 15:24   #15  |  Link
jdobbs
Moderator
 
Join Date: Oct 2001
Posts: 19,802
Default is 25/16. Those were selected because in testing they gave the best results most consistently results across the widest range of sources.

But, mileage may vary...
__________________
Help with development of new apps: Donations.
Website: www.jdobbs.net

Last edited by jdobbs; 23rd October 2005 at 15:27.
jdobbs is offline   Reply With Quote
Old 23rd October 2005, 15:26   #16  |  Link
Carpo
Registered User
 
Carpo's Avatar
 
Join Date: Dec 2002
Location: /dev/null
Posts: 1,365
i know i meant default for vbr

and i'll take that as a yes 36 is a bit much

would 25 be the default for HC aswell ?
__________________
The Internet: where men are men, women are men, and children are FBI Agents

Last edited by Carpo; 23rd October 2005 at 15:28.
Carpo is offline   Reply With Quote
Old 23rd October 2005, 16:14   #17  |  Link
SpazzHH
Registered User
 
SpazzHH's Avatar
 
Join Date: Feb 2005
Location: Cleveland, TN. USA
Posts: 429
Hank ships HC with a default Bias of 20.
__________________
I'll learn this stuff yet, and when I do, Look Out !!!
SpazzHH is offline   Reply With Quote
Old 23rd October 2005, 19:09   #18  |  Link
dvdbackup
Guest
 
Posts: n/a
Ok, let it go JDOBBS, When I have a general movie, and average is high enough (in my opinion >3500) the result of DVD-RB is unbeated in any way, comparing it to other ways of compressing. So no doubt about that one.

DVD-RB-PRO is uptill now the best tool I've ever seen.

I was only mentioning it, because of the fact I did the long movies LORD OF THE RINGS.

I used CCE with a VBR_BIAS of 0, to give it the most headroom for VBR.

But I notice that action, in chapters (=cells) that have low average bitrates, because the most of the chapter is easy going static scenes, is blocked.

Even doing 5 passes won't help.

I have the feeling that if I could steal some bitrate from the less demanding scenes the action scene would be better as having blocks is very disturbing to watch.

That was/is my problem.

You are right of the Bitrate distrubution, in the way the encoder should steal the needed bitrate automatically. But with low average bitrate I have concluded that a sudden action scene becomes in problems.

That's why I was having some thoughts of a way this could be handled in a better way.

But I agree let's stop with it. The main problem is the blocking in action scenes. I have already tried to very low bitrate matrix, as bitrate-viewer shows higher bitrates in sudden action scenes, using that matrix.

I now try the HC encoder. Some people say it's better in these paritcular cases to prevent blocking in the action scenes.

So, I was only giving this a thought because of the results I get, still leaving me with the feeling that if stealing some more bitrate of scenes that won't suffer from this too much and allocating it to the action scenes would deliver a better result, in case of the blocking of action scenes.

So I still have to try and do more experimenting with encoders and matrices.

By the way, I always have the blocking as the first problem on low bitrate material. So I set VBS_BIAS to 0. Some say this is to agrressive, but that would mean that action scenes wouldn't suffer but static scenes get in trouble.

I never have had that. It are always the action scenes becomming in trouble first. So a VBR_BIAS of 0 seems not to be too agressive, I would even say it's still not agressive enough........

If there was a way to set it to -10 I would even try that ome !

@others_asked: I don't know have to post an example of a blocked frame....
My fault. I'm just out of time and don't have it right now to try and find it out myself.

@others_asked: I personnally don't think a VBR_BIAS of 0 is a wrong setting. It's just FULL VBR. If it was too agressive the action scenes would have no blocking , but the static scenes would have. As this is not the case I still prefer VBR_BIAS=0. By the way it's only important with lower bitrates, as here it comes in focus. If you have 4000+ avergae stuff you will not notice the difference anyway. (25 or 0). But try 100 if you want to see something happen. Then the static scenes will be ok, but any demanding scene will strongly suffer. Opposed to 100 is 0. 25 will be more CBR like. So if with 0 the action scenes suffer and static wiil not, with 25 the even would more suffer and static not.

Last edited by dvdbackup; 23rd October 2005 at 19:13.
  Reply With Quote
Old 24th October 2005, 00:50   #19  |  Link
jptheripper
Registered User
 
Join Date: Aug 2003
Posts: 913
vbr bias = 0 is why you are getting blocky on low motion, it is too aggressively stealing bitrate from them
jptheripper is offline   Reply With Quote
Old 24th October 2005, 01:04   #20  |  Link
laserfan
Aging Video Hobbyist
 
Join Date: Dec 2004
Location: Off the Map
Posts: 2,443
Quote:
Originally Posted by dvdbackup
...@others_asked: I don't know have to post an example of a blocked frame.... My fault. I'm just out of time and don't have it right now...
And it's easy to see why! Just look at the length of your posts!!! Does anyone here actually READ them!!!???!!! Good grief!!
laserfan is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 14:43.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2017, vBulletin Solutions Inc.