BD-RB auto encoder quality

[xterminater]

Jdobbs, I know that this function auto select the quality output based on the time, and on the output size that is selected via BD-RB. Could you explain more on what this is about because when unchecked, you could select it to be good, better, high, and highest. When I ticked it for auto encoder quality, the log says better but does selecting high and highest really matters about the way x264 encodes? Thanks jdobbs.

[Groucho2004]

Jdobbs, I know that this function auto select the quality output based on the time, and on the output size that is selected via BD-RB. Could you explain more on what this is about because when unchecked, you could select it to be good, better, high, and highest. When I ticked it for auto encoder quality, the log says better but does selecting high and highest really matters about the way x264 encodes? Thanks jdobbs.

I know that you are keen on not wasting time and I also know that you have some difficulties with the English language so I'll try to keep it very simple:

The settings [good, better, high, etc.] determine the quality of the encode. The higher the setting, the longer it takes. Eventually, YOU have to decide what's best for a given source and how much time you are willing to "waste".

Edit: Or, just select "auto" and trust that jdobbs knows what he's doing. In your case, that's probably the best option.

[xterminater]

I do not have difficulties with the English languages. Forums are made so people typed shorthand, not to be grammatically correct with punctuations and all that. We are not in English classes so we do not have to follow a guideline to be graded upon. All my posts and comment are understandable and legit, so I do not know why you all keep saying I have trouble with my English language. I know the settings determined the quality of the encode, but what exactly does it do as far as the behind the scene work of it. I know in DVD-RB, the passes determined the quality because the more passes you pass through it, meaning the more CCE will go through and encode that segments. I do not know how x264 work in the encoding engine that's why I asked this question. But also many passes would not help if your movie is 2hr 30min long and you are trying to cram it in a DVD5 with DTS sound for example!

[extravirgin]

Unfortunately I'm not offering help on your encoding question, I'm hopefully addressing the English language issue and hope you don't take this the wrong way.

Yes forum speak can be littered with grammatical and punctuation errors but generally most on the board speak the same language natively. Problems arise when for example in this case, a non-native English speaker posts a question and certain assumptions are made by the native English speaking members. I (speaking as a native English speaker) infer certain things from the perceived tone of the post and the language used. The OP can be unaware that by choosing a particular word or phrase offence is caused. I'm sure this would be the case if for example I was posting on a board where the main language was French. Or Swahili. Or whatever.

In my opinion your original posts annoyed jdobbs because you came across as bullish and ignorant which I am sure you are not. Not only that but you were unaware that you had caused offence and took the references to the fact that you are not a native English speaker as some sort of personal attack. They were not and the fact that you are now addressing this question to jdobbs I think shows your ignorance/innocence in this matter.

[xterminater]

extravirgin, thanks for addressing to the topic. Obviously, when posting in forums, I along with anyone else want a quick response and straight to the topic. Sitting here and write in correct English and punctuation mark can take forever, and drag around the bush if you know what I mean. I already told jdobbs that I wasn't attacking him, nor offensive to him by dissing his program. Some of my posts as you can see earlier, are either asking questions which I'm not cleared about, or suggesting a future improvement of BD-RB. I may be posting in the wrong thread but don't kill me because of that. I told jdobbs many time that don't take anything offensive because if "you" want to make it sound offensive, of course it will be. Same as "make a mountain out of a mole hill" statement! I already stated that I know of nothing about high def and the way x264 works, as well as BD-RB in collaboration with each other. Also I couldn't find anything in this forum in regard to my problem which is why I made my own post.

[Mr. Movies]

Whether it takes you lots of time or not too properly address you questions and post them here is irrelevant. If you want quick accurate responses you NEED to do just that. You would be better off not asking a question than asking one improperly as all you’ll do is upset others that MAY HAVE helped you. You won’t catch flies without honey so they say. We all come across harsh at times and no one is perfect but at some point you, we all, have to improve and make changes, it seem obvious that you are bucking this requirement which you’ll pay for if you won’t learn.

I find your posts extremely difficult to follow so I can see why others are giving you flak about it.

[Groucho2004]

I know the settings determined the quality of the encode, but what exactly does it do as far as the behind the scene work of it.

No idea what you're getting at here. x264 is working behind the scenes. Is that what you mean?

I know in DVD-RB, the passes determined the quality because the more passes you pass through it, meaning the more CCE will go through and encode that segments.

That's a myth. The main factor for quality is bitrate.

But also many passes would not help if your movie is 2hr 30min long and you are trying to cram it in a DVD5 with DTS sound for example!

What is your point???

[xterminater]

@mr. movies... when I mentioned about the time it take to encode and wasting time, I mean that in a way such as is there a way we can bypass the encoding of an extra in two pass, but instead only doing 1 pass of it so that way I do not have to "waste" time on something I don't need. The problem with this is the extras in this bluray content is considered to be a "alternate movie" which is why BD-RB encoded it in two pass instead of one.

@groucho... yes I understand the x264 work behind the scene of BD-RB, but what I want to ask is the number of "passes" or the selection of different qualities, what does THAT mean. For example, like I mention in DVD-RB, the number of passes you do, ie..5, 6, or 7 means the number of time it will "filter" that segments or chapters of that DVD. But in this case x264 only do 2 passes, which shouldn't really matter what settings of quality you picked because they are all undergoing 2 pass. I want to understand what jdobbs means when he have 4 choices of quality, and how does that affect the x264 program and BD-RB. I wasn't trying to point out anything with the number of passes, or bitrate... I was making a statement that I know how DVD-RB worked with CCE but now with bluray, I have no clue! :\

[Groucho2004]

I want to understand what jdobbs means when he have 4 choices of quality, and how does that affect the x264 program and BD-RB.

The quality choices result in different encoding settings for x264, for example the "me" and "subme" parameters will have different values.

If you really want to know what all these parameters mean and how their values affect the quality I suggest you examine the command lines that BD Rebuilder produces for each setting.

Furthermore I suggest this link as a reference for all x264 command line options:
http://mewiki.project357.com/wiki/X264_Settings

[Groucho2004]

But in this case x264 only do 2 passes, which shouldn't really matter what settings of quality you picked because they are all undergoing 2 pass.

Sorry, I really can't figure out what this means.

[xterminater]

thanks groucho... that is what I really needed is that link! :) That is what I'm referring to when I want to know the behind the scene encoding steps of x264. As for the second quote you don't get it, let me re-elaborate on that. In BD-RB, during the encoding process it goes through 2 passes for each playlist of the bluray contents. That's why I asked that since it already goes through a two pass step process, it shouldn't make a difference if it's better, best, high quality, very high quality? Whereas before, in DVD-RB and CCE, you have an option to select up to 9 passes, so I'm guessing pass 3 means better and like pass 9 is more like high quality isn't it? Or does x264 work completely different? That's why I'm confused and had to ask that question.

[SquallMX]

That's why I asked that since it already goes through a two pass step process, it shouldn't make a difference if it's better, best, high quality, very high quality?

It does, 2-pass allows file size accuracy and better bitrate distribution than 1-Pass ABR, higher quality settings allows more efficacy at compressing video thus, at the same bitrate "Highest" will provide a better image than "Good" but the increases in quality come at the expense of longer encoding times.

"Automatic encoder quality" allows BD-RB to switch quality presets according to the video bitrate/target size. If the bitrate is really high (like for BD-25), there will be a great chance that the encoding differences are going to be small, so BD-RB will use the "Good" preset, but when encoding at low bitrates (BD-5) it will use "Highest" in order to squeeze every avariable bit for better quality.

:helpful:

[Groucho2004]

As for the second quote you don't get it, let me re-elaborate on that. In BD-RB, during the encoding process it goes through 2 passes for each playlist of the bluray contents. That's why I asked that since it already goes through a two pass step process, it shouldn't make a difference if it's better, best, high quality, very high quality? Whereas before, in DVD-RB and CCE, you have an option to select up to 9 passes, so I'm guessing pass 3 means better and like pass 9 is more like high quality isn't it? Or does x264 work completely different? That's why I'm confused and had to ask that question.

Why this obsession with the number of passes? It's not relevant, at least with x264.

I'm not going to explain the basics of x264 because you have all the info you need to figure it out.

Edit: Download this (http://mulder.dummwiedeutsch.de/pub/x264) document and read the "Rate control" section.

[xterminater]

Groucho, no no no obsession with passes with me. I was making a comparison that's all. Thank you for linking me to that page.

[writersblock29]

@xterminater

It's very true that certain versions of CCE allowed one to run a tremendous number of passes in order to squeeze as much quality as possible out of their encodes. I recall several posts made by people during the DVD backup days of old who would think nothing about spending three days encoding a single movie if they felt it would smooth out one stray pixel. The manufacturers of CCE had also said that anything over three passes is pretty much a waste of power and time (please don't ask for links--it's been years since I read that, and I couldn't even begin to know where to look right now). CCE, as an MPEG2 encoder, was bleeding edge for its time, and a very efficient encoder.

H264 AVC isn't MPEG2. It has an entirely different method of doing business, and is so efficient that I can't imagine what the encoder would do with several extra passes if one were to even try. Two passes is really all that's needed for optimum quality for a given bitrate. Try to think of it this way: Copying a 50GB Blu-ray to a dual-layer DVD is much on the same mindset as making a DVD backup to a CD-R... only the encoder is so efficient, many (if not most) users are very hard-pressed to be able to distinguish between the original and the copy. If you ever run into a Blu-ray movie that doesn't look good compressed down to a DVD9, chances are no amount of extra passes would have done anything to it--x264 is so efficient, it really wouldn't have anything "extra" to do with those passes: What you see is as good as the encoder's settings could do. In such an instance, you would probably be better off committing the project to a BD25 and being done with it.

This is oversimplified--forgive me for that--but I hope it helps.

[xterminater]

Well done explained writersblock! This is the type of communication between members that I am looking for, and is expecting! A "simple" explanation from you that you called oversimplified is not simple to me because it gave me great understand that what I had before, and also it is short and simple :) There's no need to bash one another, or think that someone is bashing you if you approach someone correctly as writersblock did to me. However, going back to CCE, I did compare one I did with 3 pass and one I did with 6 passes and I can see that the pixelation is less, and the picture are smoother. Most of the time I do 5 pass though, no higher.

[writersblock29]

However, going back to CCE, I did compare one I did with 3 pass and one I did with 6 passes and I can see that the pixelation is less, and the picture are smoother.

Well, beauty's in the eye of the beholder. That's the great thing about the various projects made available by the developers on this forum: They offer a larger degree of flexibility than we'd find in many commercial solutions--and sometimes only ask for donations instead of a sky-high price tag. The only downside is that a wide variety of choices can easily lead to confusion. Particularly when the person trying to make a choice is not a software engineer nor a video encoding expert.

While several members of this site have forgotten more about video encoding than you or I will ever know (bravo, guys--seriously), it helps to understand the stress that comes from being your own PR representative--or even a regular forum lurker. Nine times out of ten, any question one may consider asking has already been asked a zillion times in a zillion threads. While the intention of the asker is usually innocent, it's easy for a developer or regular visitor to feel frustrated in answering the same questions repeatedly. Add to that the fact that intent is hard to convey in text--and many members do not speak English fluently--it's hard sometimes to tell if a request or comment is trying to be light-hearted or being self-righteous; two opposites that can look the same in writing. The result is half-hearted, sarcastic, or just plain mean responses. Which is counterproductive to a forum in that it creates reluctance for participants to feel as if they are truly free to ask what's on their minds. It's catch 22.

The happy medium seems to be using the search function first, trying several different keyword combinations. If that doesn't work--or amounts to doing the same level of research that you'd expect to put into a medical essay--then sure, you betcha, ask. Then read your responses with a thick skin: In the end, some people are simply assholes. Others who sound like they are, aren't trying to be (remember it's hard to detect mood in text). In short, it can't be personal when very few of us here would know one another on the street. This place is so full of good information, it'd be a shame to allow a few bad experiences to steer you away.

[happycase]

I'm not sure if this is relevant to the discussion but to the best of my knowledge every commercial HD DVD and Blu ray encoder has supported three pass mode or greater since 2007.

[jdobbs]

I'm not sure if this is relevant to the discussion but to the best of my knowledge every commercial HD DVD and Blu ray encoder has supported three pass mode or greater since 2007. X264 supports multiple passes. In fact there is no limit to the number you can do (just use "--pass 3" for passes 2 and above). But it really doesn't improve the picture in any meaningful way. It only allows those who are obsessed with numerous passes to "feel" better. :rolleyes:

[Dark Shikari]

I'm not sure if this is relevant to the discussion but to the best of my knowledge every commercial HD DVD and Blu ray encoder has supported three pass mode or greater since 2007.Any encoder that needs 3 passes to do a good job is a pile of crap and written by idiots.

[LowDead]

Any encoder that needs 3 passes to do a good job is a pile of crap and written by idiots.

hahaa... Bullseye, Dark Shikari! :D

[writersblock29]

I'm not sure if this is relevant to the discussion but to the best of my knowledge every commercial HD DVD and Blu ray encoder has supported three pass mode or greater since 2007.

Maybe so. I wasn't stating that doing such a thing isn't possible... just that I question why anyone would. In other words, pretty much what Jdobbs and Dark Shikari said. ;)

[Biggiesized]

Any encoder that needs 3 passes to do a good job is a pile of crap and written by idiots.
Well, a third pass could be used to tune your encode if your settings aren't completely applicable to all parts of your video.

[JJB]

Jeeze, Good Luck... The developer of one of the greatest Hi-Def encoders produced and you will tell him he is wrong..forget the stupid settings that you and others will most likely not notice, your eyes are the only thing that counts.

[Dark Shikari]

More specifically --

A normal 2 pass serves the following role:

Pass 1: Gather statistics
Pass 2: Guess CRF based on statistics, encode

A third pass might be necessary in the following cases:
1. Pass 1 used a vastly different bitrate from pass 2, so the guess CRF is not very accurate.
... but this won't happen in x264, because x264 uses target bitrate mode on the first pass. This can only happen in encoders like Xvid or Theora that uses constant-quantizer first pass.

2. Pass 1 used vastly different encoding settings from pass 2, resulting in bad prediction on the second pass.
... but if this was a problem, you could just run pass 1 with the same settings, making a third pass unnecessary.

In terms of optimization algorithms, I can't think of any optimization algorithms that would require a third pass: most of the stuff x264 comes up with is done in the lookahead on the first pass. Even if you did something insane like RDRC, that would be lookahead-based (and mbtree is better anyways). This is because doing things in lookahead lets you try many possibilities (e.g. thousands of different B-frame patterns) as opposed to being restricted to one choice per pass.

[Sharc]

In other words, does it basically mean that pass 1 only serves the purpose of finding the right crf for pass 2 in order to hit the desired filesize?

[Dark Shikari]

In other words, does it basically mean that pass 1 only serves the purpose of finding the right crf for pass 2 in order to hit the desired filesize?Yes .

[writersblock29]

@Dark Shikari

Thank you for the summary. I had gotten that impression during the reading I did on how x264 does its magic--but nothing beats hearing it in simple terms from the horses' mouth. Awesome encoder.

[Sharc]

Yes .
Could the collection of the statistics for pass 2 and crf prediction be done with say every 2nd frame in pass1, i.e. reducing the time of pass1 by 50%? Would we be loosing too much information for obtaining a 'precise' (say within 2....3%) filesize?

[Dark Shikari]

Could the collection of the statistics for pass 2 and crf prediction be done with say every 2nd frame in pass1, i.e. reducing the time of pass1 by 50%? Would we be loosing too much information for obtaining a 'precise' (say within 2....3%) filesize?In theory there are many ways one could take shortcuts in the first pass for bitrate calculation. A simple one would be a low-resolution first-pass.

But this all becomes moot once you realize that the first pass's bottleneck is MB-tree and B-adapt, neither of which can be "shortcutted" because they're re-used on the second pass to avoid doing them again.*

*It would be relatively easy to write a small app to convert statsfiles between resolutions. This might be an interesting exercise, and I'm not sure how much it would hurt compression.

[Sharc]

In theory there are many ways one could take shortcuts in the first pass for bitrate calculation. A simple one would be a low-resolution first-pass.

But this all becomes moot once you realize that the first pass's bottleneck is MB-tree and B-adapt, neither of which can be "shortcutted" because they're re-used on the second pass to avoid doing them again.*

*It would be relatively easy to write a small app to convert statsfiles between resolutions. This might be an interesting exercise, and I'm not sure how much it would hurt compression.
Hmmm...., maybe a good ghost will take this up some day...:)