2009 MPEG-2 Encoder Test!

[kazyn]

lordsmurf you'd like to think your the know all about video encoding. I guess you can continue using your mainconcept piece of sh*t mpeg2 encoder. minidv2dvd you reasoning is so flawed its retarded. You guys are both tools. This guy went of out of way to make a test that is very relevant. Calling this a newbie approach just makes you guys look like newbs. HC and CCE are the most advanced mpeg2 encoders end of story. I've been a long time reader, first time poster because I felt it was warranted. lordsmurf you are a negative person, that always thinks he's right even though you are wrong (most of the time). Like thinking samsung makes good dvd burners.

[kazyn]

Or like how there's no difference between singapore and india dual layers. And like how you thought ritek made great media. I've burned lots of riteks and most of them are failing now. I do agree with you that pioneer makes the best dvd burners but lately there quality has dropped a little. They are still the best just they don't last as long as they use too. Sorry a little bit off topic. And I like your nomorecoasters site, you should just put ritek in 3rd or fourth grade media.

[lordsmurf]

Oh please....

I've never considered Ritek to be "great media" -- you've confused me with somebody else. I've always pointed out how poor they react to reading and playing -- not to mention a barely-accepted coaster loss rate when burning. Verbatim media is all made the same, this anti-regional fanboy-like rhetoric found ONLY online in forums is for the birds.

Again, this encoder test is a farce. It's completely flawed. I don't have to "know everything" to know the comparison test being done here is bullshit by somebody that doesn't know what they're doing. In fact, it takes only minimal MPEG encoding knowledge to see it for what it is. Only a "newb" would think this a quality test of any kind -- or a "test" period.

Samsung does make great burners -- or at least they did for a while there (2007-2008) --- not necessarily too sure about the drives on the market at this exact moment, summer 2009. Pioneer is still safe, always has been.

When you live in the "age of misinformation" being politically correct just isn't possible anymore. I call bullshit as I see it, and if that makes me "negative", then so be it. I'm here to help people, not coddle them and "respect their opinions" (misguided ones, usually). The difference between me and a lot of other folks is that when I don't know something, I STFU. There's a lot of topics here where I have no clue -- therefore I don't post.

[redwudz]

To all posters, please keep it civil and on topic.

Fair warning,

Moderator redwudz

[CubDukat]

I was just curious about a few things about your settings:

I always thought that DC precision for DVD encoding should be 10-bit, and I was wondering how come you decided to use that setting instead.

I alternate between using TMPGEnc Xpress 4.0 and HCEnc. I like HCEnc's speed, but the results seem a little softer than I like. I had that problem with TMPGEnc too, but there was a setting to ddisable the low-pass (?) filtering that I disabled. That ended up giving me better results with my encodes from OTA HDTV. I was wondering if you know of a way to do something similar in HCEnc, because the only other reason I wouldn't use it is because of its lack of a native batch encoding feature. TMPGEnc is way too flaky in that regard, and I need something more reliable.

I tried crunching down a few episodes of "Southland" with MediaCoder, but I was not at all impressed with its performance and /or quality at lower bitrates--not to mention setting things up was extremely fiddly.

[Sartori]

lordsmurf you are a negative person, that always thinks he's right even though you are wrong

Doesnt make him wrong though , being objective about someone opinions on different matters is what counts for a logical discussion , its assumption that even though he might have been wrong on something else , that by a deluded sense of logic , he must be wrong on this . Assumption = mother of all cockups .

Its a test for newbs wrapped up with to much tech. detail for newbs .

[kazyn]

So why is his approach flawed? I read your other post trying to explain it but it makes no sense. Saying encoding something without temporal compression is just a bunch of crap and you know it. His approach was to show you that at a certain bitrate this encoder is better while using the best possible settings. The source was blueray is completely fine, A encoder still has to rebuild the stream, still giving some idea of quality. The only thing flawed is the fact that he used stills to show the differences (he shoulda used clips). But it still gives you some idea. Most people use mpeg2 encoders to backup to dvds (ie dual layers, BD, etc). He's not trying to decieve anyone, I agree with him 100%. CCE and HC are just technically more advanced with things like AQ, Lumgain, Multipass, etc (HC is the most advanced encoder out there but its still beta so it still has flaws). It really sounds like you try to know what your talking about, but everytime you open your mouth you just make yourself sound dumb. It's like your listening but its in one ear and out the other. Which mpeg2 encoder do you think is the best? mainconcept? ppfffftttt.

temporal compression only the changes from one frame to the next are encoded as often a large number of the pixels will be the same on a series of frames.

Also the india vs singapore thing, Do you think its coincidence that the india ones are worse? I never believed it at first t'ill I bought some. I remember reading someone was trying to tell you this, but you just ignored it. The india ones never worked properly but when I burned singapores ones with the same data it works fine. There are countless number of people that have the same exact problem. They are maybe made the same way but something must be a miss here. I agree with most of your opinions on media but you tend to ignore the real life situations.

[manono]

I was just curious about a few things about your settings:

I always thought that DC precision for DVD encoding should be 10-bit, and I was wondering how come you decided to use that setting instead.

If you're addressing the guy that did the test, he's long gone. The point you bring up is only one of the many things flawed about his comparison, although I think he used the same DC Precision setting for all so all the results were degraded equally. He thought he was an expert in the use of all the encoders but he didn't know much about any of them.

I was wondering if you know of a way to do something similar in HCEnc

There's no filtering in HCEnc as there is in CCE but with a way to turn it off, and there is in Procoder but with no way to turn it off. I didn't know TMPGEnc had such a thing turned on, but I don't and won't use that one, and don't pretend to know anything about it.

All the filtering in HCEnc is done in the AviSynth script (if you filter at all there) and in the chosen quantisation matrix. I can't remember if AQ and Lumagain are on by default. They will affect the results if turned on.

[kazyn]

If you're addressing the guy that did the test, he's long gone. The point you bring up is only one of the many things flawed about his comparison, although I think he used the same DC Precision setting for all so all the results were degraded equally. He thought he was an expert in the use of all the encoders but he didn't know much about any of them.

DC precision really depends on the bitrate you are encoding at. 8 = Low bitrates, 9 = medium bitrates (mostly used for dvd5 encodes), 10 = high bitrates. So really what he did was right, nothing flawed about that. Anything else you wanna comment on genius. What else is flawed?

[lordsmurf]

Saying encoding something without temporal compression is just a bunch of crap and you know it.

And this is why people like you can barely make a passable Youtube encode, while folks like myself create videos that are sold on iTunes, Amazon, etc. And all the while, you sit at your computer, smart-ass keyboard warrior that you are, and wonder "gee, how did he make that look so good?"

. Most people use mpeg2 encoders to backup to dvds (ie dual layers, BD, etc)..

Bullshit. Most people are still using DVD Shrink or some other all-in-one transcoder.

You don't know half as much as you think you do.

[manono]

DC precision really depends on the bitrate you are encoding at. 8 = Low bitrates, 9 = medium bitrates (mostly used for dvd5 encodes), 10 = high bitrates.

It has nothing much at all to do with bitrate, but with the resulting Q. However, for DVD video 10 is usually the recommended setting. Yes, for all bitrates. If you want to go around throwing out gratuitous insults to one and all, then please know what you're talking about first. Either way you'll be thought of as a jerk, but you'll at least be a knowledgeable jerk.

[davidcw]

All the filtering in HCEnc is done in the AviSynth script (if you filter at all there) and in the chosen quantisation matrix. I can't remember if AQ and Lumagain are on by default. They will affect the results if turned on.

It's a matter of personal preference, but for 2hr movies I find the Fox 2 matrix with AQ and Lumigain gives a sharper picture than using the standard Mpeg2 matrix with AQ and Lumigain.

[kazyn]

It has nothing much at all to do with bitrate, but with the resulting Q. However, for DVD video 10 is usually the recommended setting. Yes, for all bitrates. If you want to go around throwing out gratuitous insults to one and all, then please know what you're talking about first. Either way you'll be thought of as a jerk, but you'll at least be a knowledgeable jerk.

A quote from here (http://www.digvid.info/tmpgenc/settings.php) probably explains it best:

DC component precision
The DC component represents the average brightness of a block (8x8 pixel region). In MPEG-1 video this is fixed to 8 bits, but MPEG-2 allows higher precision. It is important that the DC component is accurately represented, so a setting of 10 is recommended for MPEG-2 sequences, unless you are encoding using a low bitrate.

A quote from another forum

It's an option to pull just a tiny bit more quality out when you've already maxed out the bitrate, because it reduces the roundoff errors inherent in the DCT transform. It's only useful when decoding roundoff causes errors <= 1-2 values. If you use it at a low bitrate all you're doing is storing artifacts at a higher resolution, which is pretty much useless, and robbing it of 1% or so bitrate.

CCE has an automatic setting that will increase it if the quantizer is low enough; their guideline is "8-bit below 4mbps, 9 at 4-8, 10 above 8", but like any bitrate recommendation it varies based on compressibility.

Maybe you should do some reading manano before giving people the wrong information. I guess I do know what I'm talking about.

[kazyn]

And this is why people like you can barely make a passable Youtube encode, while folks like myself create videos that are sold on iTunes, Amazon, etc. And all the while, you sit at your computer, smart-ass keyboard warrior that you are, and wonder "gee, how did he make that look so good?"

Dude your not getting it, read a little bit. Having video without temporal compression has nothing to do with this person was trying to achieve. The mpeg2 encoder still has to encode. Thats what I'm trying to tell you. Your outright calling him a noob, but your missing the simple things that makes your look like a noob.

Bullshit. Most people are still using DVD Shrink or some other all-in-one transcoder.

You don't know half as much as you think you do. rolleyes.gif

I shoulda rephrased myself there, any smart person would know to use encoder over transcoder. But your right most people still use dvd shrink. But you still avoided a simple question whats your favorite mpeg2 encoder? What about H.264 encoder if your even into that? No comments on verbatim dual layers?? <--even though this is off topic, I've been doing this shit for 9 years, I know way more than you. It seems that when your put in your place, you avoid things that are said to you when there right. Like what deadrats said, its really a douche thing to say, worse when your wrong.

[manono]

CCE has an automatic setting that will increase it if the quantizer is low enough; their guideline is "8-bit below 4mbps, 9 at 4-8, 10 above 8", but like any bitrate recommendation it varies based on compressibility.

Maybe you should do some reading manano before giving people the wrong information. I guess I do know what I'm talking about?

Since you think that CCE and HCEnc provide the best results (and for what it's worth I agree with you), maybe you'll take what it says in the CCE user's manual (for 2.70) as being somewhat authoritative, rather than a guide or post by someone that may or may not know what he's talking about. In the case of the quote above, can you provide the link, because on the face of it it makes no sense as the DC Precision, as far as I know, has no automatic setting. It sounds to me as if he's mixing up DC Precision and AQ. Anyway, this is what the CCE users manual has to say about Intrablock DC Precision (Section 5.3.7, Page 77):

Specify the bit precision of the DC coefficient of intra-blocks. 10 is recommended for DVD video. This setting works on MPEG-2 output.

Did you join this site and post just to pick fights with the regulars around here? That's not really a recipe for winning friends and influencing people. Civil discussions are one thing; trolling and bashing for no good reason are quite another.

[lordsmurf]

A quote from here (http://www.digvid.info/tmpgenc/settings.php) probably explains it best:

Ah, perfect ..... let me translate this for everybody...

"I've started an argument with people who know more than I do. Since I don't actually know anything, I'm going to post links as my defense, and hope that nobody notices."

[kazyn]

lordsmurf there's no arquing with you, your just retarded. I'm done with you. I'm sorry Mods here, I can't takes it anymore. lordsmurf atleast try to be constructive not making yourself even look more dumb. Why are you avoiding my questions? Cause your so smart.

Manano atleast your trying, I gave you 2 links, They outright tell you that your wrong, I even put the parts in bold, so you get it... I guess you don't. That other link is from a member at doom9. Ya I mixing up AQ, with DC precision thats it...

[kazyn]

Here's another post in videohelp forums

https://www.videohelp.com/forum/archive/quot-auto-intra-dc-precision-quot-setting-in-cce-t312217.html

vcddude posted 2006 Oct 24 13:21
Hi -- I looked around in the forum and could not find an answer to these questions:

What exactly does Cinemacraft's (CCE) "Auto Intra DC Precision" setting do?
Am I better off selecting "10" for DV-to-DVD MPEG2 conversion, or better off
selecting "Auto"?

Thanks,
Kevin


adam posted 2006 Oct 24 20:01
It automatically selects a DC precision for you based on the set bitrate and the overal complexity of the source. It's not a bad idea to use it. No you should never select a DC value of 10 unless you use a very high bitrate, probably at least 7mbits. The CCE manual gives you a general rule to follow. I think its something like up to 4mbits you should use a DC of 8. Everything else should be 9 unless your bitrate is very high.

[lordsmurf]

I'm done with you.

Another perfect statement.. Here, let me translate....

"Aw, man, I'm getting beat up in this place. I thought I knew what I was talking about, and now they're talking all this techie stuff and I don't understand. I can't find it in the instruction book, or the help files. I don't see that stuff in my one-click software. I tried to post some links, but they saw through my BS. I better get the hell outta here!"

[lordsmurf]

Here's another post in videohelp forums

Translation...

"Crap, I tried to leave, but they were still giving me crap even after I left. Man, that makes me mad. I'll never be able to sleep until I prove those strangers on the Internet wrong. Maybe I can find something with another Google search. Hey, maybe a forum search! Hey, I know -- I'll find something on their own forum and quote it, then I'll teach those guys in that post. Yeah, that's it. Bastards, I'll show 'em."

[ZQX]

You crack me up, Lordsmurf!

[kazyn]

Look nows he making fun of me, I'm so beat up I can't take it anymore... Lordsmurf there's no reasoning with you, When I say reason, you reason with me, come back with something intelligent to say. Be constructive, stop trying to make yourself look cool. And stop getting off topic like I am.

[kazyn]

https://forum.videohelp.com/topic137654.html <-- lordsmurf even comments on dc precision!

https://forum.videohelp.com/topic238248.html <-- another post related to dc precision

There you go manano.

[Midzuki]

Too bad the OP apparently doesn't know mpeg2enc.

[manono]

I gave you 2 links

Then I must be blind as I see only the one link in that post of yours. I don't see a Doom9 link. And even the TMPGEnc guide recommends a setting of 10 unless for low bitrates. But that qualification is meaningless. What's a low bitrate? 1000, 2000, anything below 4000? And bitrate has very little relation to the average quant as different materials compress differently. So, a "low" bitrate for one source might be perfectly fine for a different source. In addition, being a TMPGEnc guide, what it says doesn't necessarily apply to CCE or HCEnc. Why not? Because TMPGEnc, the last I looked, allowed only its one quantisation matrix, where when using CCE or HCEnc you can adjust the matrix to the material and desired output file size. You've proven nothing at all except your lack of understanding of MPEG-2 encoding, no matter how many years you've been at it. Like simps, the emperer has no clothes.

Use any settings you like. That particular one won't make a whole lot of difference anyway. CubDukat asked a question and I answered it to the best of my ability. I still think the setting used in the encoder test was wrong and still think CubDukat is correct to use 10. The CCE manual even recommends 10. With no qualifications.

There you go manano

You're dredging up threads that are 5 and 6 years old? Gimme a break. A lot has changed since then. As for Adam's 2006 response, they must be using an older CCE version, as there's no automatic setting, and the current CCE manual recommends a setting of 10. This is especially true for reencoding of professionally produced retail DVDs (which I, and I think you, deal with mostly). I've used the 8 setting before, but only for very, very low bitrates where my matrix couldn't produce a low enough Q without artifacting. Check your own CCE. Unless it's the 2.50 version, do you see an Automatic setting?

In one of those early posts (from 2003), Adam says that most retail DVDs use 8, which is just nonsense. It was wrong then and it's even more wrong now. Adam's a fine fellow and I used to learn from him, but that statement is just wrong. Just start checking studio produced DVDs. It's not hard to do.

[kazyn]

See thats what I like manano. Your starting to ask the right questions. My personal opinion is anything under 3500 is 8, 3500-6000 is 9, and 6000 plus is 10. DC precision will be the same for every mpeg2 encoder unless theres some advanced implemenation of it (DC precision is the same as it was 8 years ago it didn't change). Most DVDs are encoded with a DC precision of 9. I will ask the author of HC and see what he says. DC precision isn't even that much of a deal when it comes to quality, but it will waste bits if its not set correctly. Looking at retail DVDs is also a good idea which I've never done before. But I'm 99% sure the load of em are 9. The point of me dredging up old posts is to give you an idea that I'm not the only one that thinks this way. Even though there old doesn't mean there not relevant. Thats also another good question, Different sources compress differently so therefore what is a low bitrate for a certain source, but you could probably be certain that anything under 3000 should and would be consider low as the quality will start to suffer with any encoder. I find HC does the best at lower bitrates. To be honest I haven't really been using CCE for a long time, I'm strictly a HC guy lately. How did I show a lack of mpeg2 encoding knowledge?

[dr.bling]

great job on the comparison, do you have how long each video took to encode for a compairson ?

[ronnylov]

I read somewhere that using DC precision of 10 is overkill at DVD bitrates but I think the discussion then was when converting interlaced DV to DVD. Maybe progressive movies which are easier to encode sometimes can justify a DC precision of 10. By an old habit I always use DC precision 9 unless the bitrate is "very low", then I use 8.

But this is just old habits I have been using. So you mean that it is probably better to choose a quantization matrix suited for the compressibility and bitrate of the actual video to be used together with DC precision of 10 instead of using a matrix suited for a little bit higher bitrate together with a lower DC precision?

And how to choose the best quantization matrix to use? Are you performing compressibility tests and optimizing for certain average quantizer by using different matrices at the actual bitrate? Or is it just a matter of a complicated guessing? When I am using HC encoder I mostly don't have a clue of which quantizer matrix to choose, only a feeling that this is a pretty high bitrate for the actual source so choose one of the better or the other way around if the bitrate is too low for the actual video...

When comparing different encoders, should you use exactly the same settings for all of them, or should you try to tweak the very best out of each encoder? Or maybe is the default settings what you should compare? I remember a couple of years ago many people said that mainconcept encoder gives bad quality. But then I tried it myself and they were right if using default settings, but going into advanced settings there was very much I could tweak and after doing some tweaking it was just as good as any other encoder I had tried. So if one encoder is very tweakable it can be better for a user that knows what to tweak, but for most people it is much better if the defualt settings are better.

In my opinion these compare encoder threads are somewhat useless. The result will depend too much on how you use the encoder and what you use it for. As an example of this thread is the test with procoder, some people may not know the problems with the levels but if you know that feeding it with yuy2 video will work then you get a different result. If I remember correctly you can not alter the quantizer matrice with procoder, then you may get a better result with HCEnc if you know how to choose the optimal set of matrices. Is it then fair to only test with default settings or should you tweak each encoder to its own maximum? Most people don't know how to tweak the very best out of each encoder (I don't for sure) so then the comparison may be useless anyway...

[manono]

Most DVDs are encoded with a DC precision of 9.

But how do you know that if you've never looked? Of the 6 on my hard drive at the moment, 5 use 10, and I was very surprised that the sixth one used 9. Then I took a closer look and saw it was from a B-grade DVD production company, that it didn't compress worth a damn, and that the encoder was justified in using the 9 setting. In any event, if I had to guess I'd say that over 95% of the movies on DVD from reputable DVD production companies use 10. To which you might respnd that they're using DVD9s while we in many, if not most, cases are recompressing for DVD5. And as perhaps you know, many movies on DVD, stripped of their menus, extras, extra audio tracks and the like, are under DVD5 size. In any event, even if compressed for DVD5 most movies benefit from the 10 setting.

In addition, something ronnylov alluded to is that many of the people here aren't reencoding movies, but are putting their own videos onto DVD. And in many cases these are interlaced and created using consumer video equipment under less than ideal conditions, and for that kind of encoding for DVD it's a completely different game.

...but you could probably be certain that anything under 3000 should and would be consider low as the quality will start to suffer with any encoder.

And again, it depends entirely on the source. DV AVI of someone's home videos taken without a tripod might not compress decently given an average bitrate of 8000. On the other hand, I've done several reencodes of films at less than 3000 with which I was satisfied. As I keep repeating, it's all about the Q - Q as in quant and Q as in quality. There's just not much correlation between bitrate and quality.

Anyway, at least you're more pleasant to talk to when you're not ranting about this and that. I've enjoyed it.

[manono]

And how to choose the best quantization matrix to use?

I want them to come out with a certain final average quant - a somewhat lower average when using a lower quality matrix, and a little bit higher average when using a better quality matrix. At first it's an educated guess. If the MPV winds up outside the wanted result I'll run a few more passes with the final matrix. I'm using CCE so it's easy and fast to just run an additional 3 passes with only the matrix changed (and sometimes the file size/bitrate also changed if too large or small). Either that or stop the encoding in the middle and change the settings when the quants being reported turn out not to be what I had hoped. I might have to run extra passes about half the time. I suppose if using HCEnc you'd have to start over.

So you mean that it is probably better to choose a quantization matrix suited for the compressibility and bitrate of the actual video to be used together with DC precision of 10 instead of using a matrix suited for a little bit higher bitrate together with a lower DC precision?

I suppose that's what I mean, since I always use 10 for the movie. If there's an extra I want but I don't want to compromise the quality of the movie itself too much by including it, and I have to give that extra a lower quality than I might like, only then will I use a DC Precision of 8 or 9. I don't pretend that these methods are suitable for everyone. For one thing the whole process is entirely manual and hands-on and most people have neither the desire nor the time to tweak and retweak everything. But since you asked, I explained it.