the Matrix

[kinneera]

First off, a very nice well-reasoned discussion of some of the technical aspects. I like the conclusions too.

Point, the first: You guys who are making generalizations on the "effective average" bitrate of a CQ encode using Kwag's approach are a little off the mark.

I would just like to clarify the point I think was trying to be made about CQ encoding (at least that I was trying to make). Basically, no one ever said there was necessarily anything wrong with the long GOP. Rather, we're addressing the fallacy of claiming the CQ encode method itself is better.

The fact that a long GOP can provide a better quantization per given bitrate is not dependent on using TMPGEnc's CQ encoding mode. Specifically, as your discussion shows, the key is optimal encoding of P and B frames over such an extremely long GOP. And the key to optimal encoding of P and B frames is multiple passes, with refinement on each pass. In other words, if TMPGEnc has gotten so good at encoding P and B frames in CQ mode, it will do even better when allowed to do 2 passes. And by extension of logic, CCE will do that much better given more passes (I will concede here that CCE does not to my recollection allow GOPs that long, however). It's no accident that the corporate version of TMPGEnc (at least the one they claim will exist according to their pricing scheme) is the one that allows up to something like 20 passes.

CQ is nothing more than a single-pass VBR encode that puts priority on maintaining a given quantization level rather than maintaining a specific bitrate average. This is fine if space is not the final consideration (pure theory would argue that CQ shouldn't even have a maximum bitrate parameter), but not for other circumstances. If CQ and multi-pass VBR produce a file of the same size, the multi-pass will always look better because the CQ will have done the equivalent of the multi-pass's first pass but without the second (or more) pass(es).

[adam]

Well actually I did not claim that the kvcd quality of Farscape is as good as dvd.

I never claimed you said kvcd has as good a quality as dvd, I claimed that you said that kvcd can "have quality as good as the source dvd."

Adam and Kineera, I could put a wager on it that that Film [Farscape] as seen on my 19inch philips 4:3 t.v carries no discernable difference when viewed from 10ft (ave viewing distance ??)as compared to the original dvd.

If there is no difference than it is just as good isn't it? If you want to promote any format or encoding technique that is fine just be realistic, consistent, and above all else honest.

[kwag]

In other words, if TMPGEnc has gotten so good at encoding P and B frames in CQ mode, it will do even better when allowed to do 2 passes. And by extension of logic, CCE will do that much better given more passes (I will concede here that CCE does not to my recollection allow GOPs that long, however). It's no accident that the corporate version of TMPGEnc (at least the one they claim will exist according to their pricing scheme) is the one that allows up to something like 20 passes.

You're wrong again kinneera.
It doesn't!
Why don't you try it. I did!.

Try TMPGEnc 2.54a against the latest CCE. You'll see what I mean.

Try a 4 pass, the most I tried, against TMPEG's CQ=80 on a +-90 minute film.
You'll see what I mean.

kwag

[bilbogod]

Actually no "discernable" (means to see or recognise clearly) not "no difference". I think that was also qualified by the inclusion of the viewing distance, which is more than just relevant. What is implied and what are stated are two different things.

[e-z-e]

Kwag,
How much longer does a "kvcd" take to encode than a standard VCD?
I might give your template a try.

e-z-e

[kwag]

Kwag,
How much longer does a "kvcd" take to encode than a standard VCD?
I might give your template a try.

e-z-e

As an example, "The Matrix", takes about 5 hours on my Pentium 4 @1.6Ghz.

A standard VCD encodes almost real-time with my current hardware.

So it takes about twice than a VCD to encode with the current KVCD template.

kwag

[kinneera]

Try TMPGEnc 2.54a against the latest CCE. You'll see what I mean.

As I indicated, CCE does not allow that type of GOP. Thus the comparison is invalid. But that GOP is not consistent with my goals, nor is cramming a movie into one disc, so I don't care. The point I was making is that from a purely technical standpoint (with contributing information from mitsui, thanks again to him) there is no valid argument that 1-pass VBR (CQ in TMPGEnc) will ever be better than multipass, no matter what encoder.

You're wrong again kinneera.
It doesn't!

From someone who already thinks 1-pass looks fantastic, your belief that 1 or more extra passes doesn't provide any improvement is not surprising nor does it carry any weight with me. In your particular case, then, my argument is purely academic and I wouldn't even expect it to fall on anything but deaf ears anyway. I'll leave it at that.

[kwag]

Try TMPGEnc 2.54a against the latest CCE. You'll see what I mean.

As I indicated, CCE does not allow that type of GOP. Thus the comparison is invalid. But that GOP is not consistent with my goals, nor is cramming a movie into one disc, so I don't care. The point I was making is that from a purely technical standpoint (with contributing information from mitsui, thanks again to him) there is no valid argument that 1-pass VBR (CQ in TMPGEnc) will ever be better than multipass, no matter what encoder.

You're wrong again kinneera.
It doesn't!

From someone who already thinks 1-pass looks fantastic, your belief that 1 or more extra passes doesn't provide any improvement is not surprising nor does it carry any weight with me. In your particular case, then, my argument is purely academic and I wouldn't even expect it to fall on anything but deaf ears anyway. I'll leave it at that.

Theory is not the same as practice.

In theory, multi-pass should be better than a single pass ( or CQ or whatever ). The encoder has more runs to re-analyze the data and distribute the bitrate to improve quality. It all boils down to that.

In practice, TMPGEnc is better in CQ mode, than 2-pass mode.

That's my point. But still, I have tested multi-passes with CCE, and TMPEG's CQ mode beats it, in every test.

At least on movies longer that 90 minutes.

Why? I don't know. Even at the standard GOP in TMPEG. It still beats CCE's x-pass.

So technically speaking, TMPEG is a better encoder. Not to mention it's a "swiss army knife" with all the included utilities it has, that don't even exist in any version of CCE.

kwag

[kinneera]

So technically speaking, TMPEG is a better encoder. Not to mention it's a "swiss army knife" with all the included utilities it has, that don't even exist in any version of CCE.

No...for your particular purposes, TMPGEnc is a better encoder. And as a side note, just about anything that can be accomplished with TMPGEnc can be accomplished faster and often better with Avisynth+CCE. But for quick learnability, you're right...you can't beat TMPGEnc. And lest I be accused of favoritism, know that I still use TMPGEnc quite often. But most importantly, since the last place this thread needs to go now is into a pissing contest about TMPGEnc vs. CCE, I'm not going to bother to say any more about it.

[Goldenaxe4000]

Hello All
for dvd stand alone :
feet a movie on 1 cd-r may available via compress vcd .
but the cd-r is 99 minutes ! not 80 !
take a template at www.geocities.com/goldenaxe4000
you need a burner with over burn feature and a media !
and u need to take the VCD and make a compress VCD


i made the matrix , rip from dvd in 2 options :
VCD 1.31 Giga !
SVCD 2.62 Giga ! <almost twice>
DVD 5.3 Giga ! <just for the test i let the FLASK make dvd 720x576>

the quality on VCD is MORE THAN VHS and it look excellent on my DVD !
( A FRIEND of mine wish to take the dvd and rip on him . when i told him it is a vcd he was in shock ! )

ig you have anymore question . u may mail me to
goldenaxe4000@yahoo.com

[Stinky]

get DVD2SVCD - adujst the settings on the bitrate tab so that the movie fits on 2 disks

it will take care of the rest

[bbb]

Does DVD2SVCD allow one to split the movie at a specific transition point of one's choice? Or does it still split according to file size. Or by frames?

[Stinky]

it will split it by size -- fitting the highest quality SVCD into the number of cdr's you specify (on the bitrate tab)

it does not allow you to pick a specific point

[stevemoley]

hi all,

right, im only gonna post this once so read if you want to.

"matrix" was the first kvcd i made and i was over the moon with the results. it was much better quality than the 2 disc vcd i had and the total file size(with most of the credits cut) was 570meg(mode 2) i put it on a 650meg cdr. when i encoded the dvd to avi, the file size was 798meg.
the picture was much crisper than a std vcd even at the low motion scenes where the bitrat went to as low as 400kbs.
please be aware that i changed the kwag template slightly to suit my dvd player and reduced the res to 352 x 288 for pal .
the tmpgenc optimizations kwag has done are very impressive and unless im making dvd files, i use his template for all my dvd rips.

thankyou kwag

-moley

[kwag]

hi all,

right, im only gonna post this once so read if you want to.

"matrix" was the first kvcd i made and i was over the moon with the results. it was much better quality than the 2 disc vcd i had and the total file size(with most of the credits cut) was 570meg(mode 2) i put it on a 650meg cdr. when i encoded the dvd to avi, the file size was 798meg.
the picture was much crisper than a std vcd even at the low motion scenes where the bitrat went to as low as 400kbs.
please be aware that i changed the kwag template slightly to suit my dvd player and reduced the res to 352 x 288 for pal .
the tmpgenc optimizations kwag has done are very impressive and unless im making dvd files, i use his template for all my dvd rips.

thankyou kwag

-moley

Thanks moley:

And if you were impressed by the results, have a look at the new parameters and the template released today.
The quality has almost doubled.

There is a link to a new sample done today, actually from "The Matrix"

Here are the changes, explanation, and links:
http://www.kvcd.net/forum/viewtopic.php?p=755#755

kwag