TMPGenc: CBR 7500kbs or CQ 90% with max 8250kbs

[Guest]

Archived from groups: rec.video.desktop (More info?)

I need to fit an mpeg2 file on DVD. So the size is pretty critical. I know
I can be pretty good with CBR with bitrate 7500kbs. Or I can try CQ say
at 90% with 8250kbs. What would give me a better quality?

--Leonid

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

On 3 May 2004 04:07:15 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>I need to fit an mpeg2 file on DVD. So the size is pretty critical. I know
>I can be pretty good with CBR with bitrate 7500kbs. Or I can try CQ say
>at 90% with 8250kbs. What would give me a better quality?

CQ shall always give you the best result, for the same given average
bitrate, because it is the only mode that gives the same quality to
each frame.

Next comes multipass VBR, which also tries to give the same quality to
each frame (and thus is an attempt to get a CQ encoding on the basis
of the average bitrate), but it's not guaranteed to get it. In fact,
examining the encoding logs you can see differences with the CQ mode.

CBR represents the most unequitative use of the available bits.

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

Bariloche <bariloche@bariloche.com> wrote:
: CQ shall always give you the best result, for the same given average
: bitrate, because it is the only mode that gives the same quality to
: each frame.

But what is given average bitrate if I specify min 2000, max 8000 and say 90%.
How do I calculate the average bitrate?

: CBR represents the most unequitative use of the available bits.

True, but if you just want to put 50 minutes of video on 1 DVD, that's the
most proper thing to do.

--Leonid

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

On 3 May 2004 23:23:21 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>Bariloche <bariloche@bariloche.com> wrote:
>: CQ shall always give you the best result, for the same given average
>: bitrate, because it is the only mode that gives the same quality to
>: each frame.
>
>But what is given average bitrate if I specify min 2000, max 8000 and say 90%.
>How do I calculate the average bitrate?

What I said would apply to the following: you perform a CQ encoding,
then you look at the log file to find out what the resulting average
bitrate has been. You can then encode in VBR mode, using that figure
as the average bitrate. Theoretically you would get the same encoding,
but in practice you don't, even if you get the same file size.

Now, the problem you wonder about is right the opposite: if I know the
average bitrate, how do I know the CQ mode that ends up using that
same average bitrate? If you have some experience with CQ encoding,
you can more or less know the bitrate given by each quality
percentage. But of course, it depends completely on the film, so you
cannot know beforehand and precisely, the bitrate attained with CQ
encoding.

This has been dealt with formerly in the newsgroup, and apparently
people is doing a tentative encoding of an Avisynth script, that can
give you the bitrate attained by a certain CQ mode, within a
reasonable degree of exactitude. Something like

AviSource("TheClip.avi")
SelectRangeEvery(300,15)

shall give you a clip made by 15 frames from every 300 (you should
encode the same number of frames that comprise a GOP, which would be
15 for PAL). If you encode this one, it can give you a reasonable
insight on the bitrate you would get when encoding the whole. Of
course, if it's not the bitrate you want, you would have to encode at
a different percentage of CQ, and keep trying, etc. If your concern is
getting an exact file size, it may not be worth that much effort, and
you should just use VBR mode instead.

>: CBR represents the most unequitative use of the available bits.
>
>True, but if you just want to put 50 minutes of video on 1 DVD, that's the
>most proper thing to do.

No. CBR allows you to calculate exactly the file size -but VBR does
exactly the same for you. You just multiply the constant bitrate of
the CBR mode, or the average bitrate of the VBR mode, by the seconds
of movie, and you get the size of the resulting mpeg video. The
mathematical definition of "average" is the very guarantee of that.

Now, let's not forget that if you intend to make a menu, that shall
also take some space on disk.

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

Bariloche <bariloche@bariloche.com> wrote:
: a different percentage of CQ, and keep trying, etc. If your concern is
: getting an exact file size, it may not be worth that much effort, and
: you should just use VBR mode instead.

I will have about 75 or 70 minutes of NTSC video all together with uncompressed
PCM sound. If it were just 65 or 60 minutes I would've fitted it. But I need
to compensate on something so that I could either go with 7500 CBR or try to
boost max bit rate to 8250 and have CQ at 90% or 80%

:>True, but if you just want to put 50 minutes of video on 1 DVD, that's the
:>most proper thing to do.

: No. CBR allows you to calculate exactly the file size -but VBR does
: exactly the same for you. You just multiply the constant bitrate of
: the CBR mode, or the average bitrate of the VBR mode, by the seconds
: of movie, and you get the size of the resulting mpeg video. The
: mathematical definition of "average" is the very guarantee of that.

My point:
If size does NOT matter , would you still go CQ? I'd definitely go with CBR
max bitrate. What is disatvantage of that as far as quality of the video goes?

--Leonid

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

"Leonid Makarovsky" <venom@cs.bu.edu> wrote in message
news:c798mr$o3m$1@news3.bu.edu...
> Bariloche <bariloche@bariloche.com> wrote:
> : a different percentage of CQ, and keep trying, etc. If your concern is
> : getting an exact file size, it may not be worth that much effort, and
> : you should just use VBR mode instead.
>
> I will have about 75 or 70 minutes of NTSC video all together with
uncompressed
> PCM sound. If it were just 65 or 60 minutes I would've fitted it. But I
need
> to compensate on something so that I could either go with 7500 CBR or try
to
> boost max bit rate to 8250 and have CQ at 90% or 80%
>
> :>True, but if you just want to put 50 minutes of video on 1 DVD, that's
the
> :>most proper thing to do.
>
> : No. CBR allows you to calculate exactly the file size -but VBR does
> : exactly the same for you. You just multiply the constant bitrate of
> : the CBR mode, or the average bitrate of the VBR mode, by the seconds
> : of movie, and you get the size of the resulting mpeg video. The
> : mathematical definition of "average" is the very guarantee of that.
>
> My point:
> If size does NOT matter , would you still go CQ? I'd definitely go with
CBR
> max bitrate. What is disatvantage of that as far as quality of the video
goes?
>
> --Leonid

Why not just try it both ways and examine the results?

I mean, we're just talkin' BITS here, no harm, no
foul in experimentation.

What everyone else thinks of your final result
does not matter-what YOU think does.

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

On 4 May 2004 23:24:11 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>If size does NOT matter , would you still go CQ? I'd definitely go with CBR
>max bitrate. What is disatvantage of that as far as quality of the video goes?

Encoding CBR at the maximum bitrate, and encoding CQ at the maximum
quality, for full D1 resolution, would bring the same results, because
the 100% quality lies somewhere between 12,000 and 15,000 kbps
(extreme cases aside).

But if you use something other than D1, and/or if you encode at lower
bitrates, then CQ gives better results than CBR. CBR would be faster
than VBR, but the latter is definitely better. And CQ is VBR made
right.

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

Bariloche <bariloche@bariloche.com> wrote:
: On 4 May 2004 23:24:11 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

:>If size does NOT matter , would you still go CQ? I'd definitely go with CBR
:>max bitrate. What is disatvantage of that as far as quality of the video goes?

: Encoding CBR at the maximum bitrate, and encoding CQ at the maximum
: quality, for full D1 resolution, would bring the same results, because

That's my point. I try to make DVDRs not more than 60 minutes long. That
brings maximum quality.

--Leonid

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

On 6 May 2004 02:39:08 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>That's my point. I try to make DVDRs not more than 60 minutes long. That
>brings maximum quality.

Even then, you can still get some advantage from using CQ mode, if you
intend something like 8500 average, but allowing, say, 9000 kbps
peaks. That gives you a small margin for variable bitrate, which you
can use either with CQ encoding (toilsome), or with the VBR one. If
you set a maximum which is not to be surpassed in any case, and you
use D1 resolution, then CBR is more than enough.

 

[Guest]

Archived from groups: rec.video.desktop (More info?)

Bariloche <bariloche@bariloche.com> wrote:
: Even then, you can still get some advantage from using CQ mode, if you
: intend something like 8500 average, but allowing, say, 9000 kbps
: peaks. That gives you a small margin for variable bitrate, which you
: can use either with CQ encoding (toilsome), or with the VBR one. If
: you set a maximum which is not to be surpassed in any case, and you
: use D1 resolution, then CBR is more than enough.

I usually don't go beyond 8250kbps because either my sound is uncompressed so
that total bitrate may exceed 9800kbps OR I'm just afraid that my DVD player is
going to choke on anything higher than 8300kbps.

The latter is due to the fact that some guy recorded something on stand alone
DVD player in XP mode (whatever gradation this player has) and my DVD player
didn't play it very well. It would play some parts slower than would speed up.
I looked at the bitrate later on and noticed that bitrate was going beyond
9800kbps.

--Leonid