Variable bitrate at different parts of a footage...

G

Guest

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

Simple dumb question.

I have a footage in form of Huffyuv AVI file. It's 80 minute long. I want to
fit it on a DVD with uncompressed sound. It wouldn't fit at maximum bitrate
of 8250kbs for video. Obviously not. I tried to compress the sound with
toolame to .mp2 at 384kbs, but wasn't too happy with the sound. I also wasn't
too happy with the video when I dropped an average bitrate to 6000kbs. Then
something striked me. Some parts of the footage can be encoded with like 4000
kbs and I can increase the more important parts. So the question is:
Is there any way to tell TMPGEnc to encode at one bitrate from time point A
to time point B and at the other bit rate from time point B to time point C.

Another thing. I have the stereo WAV file. I was thinking if I could maybe
compress it to Dolby Digital 2.0 and use that instead of MPEG-2 compression.
How do I do that? And will TMPGenc DVD Author take Dolby Digital 2.0?

Thanks.

--Leonid
 
G

Guest

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

On 15 May 2004 02:51:04 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>Simple dumb question.
>
>I have a footage in form of Huffyuv AVI file. It's 80 minute long. I want
to
>fit it on a DVD with uncompressed sound. It wouldn't fit at maximum bitrate
>of 8250kbs for video. Obviously not. I tried to compress the sound with
>toolame to .mp2 at 384kbs, but wasn't too happy with the sound. I also wasn't
>too happy with the video when I dropped an average bitrate to 6000kbs. Then
>something striked me. Some parts of the footage can be encoded with like 4000
>kbs and I can increase the more important parts. So the question is:
>Is there any way to tell TMPGEnc to encode at one bitrate from time point
A
>to time point B and at the other bit rate from time point B to time point
C.

I suppose you could use something like VirtualDUB to chop the AVI up into
pieces that correspond to the various time points. Then encode the
various pieces at the various bitrates and then MERGE the pieces.

80 minutes of footage can be encoded at approximately 7300Kb/s MPEG-2
video and 192Kb/s Dolby Digital (AC-3) audio and still fit on a 4.38GB
DVD-R. There's not a whole lot of footage that doesn't encode well at
such a high bitrate. Going up to 8000Kb/s, you run the risk of some
players not being able to read DVD-R fast enough (not a problem with
replicated discs).

>Another thing. I have the stereo WAV file. I was thinking if I could maybe

>compress it to Dolby Digital 2.0 and use that instead of MPEG-2 compression.
>How do I do that?

Get a Dolby Digital encoder, such as Sonic Foundry's SoftEncode, Surcode's
Dolby Digital Encoders, Sony's Vegas+DVD, etc.

>And will TMPGenc DVD Author take Dolby Digital 2.0?

I suppose since it has an (optional) Dolby Digital encoder plug-in, that
it should.
 
G

Guest

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

On Sat, 15 May 2004 07:31:03 +0200 (CEST), Nomen Nescio
<nobody@dizum.com> wrote:

>I suppose you could use something like VirtualDUB to chop the AVI up into
>pieces that correspond to the various time points. Then encode the
>various pieces at the various bitrates and then MERGE the pieces.

Tmpgenc has Manual VBR mode, and "Force picture type setting" (vs
"Detect Scene Change") which I understand allows you to use different
bitrates for different scenes. But I have never used it, so don't ask
me. And chopping the AVI looks a simpler way to accomplish the same.
 
G

Guest

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

Nomen Nescio <nobody@dizum.com> wrote:
: I suppose you could use something like VirtualDUB to chop the AVI up into
: pieces that correspond to the various time points. Then encode the
: various pieces at the various bitrates and then MERGE the pieces.

Yeah, that was my initial idea, but I thought there was a way to simplify the
process. Thanks though.

: 80 minutes of footage can be encoded at approximately 7300Kb/s MPEG-2
: video and 192Kb/s Dolby Digital (AC-3) audio and still fit on a 4.38GB
: DVD-R. There's not a whole lot of footage that doesn't encode well at
: such a high bitrate.

My footage is taken from LaserDisc - so the quality is really really good.

: Going up to 8000Kb/s, you run the risk of some
: players not being able to read DVD-R fast enough (not a problem with
: replicated discs).

What do you mean replicated disc?

: Get a Dolby Digital encoder, such as Sonic Foundry's SoftEncode, Surcode's
: Dolby Digital Encoders, Sony's Vegas+DVD, etc.

In general does Dolby Digital encoder give a better quality than MPEG-2 sound
encoded with tooLame?

:>And will TMPGenc DVD Author take Dolby Digital 2.0?

: I suppose since it has an (optional) Dolby Digital encoder plug-in, that
: it should.

I didn't see optional DD encoder plug-in in TMPGenc. Should I download that
plug-in?

--Leonid
 
G

Guest

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

Bariloche <bariloche@bariloche.com> wrote:
: Tmpgenc has Manual VBR mode, and "Force picture type setting" (vs
: "Detect Scene Change") which I understand allows you to use different
: bitrates for different scenes. But I have never used it, so don't ask

"Force picture type setting" allows you to insert I-frames at specific points,
but doesn't allow you to choose a bit rate.

: me. And chopping the AVI looks a simpler way to accomplish the same.

Yeah, I guess that's the only way to go. I'm also a bit afraid that when
I put pieces back together there might be some sound clicks.

--Leonid
 
G

Guest

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

> of 8250kbs for video. Obviously not. I tried to compress the sound
with
> toolame to .mp2 at 384kbs, but wasn't too happy with the sound. I also
wasn't

That's very interesting. mp2@384 _should_ be indistinguishable from the
original wav file, even 224. What are you hearing that it making you
unhappy. Sibilance, flanging... how does it sound different?
 
G

Guest

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

On 15 May 2004 11:48:09 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>Nomen Nescio <nobody@dizum.com> wrote:
>
>: 80 minutes of footage can be encoded at approximately 7300Kb/s MPEG-2
>: video and 192Kb/s Dolby Digital (AC-3) audio and still fit on a 4.38GB
>: DVD-R. There's not a whole lot of footage that doesn't encode well at
>: such a high bitrate.
>
>My footage is taken from LaserDisc - so the quality is really really good.

If the quality is "really good", then it should encode well using the
above parameters unless your MPEG encoder is garbage. TMPGEnc, while dirt
slow, produces excellent quality.

I've captured several laserdiscs via my Canopus ADVC-100 and converted
them to DVD-Video with good results.

I generally capture the disc and edit out the unnecessary parts. I then
determine the length of the edited footage and pump it into a bitrate
calculator such as Bearson's BitRate Calculator.

Let's say I have 95 minutes of source material. I enter 1 hr and 35
minutes as the length. I then enter the final filesize (~4420MB for a
4.38GB DVD-R) and the audio bitrate as 192Kb/s (standard 2-channel Dolby
Digital bitrate). That allows me to encode the video at around 6160Kb/s.
I will use something like Canopus' ProCoder at 2-pass VBR with the
average bitrate being the above with the maximum bitrate at 8Mb/s (the
possible 8Mb/s "spikes" described below haven't caused me problems). If
the footage is so long that the bitrate gets reduced to an uncomfortablty
low level (<3Mb/s), then I will encode to half-D1 or just split in into
multiple discs.

>: Going up to 8000Kb/s, you run the risk of some players not being able
>: to read DVD-R fast enough (not a problem with replicated discs).
>
>What do you mean replicated disc?

<http://www.mediacopy.co.uk/cd_cdvcdr.htm>
<http://www.absolute-disc.com/cd-duplication.html>
<http://www.proactionmedia.com/replication_basics.htm>

"Replicated" discs should be able to be encoded to the maximum allowable
bitrate per the DVD-Video standard. "Duplicated" discs may have problems
playing in standalone players when the overall bitrate 7-8+Mb/s.

>In general does Dolby Digital encoder give a better quality than MPEG-2 sound
>encoded with tooLame?

At the same bitrate, I think unquestionably it does.

>I didn't see optional DD encoder plug-in in TMPGenc. Should I download that
>plug-in?

I would. It's probably the cheapest legitimate way for you to get a Dolby
Digital (AC-3) encoder.
 
G

Guest

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

Morrmar <morrmar@myway.com-no spam> wrote:
: That's very interesting. mp2@384 _should_ be indistinguishable from the
: original wav file, even 224. What are you hearing that it making you
: unhappy. Sibilance, flanging... how does it sound different?

I do hear a tiny difference. A little sort of distortion. Tiny, but still
noticeable if I listen carefully. I wouldn't care if the source was VHS. But
the source was an LD and I was capturing the sound using S/PDIF out from LD
into my soundcard - so the sound has no loss and I'd like to keep it that way.

--Leonid
 
G

Guest

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

Nomen Nescio <nobody@dizum.com> wrote:
: If the quality is "really good", then it should encode well using the
: above parameters unless your MPEG encoder is garbage. TMPGEnc, while dirt
: slow, produces excellent quality.

I use TMPGenc.

: <http://www.mediacopy.co.uk/cd_cdvcdr.htm>
: <http://www.absolute-disc.com/cd-duplication.html>
: <http://www.proactionmedia.com/replication_basics.htm>

: "Replicated" discs should be able to be encoded to the maximum allowable
: bitrate per the DVD-Video standard. "Duplicated" discs may have problems
: playing in standalone players when the overall bitrate 7-8+Mb/s.

I've never heard of it before. So all the DVDRs are duplicated and thus I
shouldn't be using more than 8000kbs, correct? I wish I knew that before :)

If that's the case, the problem is basically solved. I can just go with like
7500kbs maximum and I'll be able to fit it pretty well.

--Leonid
 
G

Guest

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

On 15 May 2004 11:50:38 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>"Force picture type setting" allows you to insert I-frames at specific points,
>but doesn't allow you to choose a bit rate.

This is what the help hint of the program say about Manual VBR (MVBR):
"This is fixed bitrate which enable to set bitrate for each scene.
Bitrate setting is done at force picture setting." But as I have never
used that mode, that's all I know about it.

>I'm also a bit afraid that when
>I put pieces back together there might be some sound clicks.

1) Export the audio as Wav.
2) Encode the Wav as Ac3 at 192 kbps (224 or 256 kbps if you really
care for it).
2) Cut the video in pieces, and encode each of them.
3) Join the encoded video pieces.
4) Mux the Ac3 audio.
 
G

Guest

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

Bariloche <bariloche@bariloche.com> wrote:
: 1) Export the audio as Wav.
: 2) Encode the Wav as Ac3 at 192 kbps (224 or 256 kbps if you really
: care for it).
: 2) Cut the video in pieces, and encode each of them.
: 3) Join the encoded video pieces.
: 4) Mux the Ac3 audio.

Actually you gave me a good idea. I thought cutting audio and video and then
join them all in the authoring program. But I guess I can just cut video in
pieces and join them in TMPGEnc. Thanks.

--Leonid
 
G

Guest

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

Nomen Nescio <nobody@dizum.com> wrote:
: Well, here's the story:

Thanks for shding the light.

: At full D1, with clean source material and an excellent MPEG encoder, it's

I use TMPGenc and it indeed does an excellent job. However, it doesn't do a
good job on this particular footage that I recorded from a LaserDisc and
previously from VHS. This is a live concert and there're some effects like
changing cameras every split second. Also there's a smoke on the stage and
during that time I see a lot of pixalization and blockiness. Even 8000mbs
chokes on such scenes.

: commercial DVDs in a software player that shows the audio/video bitrate
: (such as PowerDVD XP v5.0) or Bitrate Viewer -- I rarely see the video
: bitrate go over 7Mb/s.

I know. Strangely enough their quality is better than higher bit rate going
from VHS.

--Leonid
 
G

Guest

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

On 16 May 2004 19:50:19 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>Nomen Nescio <nobody@dizum.com> wrote:
>
>I use TMPGenc and it indeed does an excellent job. However, it doesn't do
a
>good job on this particular footage that I recorded from a LaserDisc and
>previously from VHS. This is a live concert and there're some effects like

>changing cameras every split second. Also there's a smoke on the stage and
>during that time I see a lot of pixalization and blockiness. Even 8000mbs

>chokes on such scenes.

I would consider, as an experiment, dropping the resolution to half-D1
while keeping the bitrate the same as above.

>: commercial DVDs in a software player that shows the audio/video bitrate
>: (such as PowerDVD XP v5.0) or Bitrate Viewer -- I rarely see the video
>: bitrate go over 7Mb/s.
>
>I know. Strangely enough their quality is better than higher bit rate going
>from VHS.

The luxury of using hardware based MPEG encoders such as those by Sonic,
OptiBase or Sony.
 
G

Guest

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

Nomen Nescio <nobody@dizum.com> wrote:
: I would consider, as an experiment, dropping the resolution to half-D1
: while keeping the bitrate the same as above.

I've done this in the past (with a different footage though). The theory that
if I reduce to half D1 with twice as little bitrate is not working. I tried
352x480 footage with 4000kbs and sure enough there was a lot of pixalization.

One other thing is that I modified the TMPGenc original DVD templates. The
standard GOP structure is:
IBBPBBPBBPBBPBBPBB.

I have it customized as IBPBPBPBPB. Do you think this could cause the
pixalization problems?

One other thing I thought I could save a bit rate is that my source video is
letterboxed (widescreen within 4:3). If I crop (clip) the image in TMPGEnc,
will bits NOT be allocated outside of the clipping area while encoding?

:>I know. Strangely enough their quality is better than higher bit rate going
:>from VHS.

: The luxury of using hardware based MPEG encoders such as those by Sonic,
: OptiBase or Sony.

The software *postprocessing* compression theoretically should give a better
quality than hardware compression on the fly, shouldn't it?

--Leonid
 
G

Guest

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

On 18 May 2004 00:03:34 GMT, Leonid Makarovsky <venom@cs.bu.edu> wrote:

>Nomen Nescio <nobody@dizum.com> wrote:
>: I would consider, as an experiment, dropping the resolution to half-D1
>: while keeping the bitrate the same as above.
>
>I've done this in the past (with a different footage though). The theory that
>if I reduce to half D1 with twice as little bitrate is not working. I tried
>352x480 footage with 4000kbs and sure enough there was a lot of pixalization.

Try encoding at half-D1 (352x480) res at **8000Kb/s** and check for
pixellation is what I meant.

>One other thing is that I modified the TMPGenc original DVD templates. The
>standard GOP structure is:
>IBBPBBPBBPBBPBBPBB.
>
>I have it customized as IBPBPBPBPB. Do you think this could cause the
>pixalization problems?

Sure, since B frames generally require less bits than P frames. Your
modified GOP could be a "bandwidth hog". In hard to encode scenes, which
your source footage seems to have a lot of, that GOP could be "running out
of bits" and cause your complaints.

Try standard, or rather default, GOP settings for TMPGEnc's MPEG-2
encoding for DVD-Video.

Most MPEG encoder documentation I have seen has strongly encouraged users
to NOT modify the GOP structure for encoding that comes with the templates
unless there is a darn good reason.

>One other thing I thought I could save a bit rate is that my source video
is
>letterboxed (widescreen within 4:3). If I crop (clip) the image in TMPGEnc,
>will bits NOT be allocated outside of the clipping area while encoding?

Sure. That could help too.
 
G

Guest

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

Nomen Nescio <nobody@dizum.com> wrote:
: Try encoding at half-D1 (352x480) res at **8000Kb/s** and check for
: pixellation is what I meant.


You know what, I'll actually be encoding at half-D1. I noticed that these
2 LDs were rather noisy and have some sort of their own pixelization.

:>One other thing is that I modified the TMPGenc original DVD templates. The
:>standard GOP structure is:
:>IBBPBBPBBPBBPBBPBB.
:>
:>I have it customized as IBPBPBPBPB. Do you think this could cause the
:>pixalization problems?

: Sure, since B frames generally require less bits than P frames. Your
: modified GOP could be a "bandwidth hog". In hard to encode scenes, which
: your source footage seems to have a lot of, that GOP could be "running out
: of bits" and cause your complaints.

: Try standard, or rather default, GOP settings for TMPGEnc's MPEG-2
: encoding for DVD-Video.

I'll try that. Thanks.

BTW, I think I'll be using AVISynth to resize.
AVISource("Coroner_LiveInEastBerlin.avi")
Trim(215,2015)
Lanczos4Resize(352,480)

Now, do I need SeparateFields command before Lanczos4Resize?

--Leonid
 
G

Guest

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

On 18 May 2004 02:45:22 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>Nomen Nescio <nobody@dizum.com> wrote:
>: Try encoding at half-D1 (352x480) res at **8000Kb/s** and check for
>: pixellation is what I meant.
>
>You know what, I'll actually be encoding at half-D1. I noticed that these
>2 LDs were rather noisy and have some sort of their own pixelization.

Rather, encode half-D1 at CQ 100%. That way, you have the guarantee
that encoding shall not develop any artifact. If they show, then they
must already come with the original video.

Half-D1 at CQ 100% is almost always quite below 8000 kbps.
 
G

Guest

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

Bariloche <bariloche@bariloche.com> wrote:
: Half-D1 at CQ 100% is almost always quite below 8000 kbps.

I just encoded it with CQ 90% max 8250kbs. The average bitrate was 5836.50kbs.

I'd like to boost it up but not more than like 7000-7500kbs. Will 100% do it?

--Leonid
 
G

Guest

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

On 18 May 2004 00:03:34 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>One other thing I thought I could save a bit rate is that my source video is
>letterboxed (widescreen within 4:3). If I crop (clip) the image in TMPGEnc,
>will bits NOT be allocated outside of the clipping area while encoding?

Indeed.
 
G

Guest

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

On 18 May 2004 12:23:46 GMT, Leonid Makarovsky <venom@cs.bu.edu>
wrote:

>I just encoded it with CQ 90% max 8250kbs. The average bitrate was 5836.50kbs.
>
>I'd like to boost it up but not more than like 7000-7500kbs. Will 100% do it?

I use to encode half-D1 CQ 100%, and it never went above 7500 kbps. I
cannot speak for your material, though.

You can make a test with Avisynth's SelectRangeEvery(g.n, g), where g
would be the number of frames in your GOP, and g.n a multiple thereof.
That way, you can encode only a percentage of the movie (100/n %), and
see what average bitrate you get from the .log file