WMPTalk.com

Michael Chen · Guest

Hi, anyone from the Windows Media team,

I want a 1-pass quality-based VBR video encoder with a peak bit rate. This
encoding mode is the hybrid of quality-based VBR and constrained VBR where
one can specify a quality % number as well as a maximum bit rate. The
encoder produces a VBR stream where both parameters are guaranteed. The
main advantages are: 1) the single pass encoding; 2) the output stream will
not exceed the specified bit rate, which is essential for playback on non-PC
devices such as consumer products.

You can call it "Quality VBR (Peak)", or "Constrained Quality-base VBR".
This will be a natual complement to VBR modes currently offered: "Bit rate
VBR", "Bit rate VBR (Peak)" and "Quality-base VBR".

In the MPEG-2 world, the equivalent of "Quality VBR (Peak)" can be found in
one of the most popular encoder TMPGEnc. It allows you to create very good
DVD quality MPEG-2 video stream in a single pass.

Since the WMV format is Microsoft's proprietary close-source world
domination secret standard, this feature can only come from Microsoft. I
certainly hope you guys are reading.

Thank you

--Michael

Zarax [Digital Media MVP] · Guest

Ever tried Windows Media Encoder?
If yes try looking in the compression tab (it requires a custom session),
and you can find exactly what you are looking for, even from "the evil world
domination standard"...

"Michael Chen" wrote:

Michael Chen · Guest

Zarax,

I am afraid you are quite wrong. I know Windows Media Encoder well. What
it has is "Quality-base VBR", which has no upper bit rate limit. What I
want is "Quality-base VBR (Peak)" with a upper bit rate limit just like "Bit
rate VBR (Peak)". The only problem with the latter is that it is always
2-pass (see read my original post). So my demand is still on.

--Michael

"Zarax [Digital Media MVP]" <ZaraxDigitalMediaMVP@discussions.microsoft.com>
wrote in message news:A87345C7-614F-40D1-BA37-C7B71C82A209@microsoft.com...

Michael Chen · Guest

That is why this is not a question but a feature request, which I hope
Microsoft is watching. Besides, I don't have to elaborate how easy it is to
implement. For those who encode for hardware devices, this feature no doubt
beats 2-pass Bit Rate VBR (Peak) mode. Don't you agree, Zarax?

--Michael

"Zarax [Digital Media MVP]" <ZaraxDigitalMediaMVP@discussions.microsoft.com>
wrote in message news:6A8A12D7-1794-49CA-8CA3-0E83FF6515B1@microsoft.com...

Zarax [Digital Media MVP] · Guest

I'm not sure there are real advantages over the classical peak VBR except of
course speed, however it would be interesting to see how the encoder would
achieve peak VBR without analyzing the content first...

"Michael Chen" wrote:

Michael Chen · Guest

Zarax, it has been done. Refer to my original post.

--Michael

"Zarax [Digital Media MVP]" <ZaraxDigitalMediaMVP@discussions.microsoft.com>
wrote in message news:CABCE4E9-42BE-4194-A1A3-D3786492A3DA@microsoft.com...

Manila Dell (MSUN) · Guest

Hi, I understand that your need of having 1-pass quality-based VBR video
with a peak bitrate. However I am afraid that the encoder will be able to
produce a VBR stream where *both* parameters are guaranteed. For example,
consider the following scenario: During an encoding session you choose
1-pass VBR quality 90% and you specify the peak bitrate to be 1500 Kbps. You
press encode.
The encoder come across a video frame and determines it requires 2730 Kbps
to satisfy the Quality 90% condition. Since the peak bit rate is set to 1500
Kbps, the encoder chooses 1500 Kbps. Now Peak bitrate condition is satisfied
but the Quality 90% is *not*.

"Michael Chen" <michaelc7@earthlink.net> wrote in message
news:Bu10d.11490$w%6.9193@newsread1.news.pas.earthlink.net...

Michael Chen · Guest

Hi, you meant to say, "... the encoder wil *not* be able to produce a VBR
stream
where *both* parameters are guaranteed." I totally understand your
analysis,
and I am okay with peak bitrate taking precedence over quality. The reason
is
quite simple -- to satisfy hardware limitation.

This is the same reason why one would use the "Bit rate VBR (Peak)" mode.
I am merely requesting a feature that can achieve the same goal in 1-pass.

Does anyone from Microsoft care to comment on this thread?

Thanks

--Michael

"Manila Dell (MSUN)" <mcored@iinet.net.au.remove> wrote in message
news:#9vDECwmEHA.1716@TK2MSFTNGP12.phx.gbl...

fred.t · Guest

Michael,

What you are requesting makes perfect sense to me. I produce wedding videos
on DVD format, and many of my customers have el cheapo DVD players with peak
data throughput of about 8Mbps and a read-ahead buffer of about five seconds.
When the read ahead buffer runs empty, playback is paused until the buffer
has been filled again. I know this is below spec, but what can I do?

However it has occured to me that the best encoding method would be an
average VBR of 7Mbps with a constrained peak of 8Mbps. The quality % is not
important, often 7Mpbs is enough for 99% quality. But when the scene gets
"busy" (think about confetti falling on people) then obviously quality has to
be sacrificed in order to prevent stutteriing playback, in other words to
remain within the peak throughput of the playback device.

And obviously if all this can be done without a second pass, so much the
better (my time is worth money after all).

So here's the feature request:-
1) Ask the average desired throughput (bps) OR desired quality %
2) Ask the peak throughput (P bps) AND read-ahead buffer size (RA seconds)
3) Start encoding at the average rate.
4) Just before each new keyframe is started, check if the total bits that
have been output for the last RA seconds averages out to more than the
constrained limit P, if not then proceed. But if P was exceeded, then lower
the quality slightly, skip back RA seconds and encode that portion again
(mini local second pass).
5) Go back to step 3 unless end of clip hasbeen reached.

The above process will take longer than a single pass but not as much as a
two pass process. The more "busy" scenes you have, the closer the time will
be to a full two pass process, obviously.

I hope to see this in WMA 10.1!

-Fred

"Michael Chen" wrote:

Manila Dell (MSUN) · Guest

Yes Michael. I meant to say I am afraid that the encoder will *not* be able
to produce a VBR stream where *both* parameters are guaranteed. fred.t
pointed out some practical usage of the new proposed codec, thanks for that
explanation.
--
Cheers,
Manila

"Michael Chen" <michaelc7@earthlink.net> wrote in message
news:lQ52d.1280$0i5.675@newsread3.news.pas.earthlink.net...

Manila Dell (MSUN) · Guest

Extracted from WME Help:

1. With CBR encoding, you specify the bit rate that you want to maintain and
then set the size of the buffer. *The bit rate will fluctuate across the
stream*; however, the fluctuations are constrained by the buffer size.
2. CBR encoding is designed to work most effectively in a streaming
scenario. With CBR encoding, the bit rate remains *fairly constant* and
close to the target bit rate over the course of the stream, within a small
window of time set by the buffer size.

I tested encoding couple of video sources to 1-pass CBR WMV and the Bit Rate
(as mentioned in the Help) did fluctuate depending on the complexity of the
frame. My target bit rate was 962 Kbps (during one of the tests) and the bit
rate once lowered down to 100 Kbps for some still frames. This satisfied
your requirement of having 1-pass VBR support.
It did not go over the target bit rate (well, slightly during some complex
frames) this satisfied your requirement of having a Maximum Bit Rate.

The advantages you listed: 1) the single pass encoding and 2) the output
stream will not exceed the specified bit rate were met.

--
Cheers,
Manila

"Michael Chen" <michaelc7@earthlink.net> wrote in message
news:Bu10d.11490$w%6.9193@newsread1.news.pas.earthlink.net...

	WMPTalk.com Forum Index -> Encoders	All times are GMT
Page 1 of 1