[x264-devel] [Mohsen Shaaban <mmm5554 at cacs.louisiana.edu>] RE: Re: Using I-frame instead of IDR

[Lists administration]

----- Forwarded message from Mohsen Shaaban <mmm5554 at cacs.louisiana.edu> -----

From: Mohsen Shaaban <mmm5554 at cacs.louisiana.edu>
To: x264-devel at videolan.org
Subject: RE: [x264-devel] Re: Using I-frame instead of IDR
Date: Thu, 16 Nov 2006 12:51:31 -0600
X-Mailer: Microsoft Office Outlook 11


The deleted attachment is at:
    <http://lists.videolan.org/attachs/20061117/frequency-domain.pdf>
 

Hello Loren,

First, I'm a member of x264-devel but I do not know how to post questions
from within the x264-devel mailing list. Can you please tell me how to do
that?

Second, I have a question about H.264 transcoding from MPEG-2 bitstream, I
hope you can help me.

I want to do some experiments on the MPEG-2 bitstream before Transcoding, so
I want to see the effect of requantization error on the stream and I read in
one paper that you can do that by having a fixed triplet quantization
scaling (fixed scaling value Qi, Qp, Qb) and my question is there any
feature in MPEG-2 TM5 reference software that you can change to do that or
one have to change the rate control algorithm in there?

I attaching the paper for your reference (experiment shows in fig. 3 in the
paper)

I hope you have an answer or a guide for me to follow.
I appreciate it and thank you.

Mohsen Shaaban
PhD Student in Computer Engineering
The Center for Advanced Computer Studies
University of Louisiana at Lafayette

-----Original Message-----
From: x264-devel-bounce at videolan.org [mailto:x264-devel-bounce at videolan.org]
On Behalf Of Loren Merritt
Sent: Wednesday, November 15, 2006 10:42 PM
To: x264-devel at videolan.org
Subject: [x264-devel] Re: Using I-frame instead of IDR

On Wed, 15 Nov 2006, Axel Gunter wrote:
> Loren Merritt wrote:
>> On Wed, 15 Nov 2006, Axel Gunter wrote:
>> 
>>> I noticed all the I-frames introduced by the x264 code were IDR Access 
>>> Units. Is there a way I can have non-IDR I-frames in x264 encoding?
>> 
>> No. Why do you want non-IDR I-frames? The only reason I can think of is 
>> open-GOP B-frames, which x264 also doesn't do.
>
> I guess the compression would be better if the frames following the IDR or

> I-picture could use frames temporally before the IDR/I-picture since they 
> have more reference frames to chose from. Since IDR forces a decoder 
> reference buffer flush, this isn't possible. Correct me if I am wrong.

In that case, look at --min-keyint. If a scenecut is too close to the 
previous IDR (default: 25 frames) it gets encoded as a non-IDR I-frame. 
Note that this is not optimal, but the metric that would really optimize 
for compression ratio isn't possible to measure in 1 pass, and I haven't 
written a 2 pass scenecut/IDR algorithm yet.

--Loren Merritt

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html


----- End forwarded message -----

-- 
Lists administration <videolan at lists.videolan.org>

-- 
This is the x264-devel mailing-list
To unsubscribe, go to: http://developers.videolan.org/lists.html