Re: [rtcweb] Performance of H.264...

Stephan Wenger <> Wed, 20 November 2013 19:40 UTC

Return-Path: <>
Received: from localhost ( []) by (Postfix) with ESMTP id 2E3B61AE055 for <>; Wed, 20 Nov 2013 11:40:28 -0800 (PST)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -2.6
X-Spam-Status: No, score=-2.6 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_LOW=-0.7, SPF_PASS=-0.001] autolearn=ham
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id 1QsPcy8VQpVw for <>; Wed, 20 Nov 2013 11:40:25 -0800 (PST)
Received: from ( []) by (Postfix) with ESMTP id 3C7EA1AE162 for <>; Wed, 20 Nov 2013 11:40:24 -0800 (PST)
Received: from ( by ( with Microsoft SMTP Server (TLS) id 15.0.820.5; Wed, 20 Nov 2013 19:40:16 +0000
Received: from ([]) by ([]) with mapi id 15.00.0820.005; Wed, 20 Nov 2013 19:40:15 +0000
From: Stephan Wenger <>
To: Bernard Aboba <>, Maik Merten <>, "" <>
Thread-Topic: [rtcweb] Performance of H.264...
Thread-Index: AQHO5hk+oT4X9Wzm8UyORPNQmBT585oue4QAgAAESoD//36IAA==
Date: Wed, 20 Nov 2013 19:40:14 +0000
Message-ID: <>
In-Reply-To: <BLU169-W137AFD6B8AE1D1A206D67EC93E60@phx.gbl>
Accept-Language: en-US
Content-Language: en-US
x-originating-ip: []
x-forefront-prvs: 0036736630
x-forefront-antispam-report: SFV:NSPM; SFS:(199002)(189002)(2656002)(80022001)(81342001)(66066001)(47976001)(51856001)(87266001)(87936001)(4396001)(50986001)(49866001)(16236675002)(47736001)(65816001)(53806001)(54356001)(46102001)(77096001)(56816003)(83072001)(63696002)(85306002)(15975445006)(19580395003)(74876001)(74706001)(76786001)(54316002)(83322001)(36756003)(80976001)(19580405001)(77982001)(59766001)(74662001)(47446002)(81542001)(74366001)(81686001)(56776001)(76482001)(76796001)(69226001)(79102001)(76176001)(31966008)(81816001)(74502001)(42262001); DIR:OUT; SFP:; SCL:1; SRVR:CO1PR07MB362;; CLIP:; FPR:; RD:InfoNoRecords; A:0; MX:1; LANG:en;
Content-Type: multipart/alternative; boundary="_000_CEB277B3AA931stewesteweorg_"
MIME-Version: 1.0
Subject: Re: [rtcweb] Performance of H.264...
X-Mailman-Version: 2.1.15
Precedence: list
List-Id: Real-Time Communication in WEB-browsers working group list <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Wed, 20 Nov 2013 19:40:28 -0000

From: Bernard Aboba <<>>
Date: Wednesday, 20 November, 2013 at 14:23
To: Maik Merten <<>>, "<>" <<>>
Subject: Re: [rtcweb] Performance of H.264...

>> There's much to be said about PSNR as means for comparing image quality.

>[BA] Section III A states that the comparison really doesn't relate to interactive/low delay uses (e.g. use of B frames, etc.) which is what we'd care about on this mailing list.   But it's the first study of its kind so it seemed worth posting a link to (in hopes that someone else might have a link to something more relevant).

Ask, and it shall be given you; seek, and ye shall find...

See here:!769&app=WordPdf&authkey=!AP15CciU4zPcTHc

This paper also compares codec operation points relevant for entertainment, and does not focus on video conferencing (without I pictures).  However, since it contains data for both random access GOP structures and pure intra, one could extrapolate pure predicted picture performance and arrive some conclusion (which is left as an exercise for the reader :-)
Note that the use of B pictures is uncorrelated with the use in low delay environments, and has been since at least 2003.  (In H.264 and later standards, both predictors of B pictures can point to the past.)  The use of B pictures nowadays is almost exclusively a complexity question.