[AVT] Re: Question about "PayloadSize" field in MPEG-4 RTP payload format

Dear Phillipe and all,

I see. But the description is not clear enough. Would you revise the definition of PayloadSize, especially the second paragraph, with the text in your e-mail for the clarification?

Sincerely,
Young
----- Original Message ----- 
From: <philippe.gentric@philips.com>
To: <young@techway.co.kr>
Cc: <4on2andIP-sys@advent.ee.columbia.edu>; <avt@ietf.org>
Sent: Wednesday, August 22, 2001 9:50 PM
Subject: Re: Question about "PayloadSize" field in MPEG-4 RTP payload format

> 
> 
> 
> it is not a singleSL *MODE* issue,
> 
> the absolute rule is that the presence of PayloadSize
> (SizeLength != 0) is a total equivalence
> (mathematical symbol <=> ;-)
> with multipleSL mode
> 
> i.e.
> Single SL mode == NO PayloadSize field (never)
> Multiple SL mode == PayloadSize field (always)
> 
> HOWEVER
> 
> it may happen that in the multipleSL mode
> you have an Access Unit so large that
> it is actually larger than the MTU,
> then you will split this large AU in
> several SL packet and therefore
> transport each of these SL packets
> in one RTP packet.
> 
> ***********
> so this is a "single SL packet" but the MODE
> REMAINS multipleSL (the mode cannot change
> inside a session anyway)
> ***********
> 
> in that case since the mode is multiple
> you HAVE the PayloadSize field.
> 
> It would be stupid to waste this field by putting
> in this field the (single) SL packet size
> since UDP already gives you that !
> 
> moreover it is very helpfull to give (repeat)
> the *complete AU size* in this PayloadSIze field.
> it helps a receuver finding out if
> a complete AU has arrived or not in
> case of packet losses .
> 
> *******
> NOTE:
> 
> Actually in case this "single" SL packet
> in the RTP packet is actually a
> complete Access Unit the
> packet structure is
> exactly the same as previoulsly !
> 
> NOTE (2)
> 
> of course if you have such a stream
> (1 AU >= 1 RTP packet)
> you have NOW a choice:
> 
> (1) use the SingleSL (default) mode because
> it saves SizeLength bits per packet !
> 
> (2) use the MultipleSL mode because
> it makes AU reconstrcution easier !
> 
> but for best efficiency
> of AU reconstruction strategy
> I would recommend the usage
> of AUstartFalg and AUendFlag:
> only 2 bits per packet ...
> you have to be a Synch Layer
> fan to do that though ;-)
> 
> 
> *****
> in conclusion it is not a mode issue
> it is a "smart" exception for
> single SL transport in the MultilpleSL mode
> 
> 
> is that more clear ?
> 
> Regards,
> 
> Philippe Gentric
> Software architect
> Philips Digital Networks - MP4Net
> 51 rue Carnot B.P. 301
> 92156 Suresnes FRANCE
> tel: +33(0)147283740
> fax: +33(0)147283725
> philippe.gentric@philips.com
> http://www.mpeg-4.philips.com
> 
> 
> 
> 
> 
> "Lim, Young-Kwon" <young@techway.co.kr> on 21-08-2001 16:23:08
> 
> Please respond to "Lim, Young-Kwon" <young@techway.co.kr>
> 
> To:     Philippe Gentric/LIM/CE/PHILIPS@EMEA1
> cc:     <4on2andIP-sys@advent.ee.columbia.edu>
> Subject:  Question about "PayloadSize" field
> Classification:
> 
> 
> Dear Phillipe and all,
> 
> In the latest draft, "PayloadSize" field is defined to contain the entire size of AU in single SL case. But the parameter "SizeLength" will set to zero in default and it indicates the single SL case. This means that the length of "PayloadSize" field will
> be zero in single SL case and conflict with the definition of "PayoadSize" field. Did I missed something in between?
> 
> Sincerely,
> Young.
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 
> 

_______________________________________________
Audio/Video Transport Working Group
avt@ietf.org
http://www1.ietf.org/mailman/listinfo/avt