[Live-devel] Frame granularity

[Matthew Hodgson]

Hi all,

I'm trying to write a custom MultiFramedRTPSource and Sink (and Framer?) 
for a video payload format - but i'm slightly confused about what 
granularity of frames I should be aiming for.  Should the Source yield 
frames which refer to a discrete picture of video, or instead whatever 
sub-picture fragment the individual received packets represent (e.g. 
picture, GOB, slice or macroblocks)?

If Source frames represent discrete pictures combined from several packets 
(as H263plusVideoSource appears to), each with a set of per-packet 
specialHeaders, should those specialHeaders be used in the Sink (when 
doing a simple relay) to construct the output packets?  How would one tell 
the extended MultiFramedRTPSink to only fragment the frame on the 
boundaries implied by that frame's specialHeaders, rather than MTU-based 
fragmentation?

Alternatively, is the intention for a Framer to perform this operation - 
fragmenting the discrete picture frames into per-packet frames suitable 
for the Sink (assuming the Sink then never performs any fragmentation to 
avoid exceeding the MTU)?  If so, is the Framer allowed to sniff 
specialHeaders from the Source?

I'm trying to minimise latency when doing a simple relay by not waiting to 
combine incoming packets into discrete picture frames in the Sink - 
although I want to use the RTPSource/Sink framework rather than working at 
a blunt UDP level so as to allow more interesting Filters to be introduced 
to the chain when necessary.  Is this possible?

many thanks in advance :)

M.

--
______________________________________________________________
Matthew Hodgson   matthew at mxtelecom.com   Tel: +44 845 6667778
                 Systems Analyst, MX Telecom Ltd.