On Tue, 2002-05-07 at 05:08, Gerd Knorr wrote: > > I didn't know which one did you implemented, but I know that H.263+ is > > just TMN11... > > BTW how dificult was the wrapper ? > > It is more or less a quick&dirty hack and somewhat ugly: both encoder > end decoder heavily use static variables with plenty of name clashes. > To be able to link both into one binary I had to pipe the sources > through sed to fix the name clashes. Argh. *sigh* Anyway TMN11 is full of globals and statics.... > > Did you have to implement RTP > > payload by yourself or this was already in the code ? > > TMN just takes yuv data and spills out H.263, it does no network stuff I know I have to hack it to compile fine in Linux and added MMX iDCT + XVideo grabbed from an old XawTV :) > at all. All the rtp stuff is mine, but IIRC it wasn't that hard as vic > has some infrastructure for this (the other codecs use rtp too ...). Fine, also H.263+ RTP payload is easy if the coder spits GOBs synced streams as ffmpeg do, then I have to rip the PSC/GBSC first word.... > Somewhere I still have a command line utility which sends out H.263 > files (written by the tmn encoder) as RTP streams, which I've used for > testing purposes. If you can find it will help me a lot :) -- Juanjo sin .sig :(
