I have at various times done some experimenting with this to see what can be
done and by reducing the buffering a little and taking out some redundant
pausing I was able to get the channel change time down to around 200ms on my
DVB-T setup.
However, then some other changes went in and change times went sky high
again, and I ran out of time to fiddle with it much more.
I believe at least 50% of the channel change time can be removed from at
least the DVB channel change code. It should be possible to get the channel
retuning done in around 50-100ms with a DVB-T card (or less) and then some
small buffering on top would be less than 0.5s changes.
I have posted various questions on this in the past, but no one with more
knowlege of the code flow has shown any interest in helping out. I don't
have infinite time so I have stopped working on it for the time being.
My suggestion is to stop bitching about the time change and put in some
timing statements into the channel change code. Find out which bits are slow
and then bang on about *that* bit here on this list. At least that way it's
specific and constructive
I guess I'm curious why it's so high to begin with. It seems to
me that the card's changing shouldn't take more than a few 10's of
milliseconds. The rest is waiting on buffers to fill/flush?
I still think that the ideal situation would be to push
uncompressed video to the screen, while simultaneously sending the
compressed to the ringbuffer. If you think about it, all of the PVR
operations on LiveTV are in a different mode than "channel-surfing."
Pausing, rewinding, etc require timshifting and decoding the
recorded/encoded data. *Diplaying* the other hand (no buffer) can be a
realtime operation... data coming in gets immediately displayed as *well*
as buffered to disk. Transitioning between the two (i.e. realtime->replay
for hitting "pause" or "rewind" while watching livetv, or replay->realtime
for hitting channel up/down while watching a time-shifed livetv) need not
be as responsive as nontransitioning (realtime->realtime channel up/down).
Again, I'm not bitching that it hasn't been done. It seems like a
lot of work and most people who might implement it (understandably) don't
care.
Besides, for ivtv captures, it would require keeping the RAW YUV
part in sync and continuously functioning along with the MPEG part. For
that matter, it may be academic if the card cannot operate with both at
the same time.
I'll shut up now... :)
-Cory
--
*************************************************************************
* Cory Papenfuss *
* Electrical Engineering candidate Ph.D. graduate student *
* Virginia Polytechnic Institute and State University *
*************************************************************************
_______________________________________________
mythtv-dev mailing list
[email protected]
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
