[linux-dvb] Re: Driver 0.8.1 and encrypted channels

[Ralph Metzler <rjkm@convergence.de>]

Carsten Koch writes:
 > Klaus Schmidinger wrote:
 > 
 > > (don't know if that would be any faster than that of analog TV).
 > 
 > Should be a lot faster. At least with my d-box I had the impression
 > that teletext pages are coming in ~ 5 times faster than on analog TV.
 > Of course, I may be mislead, because the d-box also buffers the pages,
 > which is of course the best approach.

It is not faster, just buffered by the d-box.
The digital teletext stream should be exactly the same as transmitted 
through analog TV. The packages even specify which VBI line to use.
They also contain the pages ready for transmission with the right bit
order, error correction, parity, etc. just as you find them when
sampling VBI with a frame grabber.
That's why it is called re-insertion.


 > > A solution I would consider better would be to collect the TT pages
 > > just like the EIT data and have all TT data from all channels
 > > available and display it via the OSD.
 > 
 > I agree. That would be one concrete implementation of the general
 > mechanism I described above.

Setting a PES filter for the teletext PID and decoding it is no problem
but displaying it with full teletext color depth (4bit?) on the OSD
might be.
I am also not stopping anybody from implementing a small program which
parses the teletext PES packets and feeds the results to existing
programs like alevt.

 > An idea that I had a few months ago might work a lot better, though.
 > I described it here but I am afraid people did not fully understand 
 > it, so I would like to try again:
 > 
 > It seems Ralph has implemented re-insertion now, meaning that
 > he can put the teletext information into the VBI of the video
 > signal that comes out of the card's FBAS connector. 
 > If that is possible, it might also be possible to encode
 > arbitrary text (i.e. text, that does not come from the TV station
 > but from an application program) this way. 
 > Considering the numerous problems we see with the OSD, would it 
 > not be a perfect solution to simply output of the OSD this way? 
 > What I mean is: create an API that hands the text of ONE specific 
 > teletext page over to the driver for re-insertion. 
 > Please note that teletext is blindingly fast (much faster than 
 > our currect OSD) when it has to transmit only one page. 
 > You could leave the TV set in teletext mode and it would
 > display the "teletext-OSD" if there is any, otherwise it 
 > would automatically display just the picture.
 > 
 > Would that not solve all the problems with the current OSD
 > (performance and reliability)?  We would even get a larger
 > OSD which forces us to page less often. And the old OSD could
 > go where it belongs: down the drain. Freeing up all of the
 > memory in the firmware that it required.


I think I also explained before that this is not possible.
You can only choose a PID to be used for re-insertion. You cannot
provide your own data.
This is all handled by the library TI provides for the AV7110 which is
only available as binary. Complete register descriptions are also not
available.

 > Still this may be a kludge. The best solution would of course
 > be to fix the OSD so it becomes fast and reliable even when
 > it covers the entire screen. Maybe with his growing experience
 > in the firmware area, Ralph can even pull this off? Would be
 > great!

What exactly are the problems with OSD right now?
I cannot do anything about the small memory and if you want to
transmit bitmaps during playback (and thus reducing the available
bandwidth for the playback) you cannot expect not to get problems.


Ralph
 


--- 
Info:
To unsubscribe send a mail to listar@linuxtv.org with "unsubscribe linux-dvb" as subject.