With no control of what kind of hardware is used, will it ever work out of
the box for everyone?

Therein lies the rub. The quality of TVout cards generally ranges from acceptable to really awful. I have an old NV MX400 that's VIVO... both of which turn out to be so bad quality as to be useless. Once upon a time I had a brief correspondence with an nvidia support guy who knew what he was talking about. He said that with the newer cards (with tvout on-chip), some modeline resolutions cause the binary driver to go into "non-scaling" mode. That at least reduces one of the variables.

Bjorn, I believe my MSI card is using nvidias nv17 (or 18) chip for tv out.
Many ASUS cards use a different chip, according to this
database<http://ssl.jbmedia.de/cardbase/cardbase_query.asp?lang=en>but
data can be entered by anyone, so it might be inaccurate.

For PAL, if the tv out is synched to a 50Hz display refresh rate, and no
scaling occurs, there are two perfectly reasonable ways to generate a tvout
signal:

1. Grab both fields from one frame and skip one frame
2. Grab even field from one frame and odd from next

Does anyone know if both these methods are used in video cards?

If the display is running at 25hz, only the first method makes sense, so one
would hope that the tv out chip would do the right thing. In my case, with
my particular graphics card, that seems to be the case since my picture is
smooth...

Questions like these are why I built the hardware and run the tv (ntsc) straight off the VGA port. Of course that's got issues of its own, but at least you know what you've got.

Something that may be useful to throw into the mix is the DOUBLESCAN and INTERLACE flag on the modeline. Strange things can happen when you combine them, and of course it doesn't work on all drivers.

Have you tried simply programming a non-VGA-friendly (i.e. 480i) modeline and using the Xv VBLANK sync as-is? I don't have a tvout card that doesn't suck, but perhaps the chip gets smarter about it when it doesn't have to temporally scale anything.

-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

Reply via email to