[MPlayer-dev-eng] Re: Fwd: [MPlayer-users] Re: MPlayer 0.90RC2 and cvs - VIDIX bug with RADEON

[James Stembridge]

I have further investigated this problem and seem to have made some progress.

I now think that the change I was complaining about is actually correct. With
X set to 24-bit colour the overlay worked perfectly. It seems that the 16-bit
support was broken anyway, and this change simply made it a bit more broken.

I investigated the values of some of the registers while playing a video in
windows and discovered that (at least for my card) the colour key registers
are being set incorrectly by vidix. I have attached a patch that corrects the
problem for me.

I suspect that as is the patch will break other cards. I'm assuming the
 colour key value was calculated that way for a particular reason. I do not
 have any other cards to test this with and so am unable to determine for
 which cards this fix needs to be activated. Anyone willing to test?

To add to this mess during my experimentations I have found that BGR 15-bit
support does not seem to be working properly. The video displays but has
somewhat psychedelic colours. I think I've spent enough time staring at hex
values over the past couple of days so will investigate this in the near
future (that is unless someone else wants to fix it!).

James.

Arpi wrote:
> argh.
> when will this mess end?
>
> --------- Forwarded message ---------
> From: James Stembridge <jstembridge at gmx.net>
> To: mplayer-users at mplayerhq.hu
> Subject: [MPlayer-users] Re: MPlayer 0.90RC2 and cvs - VIDIX bug with
> RADEON
>
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> Current CVS does not work for me, I just get a pink window with no
> video.
>
> Prior to this change being added to CVS:
>
> $ cvs diff -r 1.57 -r 1.58 vidix/drivers/radeon_vid.c
> Index: vidix/drivers/radeon_vid.c
> ===================================================================
> RCS file: /cvsroot/mplayer/main/vidix/drivers/radeon_vid.c,v
> retrieving revision 1.57
> retrieving revision 1.58
> diff -u -3 -p -r1.57 -r1.58
> --- vidix/drivers/radeon_vid.c  25 Dec 2002 20:57:11 -0000      1.57
> +++ vidix/drivers/radeon_vid.c  4 Jan 2003 14:56:45 -0000       1.58
> @@ -960,13 +960,13 @@ int vixInit( void )
>  #ifdef RADEON
>    switch(def_cap.device_id)
>      {
> -    case DEVICE_ATI_RADEON_QW:
>      case DEVICE_ATI_RADEON_MOBILITY_M6:
>      case DEVICE_ATI_RADEON_MOBILITY_M62:
>      case DEVICE_ATI_RADEON_MOBILITY_M63:
>      case DEVICE_ATI_RADEON_MOBILITY_M64:
>      case DEVICE_ATI_RADEON_9000_IF:
>        rage_ckey_model=1;
> +    case DEVICE_ATI_RADEON_QW:
>      case DEVICE_ATI_RADEON_QD:
>      case DEVICE_ATI_RADEON_QE:
>      case DEVICE_ATI_RADEON_QF:
>
> I got video but colour keying was not working correctly, ie video was
> visible even if the mplayer window was hidden behind others.

-------------------------------------------------------


-------------- next part --------------
A non-text attachment was scrubbed...
Name: vidix_radeon_qw_16_bit.diff
Type: text/x-diff
Size: 701 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-dev-eng/attachments/20030106/657d0078/attachment.diff>