[MPlayer-users] Bicubic scaling in vo_gl - does it ever work?
mosgalin at VM10124.spb.edu
Thu Jun 15 11:30:43 CEST 2006
Hi Reimar D?ffinger!
On 2006.06.15 at 10:34:25 +0200, Reimar D?ffinger wrote next:
> Sorry for giving you a bad commandline, it missed -demuxer rawvideo.
Makes a difference. I'm not familiar with rawvideo, you see...
> This one is actually tested (couldn't test the other one since windows
> has no /dev/zero):
> mplayer -vf test,crop=100:70:0:0 -geometry 1000x700+0+0 -vo
> gl:yuv=2:lscale=1 -demuxer rawvideo -rawvideo w=100:h=100 /dev/zero
> > Both with yuv=2 and yuv=3 there are no problems. However, should I try
> > to change brightness, contrast or saturation, I immediately see those
> > broken lines defects. Even when I don't change them, i.e. my config is
> This is weird, since none of all these have any effect on the bicubic
> interpolation code...
> One thing you could try: in libvo/gl_common.c, change
> OPTION ARB_precision_hint_fastest;
> OPTION ARB_precision_hint_nicest;
I'll try it. And yeah, I want nicest quality, why would I bother with
bicubic scaling in the first place otherwise? ;)
> Can anyone else reproduce these problems? If yes, can you also test
> under windows and see if that makes a difference?
I can test it in windows, but I have no means of compiling mplayer for
windows. Do you know where I can get win32 build with latest vo_gl
changes (after pre8)?
> > So, what do you think?
> That I can't make any sense of it...
*sigh* so I thought. The fact that sometimes it works for sure,
sometimes it doesn't and sometimes it's random really doesn't make a
sense. I mean, what's the difference between libmpeg2 and ffmpeg
decoders in absence of dr/slices/etc? There should be none...
More information about the MPlayer-users