|
From: Arne S. <ar...@sc...> - 2000-11-09 21:48:07
|
I used libdv patched with xvimage patch, but I will check it with the SDL version too. However I am more of an end-user of libdv and quite clueless how the decoding actually works and where to look.... Arne -----Original Message----- From: Stefan Lucke Sent: Thursday, November 09, 2000 9:07 PM To: Libdv-Dev (E-Mail) Subject: RE: [libdv-dev] libdv bug? On Don, 09 Nov 2000, Arne Schirmacher wrote: > I tried the hardware version too, but there are still errors (green spots > every once in a while). Check out the file problem3.dv, where I isolated a > few frames. My card is a Matrox G200. I can't verify the green spots in this file. Did you used the SDL or the plain Xv version? My card is a Matrox G400DH (much faster AGP transfers than my previous G200 with my MB ASUS-P5A). > > Anyway, if the hardware version works better this suggests that the problem > is somewhere in the yuv to rgb conversion code. I don't thing that the source of the problem is YUV -> RGB conversion. The table lookup for Y values is sometimes fed with large negative values -133 .. (not true for your examples but for some of mine). Such negative values should not exist !! Regions affected by your example are produced by idct88 and idct248. So where to look for ? -- mfg Stefan Lucke (lu...@sn...) _______________________________________________ libdv-dev mailing list lib...@li... http://lists.sourceforge.net/mailman/listinfo/libdv-dev |