Lucian Muresan wrote: > Ville Syrjälä wrote: >> On Tue, Nov 01, 2005 at 03:48:00PM +0100, Lucian Muresan wrote: >>> Hi folks, >>> >>> as you may remember from previous postings of mine, I have serious >>> trouble in using df_xine on my G400 DH TV-out (even with latest DirectFB >>> and matroxfb Matrox-relevant patches from Ville), when playing back >>> interlaced video, which obviously I don't need and want to be >>> deinterlaced by the software when viewing it on my standard PAL TV. The >>> problem is that I can't have both correct aspect ratio AND smooth >>> playback at the same time with some let's call them weird actual aspect >>> ratios they broadcast on DVB. I posted for testing 2 samples which >>> illustrate this, one with a standard 720x576 resolution, which is played >>> ok when selecting "square" (R + 1) in df_xine, and one non-standard, >>> which has to be correctly scaled to look well, 528x576. Both contain >>> horizontal scrolling text on which one can see what I mean. The one with >>> the "correct" resolution of 720x576 is the only one which can beplayed >>> back correctly, but with the other one, as I said, I get either wrong >>> aspect ratio, or non-smooth playback. >> >> StretchBlit() on interlaced content is a bad idea as we (still) scale the >> whole frame in one operation :( >> > > So here it is again, my patch. This time it works well for me on the > CRTC2 for different resolutions of 4:3 and standard 720x576 16:9 DVB > interlaced content, I get almost smooth playback AND correct aspect > ratio (well, once in few seconds, there seems to be something like a > little sync catchup, visible as a speed jump on scrolling text, but it's
[...] > Is it possible that this occasional "jump" is due to using StretchBlit > (I don't know exactly when and where, it's Claudio's code)? Is there > some way to improve the situation with df_xine? Thank you Ville, the "jump" is gone!!!! In the meantime I succeeded to manually apply version 3 of the triple buffering/field parity ioctl patch to DirectFB (I had a fresh checkout, that's why it didn't apply cleanly in the first attempt, I think). So I'm happy with the result for now, will check again tomorrow after work, maybe someone else can also try my patch to df_xine. Lucian _______________________________________________ directfb-dev mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-dev
