> Maybe he does it now. It looks better. But what I don't understand is > following. I get complete encoded frames. The pts-difference tells me, that > from picture to picture is a delay of 40ms. That means I get a complete > frame. What I want to get is an encoding by fields. Don't get me wrong. I > know that x264 only produces MBAFF streams, this is ok for me, but I think > that x264 can also produce interlaced fields with a duration of 20 ms right?
AFAIK you always get complete frames from the decoder. It doesn't matter if you encode interlaced or progressive. In the interlaced case, a frame is made up of two fields. The PTS value describes the FRAME timestamp. If two frames have a PTS distance of 3600 on a 90000 timebase, one FRAME has a duration of 40ms. Therefore if a frame lasts 40ms (25fps), the two FIELDS then have a duration of 20ms each. So, in the end, your result seems to be correct already. Regards Andre p.s.: Please don't top post. >>> could anyone please tell me, how I have to configure my AVCodecContext if I >>> want to encode my video with x264 interlaced? >>> >>> A simple >>> >>> ctx->flags |= CODEC_FLAG_INTERLACED_DCT; >>> >>> does not work. (I got this from the libx264.c - file). >>> >>> Any ideas? >> >> try additionally: >> >> ctx->flags |= CODEC_FLAG_INTERLACED_ME; _______________________________________________ libav-user mailing list [email protected] https://lists.mplayerhq.hu/mailman/listinfo/libav-user
