[10:55] <Shimmy> https://ffmpeg.org/trac/ffmpeg/ticket/747 [11:42] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r5932938c9a 10ffmpeg/libswresample/x86/ (audio_convert.asm swresample_x86.c): [11:42] <CIA-17> ffmpeg: swr: float_to_int32_sse2() [11:42] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [12:33] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r73edb58c3c 10ffmpeg/libswresample/x86/ (audio_convert.asm swresample_x86.c): [12:33] <CIA-17> ffmpeg: swr: float_to_int16_sse2() [12:33] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [13:03] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r1d128e5814 10ffmpeg/libavformat/nutdec.c: [13:03] <CIA-17> ffmpeg: nutdec: replace assert with av_assert0. [13:03] <CIA-17> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de> [13:03] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r370211f470 10ffmpeg/libavformat/nutdec.c: [13:03] <CIA-17> ffmpeg: nutdec: minor simplification. [13:03] <CIA-17> ffmpeg: Also fixes an (incorrect) "control reaches end of non-void function" [13:03] <CIA-17> ffmpeg: warning with some compilers. [13:03] <CIA-17> ffmpeg: Signed-off-by: Reimar Döffinger <Reimar.Doeffinger@gmx.de> [13:03] <CIA-17> ffmpeg: 03Reimar Döffinger 07master * r755667eebc 10ffmpeg/libavformat/rmenc.c: (log message trimmed) [13:03] <CIA-17> ffmpeg: rmenc: more than 2 streams are not currently possible. [13:03] <CIA-17> ffmpeg: It is possible that just extending the RMMuxContext.streams [13:03] <CIA-17> ffmpeg: array would avoid it. [13:03] <CIA-17> ffmpeg: It is also possible that two audio streams will fail to mux [13:03] <CIA-17> ffmpeg: correctly as well, though at least it should not crash for [13:03] <CIA-17> ffmpeg: this reason. [14:26] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r65722e7fc5 10ffmpeg/libswresample/x86/ (audio_convert.asm swresample_x86.c): [14:26] <CIA-17> ffmpeg: swr: int32_to_int16_mmx/sse [14:26] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [14:28] <Shimmy> https://ffmpeg.org/trac/ffmpeg/ticket/747 [14:44] <burek> Shimmy, http://ffmpeg.org/consulting.html [15:45] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r2d6c29f566 10ffmpeg/libswresample/ (swresample.c swresample_internal.h): [15:45] <CIA-17> ffmpeg: swr: add set_audiodata_fmt() and use it to simplify code [15:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [15:45] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r106789dfa0 10ffmpeg/libswresample/ (swresample.c swresample_internal.h): [15:45] <CIA-17> ffmpeg: swr: add sample format to AudioData [15:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [15:45] <CIA-17> ffmpeg: 03Michael Niedermayer 07master * r9f98a8e9bc 10ffmpeg/libswresample/audioconvert.c: [15:45] <CIA-17> ffmpeg: swr: use memcpy when input and output match [15:45] <CIA-17> ffmpeg: Signed-off-by: Michael Niedermayer <michaelni@gmx.at> [17:17] <burek> one noob question.. if I want to use 1 ffmpeg git cloned on my machine to produce several different builds (one gpl, another non-free), how can I specify in the configure script to put all object/binary/make files into separate directory for each build? [17:23] <Compn> set the bin dir ? [17:23] <Compn> ./configure --help | grep -i bin [17:24] <burek> --bindir=DIR install binaries in DIR [PREFIX/bin] <--- that? [17:24] <burek> well, that installs binaries into some dir [17:24] <burek> I need all *.o files to put into /obj/ [17:27] <Compn> oh [17:27] <Compn> no clue [17:27] <Compn> yeah that [17:29] <burek> would it be enough, for example, to just cd into the obj/ directory and run ../configure ...? [17:33] <Compn> sounds like a user question :P [17:33] Action: Compn is helpful! [17:33] <burek> :beer: :) [17:33] <Compn> sounds like you want 'to build ffmpeg out of tree' [17:33] <Compn> try google that [17:36] <burek> that's it :) thanks a lot! :) [18:39] <CIA-17> ffmpeg: 03Carl Eugen Hoyos 07master * r3dafde0b96 10ffmpeg/libavcodec/ (cscd.c version.h): [18:39] <CIA-17> ffmpeg: Simplify camstudio decoding and fix odd 16bit decoding. [18:39] <CIA-17> ffmpeg: Fixes ticket #1220. [18:47] <CIA-17> ffmpeg: 03Stefano Sabatini 07master * r46eba43e0e 10ffmpeg/libavfilter/buffersink.h: lavfi/buffersink: fix header inclusion guard name [19:30] <burek> Compn, do you know maybe how can I compile libx264 in some tmp directory without make install and then tell ffmpeg to find what it needs for libx264 in that tmp directory? [19:30] <burek> something like ./configure --enable-libx264=/my/tmp/dir/x264 :) [20:40] <burek> also, is it more appropriate (if I'd like to compile 32bit ffmpeg on 64bit machine) to use "--arch=i386" or "--cc='gcc -m32'" [20:41] <Daemon404> --extra-cflags and --extra-ldflags [20:42] <burek> i see, it's better to use those 2 instead of -cc, but does --arch set up all that automatically? [20:43] <Daemon404> that was in response to the part about a local libx264 [20:43] <burek> oh, ok :) thanks :) [21:31] <burek> hm, I hit a wall now.. I've built libx264 as 32bit lib/bin and it works on 32bit machine [21:32] <burek> I've also built bare ffmpeg (just ./configure) as 32bit bin and it also works on 32bit machine [21:32] <burek> but trying now to build ffmpeg+libx264 as 32bit bin gives me this: [21:33] <burek> http://pastebin.com/kKpuGtwn [21:33] <burek> shortly, undefined reference to `x264_encoder_open_124' [21:33] <burek> :S [21:49] <michaelni> burek, you compile ffmpeg against the headers for the x264 you link it too ? aka they are matching ? [21:49] <michaelni> both version 124 ? [21:49] <burek> I figured I've put --arch=i386 which I guess is invalid.. I put --arch=x86 and doing make again [21:50] <burek> both x264 and ffmpeg are from latest git if that's what you are asking :) [21:59] <burek> woo hoo it works \o/ :))))))))) [22:01] <burek> or not :/ [22:01] <burek> ./ffmpeg: error while loading shared libraries: libx264.so.124: cannot open shared object file: No such file or directory [22:01] <burek> back to the lab.. [22:09] <Daemon404> protip: have the shared lib somewhere ffmpeg can see [22:10] <burek> I figured it's looking in usr/local/lib and /usr/lib but I've used --prefix=/custom/dir to install compiled x264 [22:10] <burek> I'm trying static build now and if it works, that will be ok too :) [22:11] <burek> it would be cool if I could just speed up make like not to build everything, just the very basic set fo formats, codecs, etc [22:13] <Daemon404> burek, you can enable/disable anything you want [22:13] <Daemon404> configure --help [22:13] <Daemon404> also, this may be mroe suited to #ffmpefg [22:13] <Daemon404> er [22:13] <Daemon404> #ffmpeg [22:13] <burek> ok, sorry [22:13] <burek> :X [22:52] <CIA-17> ffmpeg: 03Christophe Gisquet 07master * rfe5ed69dc7 10ffmpeg/libavcodec/x86/ (rv34dsp.asm rv34dsp_init.c): [22:52] <CIA-17> ffmpeg: rv34dsp x86: implement MMX2 inverse transform [22:52] <CIA-17> ffmpeg: 141 cycles down to 51. [22:52] <CIA-17> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> [22:52] <CIA-17> ffmpeg: 03Christophe GISQUET 07master * re75d1d4f73 10ffmpeg/libavcodec/x86/dsputil_mmx.c: [22:52] <CIA-17> ffmpeg: dsputil x86: revert a test back to its previous value [22:52] <CIA-17> ffmpeg: Commit 356ee8d caused the initial inversion. [22:52] <CIA-17> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> [22:52] <CIA-17> ffmpeg: 03Christophe Gisquet 07master * r8802799991 10ffmpeg/libavcodec/rv34.c: [22:52] <CIA-17> ffmpeg: rv34: remove constant parameter [22:52] <CIA-17> ffmpeg: is_block2 was always 0, so just remove it, and change accordingly the code. [22:52] <CIA-17> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> [22:52] <CIA-17> ffmpeg: 03Ronald S. Bultje 07master * r774b681b80 10ffmpeg/libavcodec/rv34.c: [22:52] <CIA-17> ffmpeg: rv34: remove inline keyword from rv34_decode_block(). [22:52] <CIA-17> ffmpeg: Force-inlining increases object size by 8kB and causes a 2% slowdown. [22:52] <CIA-17> ffmpeg: 03Christophe Gisquet 07master * rd834ab0f2a 10ffmpeg/libavcodec/rv40.c: [22:52] <CIA-17> ffmpeg: rv40: perform bitwise checks in loop filter [22:52] <CIA-17> ffmpeg: Down from 95 kcycles to 93 (including all called functions). [22:53] <CIA-17> ffmpeg: Signed-off-by: Ronald S. Bultje <rsbultje@gmail.com> [00:00] --- Mon Apr 30 2012