On 19.08.2014, at 15:46, Clément Bœsch <u...@pkh.me> wrote: > On Tue, Aug 19, 2014 at 01:41:49PM +0000, Carl Eugen Hoyos wrote: >> Clément Bœsch <u <at> pkh.me> writes: >> >>> On Mon, Aug 18, 2014 at 11:39:26PM +0000, Carl Eugen Hoyos wrote: >>>> Clément Bœsch <u <at> pkh.me> writes: >>>> >>>>> + { AV_CODEC_ID_PCM_S24BE, MKTAG('n', 'i', '2', '4') }, >>>>> BBC typo */ >>>>> + { AV_CODEC_ID_PCM_S24LE, MKTAG('n', 'i', '2', '4') }, >>>>> /* BBC typo */ >> >>> I'm sorry, I don't understand what you mean. >> >> I believe that FFmpeg would not be able to >> decode "ni24" samples with your patch above. >> If I understand the patch from Elemental >> correctly, all ni24 files they encountered >> are PCM_S24LE, while your patch would identify >> such files as PCM_S24BE. >> Or do I misunderstand? >> > > There is an endianess chunk which is used to swap BE <-> LE if necessary > (see lavf/mov.c:mov_read_enda()). If you look into the array, I'm > changing, there are a bunch of similar occurences around.
Considering that they got ni/in wrong, is there any particular reason to believe they got that one right? It would be kind of silly if we added a workaround/hack that hasn't been tested by anyone and in the end doesn't even work, especially if the risky version is the one with (trivially) more code... _______________________________________________ ffmpeg-devel mailing list ffmpeg-devel@ffmpeg.org http://ffmpeg.org/mailman/listinfo/ffmpeg-devel