[FFmpeg-devel] [FFmpeg-cvslog] avcodec/libx264: fix compilation with x264 builds >= 153

[James Almer]

On 12/26/2017 7:22 PM, Michael Niedermayer wrote:
> On Tue, Dec 26, 2017 at 02:58:19PM +0000, James Almer wrote:
>> ffmpeg | branch: master | James Almer <jamrial at gmail.com> | Mon Dec 25 19:41:09 2017 -0300| [2a111c99a60fdf4fe5eea2b073901630190c6c93] | committer: James Almer
>>
>> avcodec/libx264: fix compilation with x264 builds >= 153
>>
>> x264 now supports multibitdepth builds, with a slightly changed API to
>> request bitdepth during initialization.
>>
>> Reviewed-by: Ricardo Constantino <wiiaboo at gmail.com>
>> Signed-off-by: James Almer <jamrial at gmail.com>
>>
>>> http://git.videolan.org/gitweb.cgi/ffmpeg.git/?a=commit;h=2a111c99a60fdf4fe5eea2b073901630190c6c93
>> ---
>>
>>  libavcodec/libx264.c | 29 ++++++++++++++++++++++++++---
>>  1 file changed, 26 insertions(+), 3 deletions(-)
> 
> This breaks:
> LD_PRELOAD=/usr/local/x264-10/lib/libx264.so.146  ./ffmpeg -i ~/videos/mm-short.mpg -avcintra-class 100 -tune psnr -flags +ildct-global_header -t 0.5 -pix_fmt yuv422p10 -vf scale=1920:1080  -an x.mov
> 
> previously it encoded with high bit depth after the patch it attempts to use 8bit
> and fails:
> [libx264 @ 0x3c36a80] This build of x264 requires high depth input. Rebuild to support 8-bit input.
> 
> IIUC this patch changes from the value at runtime to the value at
> build time. That seems like a bad idea as they can differ and the value
> of the library linked to is the one mattering

Changed back to runtime.