[x265] My bug reports for rev. ~400-450

[Mario *LigH* Rohkrämer]

Am 21.10.2013, 20:01 Uhr, schrieb Steve Borho <steve at borho.org>:

> On Mon, Oct 21, 2013 at 12:21 PM, LigH <contact at ligh.de> wrote:
>
>>   * when --weightp is used, the encoders (GCC-x86, VC11-x86, VC11-AMD64)
>> crash at the end of the encoding (finished or cancelled by Ctrl+C)
>>
>
> weightp is also only partially implemented and still being debugged.
>
> a command line would be help for reproduction.

+----
x265.exe sintel_trailer_360.y4m -o sintel_trailer_360.hevc -F 3 -q 12 -b 3  
--b-adapt 2 --bframe-bias 30 --weightp --ref 3 -i 240 --no-psnr --ssim  
--rdpenalty 1 --csv sintel_trailer_360.csv
+----

The Sintel trailer was resized to fit inside 640x360 (one of the YouTube  
standard resolutions) and cropped (due to the Cinemascope AR, to exactly  
640x272). The full 1080p resolution is tested only rarely due to the low  
speed.

The crash (APPCRASH, exception c0000005 in ntdll.dll) leaves the created  
HEVC bitstream incomplete, so I believe it must happen very late, but  
before gracefully closing the output file handle.


One more bug: Although the video was set to exactly 24.0 fps, x265 reports  
"23Hz":

+----
y4m  [info]: 640x272 23Hz, frames 0 - 1252 of 1253
+----

-- 
__________

Fun and success!
Mario *LigH* Rohkrämer
mailto:contact at ligh.de