[x264-devel] [PATCH] Relaxed parameter checking at encoder/encoder.c

Alexander Strange astrange at ithinksw.com
Sun Dec 12 00:05:20 CET 2010


On Dec 11, 2010, at 6:00 PM, Jason Garrett-Glaser wrote:

> 2010/12/11 Marco Aurélio Graciotto Silva <magsilva at gmail.com>:
>> Dear x264 developers,
>> 
>> Lately, I'm trying to compress some movies using MPEG-4 AVC
>> Predictive Profile (aka lossless compression). I wasn't using
>> x264 directly, but FFmpeg. Then I strumbled across an issue
>> regarding some parameters that FFmpeg defined by itself and
>> which, when using x264 in lossless mode, results in an error:
>> 'broken ffmpeg default settings detected'. Further information
>> and instructions to reproduce this are described at
>> http://permalink.gmane.org/gmane.comp.video.ffmpeg.user/29323
>> 
>> I traced the issue to encoder/encoder.c, more precisely at
>> line 471, in a block of code that, as its comment says, "Detect
>> default ffmpeg settings and terminate with an error."
>> 
>> From the FFmpeg side, I can workaround this by setting
>> some parameters (such as qmin, qmax, etc), even though
>> such parameters will be overriden by x264 (because of the
>> lossless mode).
>> 
>> However, a better solution would be to, in the code that x264 tries
>> to detect FFmpeg defaults, ignore some parameters that will be
>> overwritten anyway in lossless mode. I created a patch
>> x264-lossless-ffmpeg_parameter_check.patch and it worked
>> just fine here. Is it okay?
>> 
>> 
>> Regards,
>> 
>> Marco Aurélio
> 
> As I already told you on the ffmpeg thread, this is not a bug; this is
> entirely intentional behavior.  x264 uses the broken defaults of
> ffmpeg to detect when a user has not set a -vpre.  Relaxing these to
> allow users to remove parts of the ffmpeg preset is entirely
> contradictory to the purpose of said code.
> 
> Patch, again, rejected.
> 
> Jason

I sent a patch series to ffmpeg-devel with a superficial (but effective) fix for this, but haven't had time to finish it.
See "load -vpre/apre default by default".



More information about the x264-devel mailing list