[FFmpeg-user] Rotate 180 degrees

Cecil Westerhof Cecil at decebal.nl
Tue May 20 13:53:46 CEST 2014


Op Tuesday 20 May 2014 11:22 CEST schreef Moritz Barsnick:

> On Tue, May 20, 2014 at 11:05:33 +0200, Cecil Westerhof wrote:
>> file-size   620.4 MB
>> file-size   876.1 MB
>
> I've seen much bigger differences. Of course, mileage varies
> according to input material. I did tests on some example material of
> mine, and found that "veryfast" (or was is "fast") gave me an
> acceptable ratio. "Acceptable" remains to be defined. ;-) I just
> noticed that the slower presets hadn't given me any "significant"
> file size reduction, related to the added hours of wait time. (I
> have a very slow CPU.)

I will look into those other options also.


> What DID help _very much_ was ensuring that my libx264 was compiled
> with assembler support. Originally, my yasm was too old, and x264
> dropped the assembler-optimized routines. Recompiling with the
> proper yasm actually made H.264 encoding more than 4x faster for
> me!!!

Well, I use the openSUSE version. I could compile ffmpeg myself, but
with the current results I think it is not necessary. But something to
keep in mind.


>> And because I need to do ten other operations (at the moment I do
>> not know how to do it in one) it is very welcome indeed. :-D
>
> Please do ask - there's always some way to combine and optimize,
> that can save you even more time.

Problem solved. I had asked a question and Raúl Mansilla gave an
answer I could us. I now use:
  -filter_complex \
    "   [0:v][1:v]    overlay=0:0:enable=between(t\,0\,7) [stage1];
     [stage1][2:v] overlay=0:0:enable=between(t\,40\,45)  [stage2];
     [stage2][3:v]overlay=0:0:enable=between(t\,105\,110) [stage3];
     [stage3][4:v]overlay=0:0:enable=between(t\,210\,214) [stage4];
     [stage4][5:v]overlay=0:0:enable=between(t\,270\,275) [stage5];
     [stage5][6:v]overlay=0:0:enable=between(t\,306\,310) [stage6];
     [stage6][7:v]overlay=0:0:enable=between(t\,415\,421) [stage7];
     [stage7][8:v]overlay=0:0:enable=between(t\,515\,523) [stage8];
     [stage8][9:v]overlay=0:0:enable=between(t\,622\,642) [stage9];
     [stage9][10:v]overlay=0:0:enable=between(t\,722\,792)" \

It is not 10 times as fast (this step takes more as the single steps I
did first), but I think it is about 6 times as fast. Certainly a good
improvement.

I need to spend more time with ffmpeg: you can do very nice thinks
with it, if you know how.

-- 
Cecil Westerhof
Senior Software Engineer
LinkedIn: http://www.linkedin.com/in/cecilwesterhof


More information about the ffmpeg-user mailing list