[x264-devel] Generating H264 for iPad?

[Reindl Harald]

Am 23.02.2012 17:24, schrieb Jason Garrett-Glaser:
>>> Use the -preset, -profile, etc options, not a huge pile of cargo-cult options.
>>
>> the damned -preset / -profile parameters including their
>> x264-values are changing permanently - this way you can
>> NEVER make a useful wrapper without maintain hughe
>> code-switches and ways control them by a specific
>> ffmpeg-version
> 
> I have no idea what you're even saying here.  The -preset/-profile
> names have never changed since their introduction many years ago.
> ffmpeg has had per-codec option support for ages now too, so there
> really isn't any excuse to continue this mess.

this is simply not true

ffmpeg -i demo.mp4 -vcodec 'libx264' -acodec 'libfaac' -vpre 'baseline' -vpre '-vpre fastfirstpass'  xxx.mp4
File for preset '-vpre fastfirstpass' not found

this worked 1-5 years ago and was renamed multiple times
and yes i do two-pass-ecoding and with throw away the
presets it is no more randomly broken after ffmpeg updates

> Please stop spreading bad advice on the mailing list.
> 
> For specifics:
> 
>>>> ffmpeg -i 'inputfile.mp4' -y -vb '448k' -ab '96k' -ar '44100' -ac '2' -r '25' -async '1' -maxrate '544k' -f 'ipod'
>>>> -vcodec 'libx264' -acodec 'libfaac' -flags '+loop+mv4' -cmp '256' -partitions
>>>> '+parti4x4+parti8x8+partp4x4+partp8x8+partb8x8' -me_method 'hex' -me_range '16' -subq '7' -trellis '1' -refs '5' -g
>>>> '250' -keyint_min '25' -sc_threshold '40' -i_qfactor '0.71' -qcomp '0.6' -qmin '0' -qmax '69' -qdiff '4' -bufsize
>>>> '2M' -directpred '3' -rc_lookahead '50' -threads '2' -coder '0' -bf '0' -flags2 '-wpred-dct8x8' -wpredp '0'
>>>> 'outputfile.mp4'
> 
> 1.  You're using ABR mode, which is strongly discouraged

as you above i have no idea what you are saying here

> 2.  async 1 is set for absolutely no reason.  Does that option even
> exist anymore?

[harry at rh:/data/autotest/parts/ffmpeg]$ ffmpeg --help | grep async
-async              audio sync method

> 3.  +mv4 doesn't even do anything

well, maybe there are too much bad tutorials out there

> 4.  At least half of those options do nothing, because they are the defaults

and that is one MORE reason to set them because if someone
changes the defaults it is NOT my intention to get changed
any behavior in my wrapper

> 5.  You are crippling x264's speed by reducing the number of threads

do not make assumptions without knowing the context

this was a LOG-SNIPPET from a webserver with some
encoding apps running

x264 has NOT to use more than 50% of CPU especially
because we are speaking about a VMware-Cluster so
even the machine has it's hardware not alone

> 6.  Your bufsize is chosen without any apparent reason.  It doesn't
> match any typical number of seconds either.

as long it does not hurt....
the machine has ( GB memory

> 7.  You have some options enabled that are pointless except at very
> insane speed settings (e.g. p4x4).
> 
> Seriously, why?

because after many hours of testing and optimizing this were the
best defaults with minimum file size, maximum converting speed
and best quality compared a hughe amount of input/output files
and this until know has worked for each convert with expected
good result and without touching the ffmpeg-php-wrapper for more
than 1.5 years so it is currently useable with older ffmpeg
versions as also with the newest one

respect the fact that i have other work beside changing permanently
working things by making any software-updates as i had to do
for 'vpre'-profiles in the past

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 262 bytes
Desc: OpenPGP digital signature
URL: <http://mailman.videolan.org/pipermail/x264-devel/attachments/20120223/e99917fd/attachment.pgp>