[FFmpeg-user] Setting up a watch folder to encode adaptive bit rate videos

Moritz Barsnick barsnick at gmx.net
Mon Nov 2 12:12:10 CET 2015


On Sat, Oct 31, 2015 at 20:36:54 +0100, Henk D. Schoneveld wrote:
> On 30 Oct 2015, at 03:40, Joel Lopez <badassmexican at gmail.com> wrote:
> > 1.  Single or 2-pass encoding?  I'm reading quality isn't affected
> > much and that it may be possible to have aligned key frames with
> > either one.  What do you guys do?
> 2-pass in my opinion only makes sense when the resulting file has to be stored max-quality to finite storage medium. CD/DVD etc.
> Do some testing and look what % is saved and relate to the extra time/energy invested in that 2nd pass. 

I think the point was to use two-pass encoding in order to use the
intermediate analysis file for various second-pass conversions
(resolutions, qualities), so that those would result in the same key
frame locations. This was suggested on "the net" somewhere (Wowza?). I
neither know whether it is valid to do the second pass with differing
resolutions, nor whether the key frame locations are absolutely
determined by the first pass. (Does the first pass analysis determine
those positions?)

Joel, isn't is worth trying?

> > -x264opts 'keyint=24:min-keyint=24:no-scenecut’
> AFAIK 
> blocking key frames on scene cuts/changes will result in more or less visible degraded quality.

I totally agree. Overriding automatic scene change key frames by
constant-distance keyframes is just a hack for the point of having the
same positions.

Moritz


More information about the ffmpeg-user mailing list