[FFmpeg-devel] [PATCH 1/4] libavcodec: new API for frame threading by step

[Michael Niedermayer]

On Sun, Aug 03, 2014 at 06:39:38PM +0200, Christophe Gisquet wrote:
> Hi,
> 
> note: I'm using "step" throughout the patch because of the step
> function and what the causal part most often looks like. I have no
> idea for another less confusing wording.
> 
> 2014-07-28 23:15 GMT+02:00 Michael Niedermayer <michaelni at gmx.at>:
> > maybe i misunderstand but the "progress[1] > y" check looks odd,
> > shouldnt progress[0] be updated to a larger value even if
> > progress[1] > y
> > ?
> 
> Let's look at the following schema, that maybe should be added to the
> documentation:
>          ___ p[0]
> p[1] ___|
>        p[2]
> 
> p(rogress)[0] and p[1] are the ordinates of the step. p[2] is the abscissa.

another way to vissualize the same thing is to consider

1)  0..p[2]     2) 0.......w
    .....|         .........
p[1]_____|     p[0]_________

that is two convex shapes, first a rectangle from (0,0) to (p[2], p[1])
and second the old "all above line"

and consider them finished&available to be all points that are in
either.

I had based my suggestions of what how to update based on this
vissualization

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

Avoid a single point of failure, be that a person or equipment.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 181 bytes
Desc: Digital signature
URL: <https://ffmpeg.org/pipermail/ffmpeg-devel/attachments/20140804/c876e838/attachment.asc>