[x265] [PATCH 1 of 2] slicetype: Modify Scenecut algorithm to detect scene transition points

Deepthi Nandakumar deepthi at multicorewareinc.com
Mon Oct 5 18:12:30 CEST 2015


Hello,

We cannot reproduce this scene cut error. We tested with 4 different clips,
and the current tip usually produces one I-slice less than the tip before
this changeset. This is not necessarily a bug.

We also tested all cases in both 8bpp and 10bpp modes, and the behaviour
was fairly predictable in both.

Do you have a specific source clip/command line where you saw this issue?

Thanks,
Deepthi



On Mon, Oct 5, 2015 at 9:33 AM, Xinyue Lu <maillist at 7086.in> wrote:

> Hi,
>
> After this patch, x265 seems to always produce constant length gop,
> where length is max keyint, completely ignoring scene cuts.
>
> For example, having --keyint 360 will produce a stream where IDR is
> inserted at 0, 360, 720, 1080, 1440, 1800, ... all the way to the last
> frame.
>
> I back out the related patches (rev 10912, 10923, 11005) and it's back
> to normal.
>
> Can you reproduce this behavior? Is it a feature or a bug?
>
> I've also left a post in doom9 thread.
>
> Thanks
>
> On Mon, Aug 17, 2015 at 1:41 AM,  <aarthi at multicorewareinc.com> wrote:
> > # HG changeset patch
> > # User Aarthi Thirumalai
> > # Date 1437505166 -19800
> > #      Wed Jul 22 00:29:26 2015 +0530
> > # Node ID 51320561d41f32faa75dd3d04d88ea68500ce995
> > # Parent  d56b2466c04459205287e1581d8a36eebf372ba6
> > slicetype: Modify Scenecut algorithm to detect scene transition points
> > to improve Rate Control (refs #160).
> >
> > identify scene trasitions, fade-ins, fadeouts, sceneCuts and signal the
> flag bSceneCut
> > in Lowres structure. This flag will be used by RateControl to adjust the
> wps during scene cuts.
> >
> _______________________________________________
> x265-devel mailing list
> x265-devel at videolan.org
> https://mailman.videolan.org/listinfo/x265-devel
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20151005/5eed21b8/attachment.html>


More information about the x265-devel mailing list