[x264-devel] commit: Make B-pyramid spec-compliant (Lamont Alston )

[Jason Garrett-Glaser]

On Mon, Oct 19, 2009 at 5:57 AM, Loic Le Loarer <lll+vlc at m4x.org> wrote:
> Hi,
>
> On Monday October 19 2009 at 11:29:45 AM +0200, git version control wrote:
>> Make B-pyramid spec-compliant
>> The rules of the specification with regard to picture buffering for pyramid coding are widely ignored.
>> x264's b-pyramid implementation, despite being practically identical to that proposed by the original paper, was technically not compliant.
>> Now it is.
>
> Can you elaborate which rules aren't respected in the pyramid coding
> before this commit ? I'm wondering how wrong the generated streams were.

Pyramid coding, as defined in the original paper and JM, assumes that
if we decode frame X but do not display frame X until time X+1, there
is no memory cost in storing frame X in the meantime.  According to
the spec though, it takes one DPB slot.

As such, the streams took one more DPB slot than x264 claimed that they did.

Dark Shikari