[Bug 762352] decoder: hevc: Fill dependent slice segment headers while parsing

[GStreamer (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=762352

--- Comment #3 from Scott D Phillips <scott.d.phillips at intel.com> ---
(In reply to sreerenj from comment #2)
> (In reply to Scott D Phillips from comment #0)
> If the picture has multiple
> > slices with dependent slice segments then the dependent
> > segments from all slices will get populated with the fields
> > from the last independent slice segment header instead of its
> > corresponding independent slice segment header.
> 
> Is it written like that in the spec?
> I am not sure whether the commit message here is correct,

Maybe I worded that poorly. The spec says:

H.265 7.4.7.1:
> dependent_slice_segment_flag equal to 1 specifies that the
> value of each slice segment header syntax element that is not
> present is inferred to be equal to the value of the
> corresponding slice segment header syntax element in the slice
> header.

So this bug and patch all deal with that "corresponding slice
segment header" part. H.265 3.4.6 says "the values of some syntax
elements of the slice segment header are inferred from the values
for the preceding independent slice segment in decoding order."

So what I mean to say in the commit message is that before my
patch we're using the wrong slice segment header as
the "corresponding" one. Before my patch we always use the AU's
*last* independent slice segment header instead of correctly
using the preceding independent header.

> Missing syntax elements of the dependent slice segment headers
> should be always inferred from the previous independent slice
> segment. In that way the current code is right IMO...

The current code has a parsing phase followed by a decoding
phase. We keep the 'previous independent' reference up to date
through the parsing phase but it is actually dereferenced during
the decoding phase. So we're not pulling the values from
the "corresponding" header as we go, but rather we currently pull
the values from the *last* header after all parsing is done. In
that way we wind up populating all dependent slice segment
headers with the values from the last independent slice segment
header in decoding order.

> BTW, Could you please share a sample video ??

I'm testing against a licensed compliance test stream that I
won't be able to share publicly. I can mail it to you of course.

-- 
You are receiving this mail because:
You are the QA Contact for the bug.
You are the assignee for the bug.