question on deblock in B frames

Please take discussion of restricted code to a rarvcode
development
list (restricted list). We should not discuss it on the
public
mailing lists.

--greg.


Jay N. Shingala wrote:
> Hi,
> 
>  I am finding it difficult to understand the derivation
of filtering 
> strength for a B 4x4 sub block. While the two adjacent
sub blocks 
> referring to different frames and corresponding RefDiff
being true or 
> false is well understood, the confusion is regarding
the motion vectors 
> of forward and backward direction. I am referring to
rv89combo_c code 
> and it appears that only the forward mv differences are
used for 
> determining the strength to 1(when the mvds exceed or
equal one integer 
> pel). However, the same is not done for  the backward
mvds. If my 
> observation is correct, can anybody help me with the
reason for this ? 
> If not, them I am missing  something and can somebody
clarify this in 
> detail.
> 
>  To be more precise, if we have two adjacent B mbs
sharing an edge, both 
> being backward, and the mv differences exceed one
integer pel, is the 
> filtering strength for both the 4x4 blocks 1 ? It seems
logical but I 
> find it difficult to see the implementation for  same
in rv89combo_c  
> branch.
> 

_______________________________________________
Video-dev mailing list
Video-devhelixcommunity.org
http://lists.helixcommunity.org/mailman/listinfo/video
-dev