Changes between Initial Version and Version 1 of Ticket #636, comment 3


Ignore:
Timestamp:
31 Oct 2019, 14:49:18 (5 years ago)
Author:
afilippov
Comment:

Legend:

Unmodified
Added
Removed
Modified

  • Ticket #636, comment 3

    initial v1  
    11Since multiplication by 0 is well defined, to refer to this reference sample (ref[18] in your example) or not is just an implementation question that can be handled in different ways. Thus, to me, the reported issue is irrelevant to the spec as a result of equation (8-138) is always defined and doesn't depend on a value of this reference sample (ref[18] in your example) that may take an arbitrary value which is called "undefined" in your report.
    22
    3 JVET-P0626 is aimed at removing the redundant operations that forced implementers to perform an extra padding operation. If JVET-P0626 is not applied, it's unclear why the reference sample (ref[18] in your example) should be set to p[ −1 + refW ][ −1 − refIdx ], although any value can be used instead. Thus, the idea to define a variable that can take an arbitrary value without any affect on a result is not quite clear to me.
     3JVET-P0626 is aimed at removing the redundant operations that forced implementers to perform an extra padding operation. If JVET-P0626 is not applied, it's unclear why the reference sample (ref[18] in your example) should be set to p[ −1 + refW ][ −1 − refIdx ], although any value can be used instead. Thus, the idea to define a variable that can take an arbitrary value without any effect on a result is not quite clear to me.
    44
    55I don't see any spec related issues. The SW bug mentioned at the end of your report is fixed in the merge request ​https://vcgit.hhi.fraunhofer.de/jvet/VVCSoftware_VTM/merge_requests/1096 with JVET-P0626 inside.