Skip to content

Cosmetic change to avoid confusion in the getWideAngle function

There have been already two tickets (#1064 and more recently #1096) reporting a mismatch between the software and the spec regarding the getWideAngle function behaviour. As replied in #1064, there is no mismatch, given that the software can implement what the spec says in many different ways. In particular, in the VTM implementation the concerned function returns an index that will be used in the intra prediction functions to generate the corresponding angular direction. Therefore, there is no need to 'jump' from 2 to -1, as the spec does, given that the values 0 and 1 are reserved for the Planar and DC modes.

For this reason, the present MR introduces two cosmetic changes to try to avoid this perfectly understandable confusion:

  1. The getWideAngle function has been renamed to getModifiedWideAngle. A comment has been added inside this function to explicitly mark the behavior.
  2. The PU::getWideAngIntraMode function has been renamed to PU::getWideAngle, given that it returns a wide angle index exactly in the way described in the draft's section 8.4.2.5.6 Wide angle intra prediction mode mapping process. A comment has also been added to explicitly state this.
Edited by Santiago de Luxán Hernández

Merge request reports