Opened 12 years ago Closed 12 years ago #594 closed defect (fixed)14-bit issues in Weighted Prediction
Description
In 8.5.2.2.3 Weighted sample prediction process, variable offset1 is set equal to 1<<(shift1-1) but shift1 is set equal to 14 - bitDepth. If bitDepth equals to 14 then we have a negative shift. Basic non-invasive lightweight solution would be to set offset1 to 0 when bitDepth equals to 14. Another hardcore one would be to shift all MC (WP, IF...) precision on 15 bits instead of 14 bits that would be very invasive in the code. Btw, the same issue happens in HM7.0. (WP, IF). Change History (4)comment:1 Changed 12 years ago by DefaultCC Plugin
comment:2 Changed 12 years ago by pandrivoncomment:3 Changed 12 years ago by bbross
comment:4 Changed 12 years ago by bbross
Fixed in JCTVC-I1003_d6. Note: See
TracTickets for help on using
tickets. | This list contains all users that will be notified about changes made to this ticket. These roles will be notified: Reporter, Owner, Subscriber, Participant
|
Quoted in 8.5.2.2.3:
"– The variable offset1 is set equal to 1 << ( shift1 − 1 ) and the variable offset2 is set equal to 1 << ( shift2 − 1 ). "
Suggestions:
"– The variable offset1 is derived as follows.
or all could be written as a text (it allows removing to equation figures) like:
"- If shift1 is greater than 0, the variable offset1 is set equal to 1<<( shift1 - 1 ), otherwise shift 1 is set equal to 0.
HM should be fixed accordingly Cf. ticket #595.