generic-library/ffmpeg
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

error_concealment: Use previous pictures motion vectors when the current ones have been lost.

Browse Source 
Looks better for some cases, worse for others, overall not much difference.
Its more correct though.
Signed-off-by: Michael Niedermayer <michaelni@gmx.at>
This commit is contained in:
Michael Niedermayer
2011-05-11 00:47:55 +02:00
parent a811ec7cb5
commit 3732c48304
3 changed files with 12 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
libavcodec/error_resilience.c
View File

@@ -381,6 +381,14 @@ static void guess_mv(MpegEncContext *s){
fixed[mb_xy]= f;
if(f==MV_FROZEN)
num_avail++;
else if(s->last_picture.data[0]){
const int mb_y= mb_xy / s->mb_stride;
const int mb_x= mb_xy % s->mb_stride;
const int mot_index= (mb_x + mb_y*mot_stride) * mot_step;
s->current_picture.motion_val[0][mot_index][0]= s->last_picture.motion_val[0][mot_index][0];
s->current_picture.motion_val[0][mot_index][1]= s->last_picture.motion_val[0][mot_index][1];
s->current_picture.ref_index[0][4*mb_xy] = s->last_picture.ref_index[0][4*mb_xy];
}
}
if((!(s->avctx->error_concealment&FF_EC_GUESS_MVS)) || num_avail <= mb_width/2){