render->p_past_surface = NULL;
switch(s->pict_type) {
- case FF_I_TYPE:
+ case AV_PICTURE_TYPE_I:
return 0; // no prediction from other frames
- case FF_B_TYPE:
+ case AV_PICTURE_TYPE_B:
next = (struct xvmc_pix_fmt*)s->next_picture.data[2];
if (!next)
return -1;
return -1;
render->p_future_surface = next->p_surface;
// no return here, going to set forward prediction
- case FF_P_TYPE:
+ case AV_PICTURE_TYPE_P:
last = (struct xvmc_pix_fmt*)s->last_picture.data[2];
if (!last)
last = render; // predict second field from the first