The AVC software decoder always aligns the decoded dimensions to 16 pixel boundaries, report this properly to callers.
This commit is contained in:
@@ -322,8 +322,10 @@ status_t AVCDecoder::read(
|
||||
crop_top = crop_left = 0;
|
||||
}
|
||||
|
||||
mFormat->setInt32(kKeyWidth, crop_right - crop_left + 1);
|
||||
mFormat->setInt32(kKeyHeight, crop_bottom - crop_top + 1);
|
||||
int32_t aligned_width = (crop_right - crop_left + 1 + 15) & ~15;
|
||||
int32_t aligned_height = (crop_bottom - crop_top + 1 + 15) & ~15;
|
||||
mFormat->setInt32(kKeyWidth, aligned_width);
|
||||
mFormat->setInt32(kKeyHeight, aligned_height);
|
||||
|
||||
mInputBuffer->release();
|
||||
mInputBuffer = NULL;
|
||||
|
||||
Reference in New Issue
Block a user