Hi,
|
I use an example from wiki.danicidsp (Accessing pixels in a frame on DM643x). As I understand in this example, three video buffer operates on a ping-pong, that is, one is treated while the other two getting data. I only insert function getting Luma from image and then restoring this image.
//---------------------------------------------
while (!done && status ==0)
FVID_exchange(hGioVpfeCcdc, &frameBuffPtr);
VLIB_extractLumaFromUYUV(input_Ptr,
720, 720,
576,
iPtr);
for (i=1, index=0; i<(720*576)*2; i+=2, index++)
{
*(((unsigned char *)frameBuffPtr->frame.frameBufferPtr) + i) = *(((unsigned char *) iPtr) + index);
}
........
//-------------------------------
It loads processor to 72% and video very "stopped". But I have not even started the desired manipulation of the pixels. This is not good. What to do in this example to increase the speed of processing code or decrease load processor.
Tnanks,
Igor.
|
