image.c

用ADI DSP汇编指令编写的全内插函数（包括水平、垂直、交叉内插）

			interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width2, rounding);
			interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width2, rounding);
			interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width2, rounding);

			n_ptr += 8;
			h_ptr += 8;
			v_ptr += 8;
			hv_ptr += 8;
		}
		h_ptr += stride_add2;
		v_ptr += stride_add2;
		hv_ptr += stride_add2;
		n_ptr += stride_add2;
	}
#endif
*/
	/*
	   interpolate_halfpel_h(
	   refh->y - offset,
	   refn->y - offset,
	   edged_width, edged_height,
	   rounding);

	   interpolate_halfpel_v(
	   refv->y - offset,
	   refn->y - offset,
	   edged_width, edged_height,
	   rounding);

	   interpolate_halfpel_hv(
	   refhv->y - offset,
	   refn->y - offset,
	   edged_width, edged_height,
	   rounding);
	 */

	/* uv-image-based compensation
	   offset = EDGE_SIZE2 * (edged_width / 2 + 1);

	   interpolate_halfpel_h(
	   refh->u - offset,
	   refn->u - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);

	   interpolate_halfpel_v(
	   refv->u - offset,
	   refn->u - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);

	   interpolate_halfpel_hv(
	   refhv->u - offset,
	   refn->u - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);


	   interpolate_halfpel_h(
	   refh->v - offset,
	   refn->v - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);

	   interpolate_halfpel_v(
	   refv->v - offset,
	   refn->v - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);

	   interpolate_halfpel_hv(
	   refhv->v - offset,
	   refn->v - offset,
	   edged_width / 2, edged_height / 2,
	   rounding);
	 */
}


/*
chroma optimize filter, invented by mf
a chroma pixel is average from the surrounding pixels, when the
correpsonding luma pixels are pure black or white.
*/

void
image_chroma_optimize(IMAGE * img, int width, int height, int edged_width)
{
	int x,y;
	int pixels = 0;

	for (y = 1; y < height/2 - 1; y++)
	for (x = 1; x < width/2 - 1; x++)
	{
#define IS_PURE(a)  ((a)<=16||(a)>=235)
#define IMG_Y(Y,X)	img->y[(Y)*edged_width + (X)]
#define IMG_U(Y,X)	img->u[(Y)*edged_width/2 + (X)]
#define IMG_V(Y,X)	img->v[(Y)*edged_width/2 + (X)]

		if (IS_PURE(IMG_Y(y*2  ,x*2  )) &&
			IS_PURE(IMG_Y(y*2  ,x*2+1)) &&
			IS_PURE(IMG_Y(y*2+1,x*2  )) &&
			IS_PURE(IMG_Y(y*2+1,x*2+1)))
		{
			IMG_U(y,x) = (IMG_U(y,x-1) + IMG_U(y-1, x) + IMG_U(y, x+1) + IMG_U(y+1, x)) / 4;
			IMG_V(y,x) = (IMG_V(y,x-1) + IMG_V(y-1, x) + IMG_V(y, x+1) + IMG_V(y+1, x)) / 4;
			pixels++;
		}

#undef IS_PURE
#undef IMG_Y
#undef IMG_U
#undef IMG_V
	}

	DPRINTF(XVID_DEBUG_DEBUG,"chroma_optimized_pixels = %i/%i\n", pixels, width*height/4);
}





/*
  perform safe packed colorspace conversion, by splitting
  the image up into an optimized area (pixel width divisible by 16),
  and two unoptimized/plain-c areas (pixel width divisible by 2)
*/

static void
safe_packed_conv(uint8_t * x_ptr, int x_stride,
				 uint8_t * y_ptr, uint8_t * u_ptr, uint8_t * v_ptr,
				 int y_stride, int uv_stride,
				 int width, int height, int vflip,
				 packedFunc * func_opt, packedFunc func_c, int size)
{
	int width_opt, width_c;

	if (func_opt != func_c && x_stride < size*((width+15)/16)*16)
	{
		width_opt = width & (~15);
		width_c = width - width_opt;
	}
	else
	{
		width_opt = width;
		width_c = 0;
	};
//	func_opt(x_ptr, x_stride,
//			y_ptr, u_ptr, v_ptr, y_stride, uv_stride,
//			width_opt, height, vflip);        
////////////////////////////////////////////////////////
       UYVYtoYYUV( x_ptr, x_stride,
			y_ptr, u_ptr, v_ptr, y_stride, uv_stride,
			width_opt, height, vflip);
////////////////////////////////////////////////////////
	if (width_c)
	{
		func_c(x_ptr + size*width_opt, x_stride,
			y_ptr + width_opt, u_ptr + width_opt/2, v_ptr + width_opt/2,
			y_stride, uv_stride, width_c, height, vflip);
	}
}


int
image_input(IMAGE * image,
			uint32_t width,
			int height,
			uint32_t edged_width,
			uint8_t * src[4],
			int src_stride[4],
			int csp,
			int interlacing)
{
	const int edged_width2 = edged_width/2;
	const int width2 = width/2;
	const int height2 = height/2;
#if 0
	const int height_signed = (csp & XVID_CSP_VFLIP) ? -height : height;
#endif

	switch (csp & ~XVID_CSP_VFLIP) {
	case XVID_CSP_RGB555:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?rgb555i_to_yv12  :rgb555_to_yv12,
			interlacing?rgb555i_to_yv12_c:rgb555_to_yv12_c, 2);
		break;

	case XVID_CSP_RGB565:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?rgb565i_to_yv12  :rgb565_to_yv12,
			interlacing?rgb565i_to_yv12_c:rgb565_to_yv12_c, 2);
		break;


	case XVID_CSP_BGR:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?bgri_to_yv12  :bgr_to_yv12,
			interlacing?bgri_to_yv12_c:bgr_to_yv12_c, 3);
		break;

	case XVID_CSP_BGRA:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?bgrai_to_yv12  :bgra_to_yv12,
			interlacing?bgrai_to_yv12_c:bgra_to_yv12_c, 4);
		break;

	case XVID_CSP_ABGR :
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?abgri_to_yv12  :abgr_to_yv12,
			interlacing?abgri_to_yv12_c:abgr_to_yv12_c, 4);
		break;

	case XVID_CSP_RGBA :
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?rgbai_to_yv12  :rgba_to_yv12,
			interlacing?rgbai_to_yv12_c:rgba_to_yv12_c, 4);
		break;
            
	case XVID_CSP_ARGB:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?argbi_to_yv12  : argb_to_yv12,
			interlacing?argbi_to_yv12_c: argb_to_yv12_c, 4);
		break;

	case XVID_CSP_YUY2:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yuyvi_to_yv12  :yuyv_to_yv12,
			interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2);
		break;

	case XVID_CSP_YVYU:		/* u/v swapped */
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->v, image->u,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yuyvi_to_yv12  :yuyv_to_yv12,
			interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2);
		break;

	case XVID_CSP_UYVY:
		safe_packed_conv(
			src[0], src_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?uyvyi_to_yv12  :uyvy_to_yv12,
			interlacing?uyvyi_to_yv12_c:uyvy_to_yv12_c, 2);
		break;

	case XVID_CSP_I420:	/* YCbCr == YUV == internal colorspace for MPEG */
		yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2,
			src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2,
			src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP));
		break;

	case XVID_CSP_YV12: /* YCrCb == YVA == U and V plane swapped */
		yv12_to_yv12(image->y, image->v, image->u, edged_width, edged_width2,
			src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2,
			src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP));
		break;

	case XVID_CSP_PLANAR:  /* YCbCr with arbitrary pointers and different strides for Y and UV */
		yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2,
			src[0], src[1], src[2], src_stride[0], src_stride[1],  /* v: dst_stride[2] not yet supported */
			width, height, (csp & XVID_CSP_VFLIP));
		break;

	case XVID_CSP_NULL:
		break;

	default :
		return -1;
	}


	/* pad out image when the width and/or height is not a multiple of 16 */

	if (width & 15)
	{
		int i;
		int pad_width = 16 - (width&15);
		for (i = 0; i < height; i++)
		{
			memset(image->y + i*edged_width + width,
				 *(image->y + i*edged_width + width - 1), pad_width);
		}
		for (i = 0; i < height/2; i++)
		{
			memset(image->u + i*edged_width2 + width2,
				 *(image->u + i*edged_width2 + width2 - 1),pad_width/2);
			memset(image->v + i*edged_width2 + width2,
				 *(image->v + i*edged_width2 + width2 - 1),pad_width/2);
		}
	}

	if (height & 15)
	{
		int pad_height = 16 - (height&15);
		int length = ((width+15)/16)*16;
		int i;
		for (i = 0; i < pad_height; i++)
		{
			memcpy(image->y + (height+i)*edged_width,
				   image->y + (height-1)*edged_width,length);
		}

		for (i = 0; i < pad_height/2; i++)
		{
			memcpy(image->u + (height2+i)*edged_width2,
				   image->u + (height2-1)*edged_width2,length/2);
			memcpy(image->v + (height2+i)*edged_width2,
				   image->v + (height2-1)*edged_width2,length/2);
		}
	}

/*
	if (interlacing)
		image_printf(image, edged_width, height, 5,5, "[i]");
	image_dump_yuvpgm(image, edged_width, ((width+15)/16)*16, ((height+15)/16)*16, "\\encode.pgm");
*/
	return 0;
}



int
image_output(IMAGE * image,
			 uint32_t width,
			 int height,
			 uint32_t edged_width,
			 uint8_t * dst[4],
			 uint32_t dst_stride[4],
			 int csp,
			 int interlacing)
{
	const int edged_width2 = edged_width/2;
	int height2 = height/2;

/*
	if (interlacing)
		image_printf(image, edged_width, height, 5,100, "[i]=%i,%i",width,height);
	image_dump_yuvpgm(image, edged_width, width, height, "\\decode.pgm");
*/

	switch (csp & ~XVID_CSP_VFLIP) {
	case XVID_CSP_RGB555:
		safe_packed_conv(
			dst[0], dst_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yv12_to_rgb555i  :yv12_to_rgb555,
			interlacing?yv12_to_rgb555i_c:yv12_to_rgb555_c, 2);
		return 0;

	case XVID_CSP_RGB565:
		safe_packed_conv(
			dst[0], dst_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yv12_to_rgb565i  :yv12_to_rgb565,
			interlacing?yv12_to_rgb565i_c:yv12_to_rgb565_c, 2);
		return 0;

    case XVID_CSP_BGR:
		safe_packed_conv(
			dst[0], dst_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yv12_to_bgri  :yv12_to_bgr,
			interlacing?yv12_to_bgri_c:yv12_to_bgr_c, 3);
		return 0;

	case XVID_CSP_BGRA:
		safe_packed_conv(
			dst[0], dst_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yv12_to_bgrai  :yv12_to_bgra,
			interlacing?yv12_to_bgrai_c:yv12_to_bgra_c, 4);
		return 0;

	case XVID_CSP_ABGR:
		safe_packed_conv(
			dst[0], dst_stride[0], image->y, image->u, image->v,
			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
			interlacing?yv12_to_abgri  :yv12_to_abgr,
			interlacing?yv12_to_abgri_c:yv12_to_abgr_c, 4);
		return 0;

	case XVID_CSP_RGBA:
		safe_packed_conv(