--- branches/dev-api-3/xvidcore/src/image/image.c	2002/12/10 11:13:50	702
+++ branches/dev-api-3/xvidcore/src/image/image.c	2003/01/11 14:59:24	769
@@ -476,6 +476,48 @@
 }
 
 
+/*
+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(DPRINTF_DEBUG,"chroma_optimized_pixels = %i/%i", pixels, width*height/4);
+}
+
+
+
+
 
 /*
   perform safe packed colorspace conversion, by splitting
@@ -635,14 +677,14 @@
 
 	case XVID_CSP_I420:
 		yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2,
-			src, src + width*height, src + width*height + width2*height2,
-			width, width2, width, height, (csp & XVID_CSP_VFLIP));
+			src, src + src_stride*height, src + src_stride*height + (src_stride/2)*height2,
+			src_stride, src_stride/2, width, height, (csp & XVID_CSP_VFLIP));
 		break
 			;
 	case XVID_CSP_YV12:		/* u/v swapped */
 		yv12_to_yv12(image->y, image->v, image->u, edged_width, edged_width2,
-			src, src + width*height, src + width*height + width2*height2,
-			width, width2, width, height, (csp & XVID_CSP_VFLIP));
+			src, src + src_stride*height, src + src_stride*height + (src_stride/2)*height2,
+			src_stride, src_stride/2, width, height, (csp & XVID_CSP_VFLIP));
 		break;
 
 	case XVID_CSP_USER:
@@ -723,7 +765,6 @@
 			 int interlacing)
 {
 	const int edged_width2 = edged_width/2;
-	int width2 = width/2;
 	int height2 = height/2;
 
 /*
@@ -832,15 +873,15 @@
 		return 0;
 
 	case XVID_CSP_I420:
-		yv12_to_yv12(dst, dst + width*height, dst + width*height + width2*height2,
-			width, width2,
+		yv12_to_yv12(dst, dst + dst_stride*height, dst + dst_stride*height + (dst_stride/2)*height2,
+			dst_stride, dst_stride/2,
 			image->y, image->u, image->v, edged_width, edged_width2,
 			width, height, (csp & XVID_CSP_VFLIP));
 		return 0;
 
 	case XVID_CSP_YV12:		// u,v swapped
-		yv12_to_yv12(dst, dst + width*height, dst + width*height + width2*height2,
-			width, width2,
+		yv12_to_yv12(dst, dst + dst_stride*height, dst + dst_stride*height + (dst_stride/2)*height2,
+			dst_stride, dst_stride/2,
 			image->y, image->v, image->u, edged_width, edged_width2,
 			width, height, (csp & XVID_CSP_VFLIP));
 		return 0;
@@ -993,7 +1034,6 @@
 }
 
 
-#define ABS(X)    (((X)>0)?(X):-(X))
 float
 image_mad(const IMAGE * img1,
 		  const IMAGE * img2,
@@ -1058,66 +1098,105 @@
 }
 
 
-/* reduced resolution deblocking filter */
+void
+image_clear(IMAGE * img, int width, int height, int edged_width,
+					int y, int u, int v)
+{
+	uint8_t * p;
+	int i;
+
+	p = img->y;
+	for (i = 0; i < height; i++) {
+		memset(p, y, width);
+		p += edged_width;
+	}
+
+	p = img->u;
+	for (i = 0; i < height/2; i++) {
+		memset(p, u, width/2);
+		p += edged_width/2;
+	}
+
+	p = img->v;
+	for (i = 0; i < height/2; i++) {
+		memset(p, v, width/2);
+		p += edged_width/2;
+	}
+}
+
+
+/* reduced resolution deblocking filter 
+	block = block size (16=rrv, 8=full resolution)
+	flags = XVID_DEC_YDEBLOCK|XVID_DEC_UVDEBLOCK
+*/
 void
 image_deblock_rrv(IMAGE * img, int edged_width,
-				const MACROBLOCK * mbs, int mb_width, int mb_height, int mb_stride)
+				const MACROBLOCK * mbs, int mb_width, int mb_height, int mb_stride,
+				int block, int flags)
 {
 	const int edged_width2 = edged_width /2;
+	const int nblocks = block / 8;	/* skals code uses 8pixel block uints */
 	int i,j;
 
-	/* horizontal deblocking */
-
-	for (j = 1; j < mb_height*2; j++)	// luma: j,i in block units
-	for (i = 0; i < mb_width*2; i++)
+	/* luma: j,i in block units */
+	if ((flags & XVID_DEC_DEBLOCKY))
 	{
-		if (mbs[(j-1)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED ||
-			mbs[(j+0)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED)
+		for (j = 1; j < mb_height*2; j++)		/* horizontal deblocking */
+		for (i = 0; i < mb_width*2; i++)
 		{
-			xvid_HFilter_31_C(img->y + (j*16 - 1)*edged_width + i*16,
-							  img->y + (j*16 + 0)*edged_width + i*16, 2);
+			if (mbs[(j-1)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED ||
+				mbs[(j+0)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED)
+			{
+				hfilter_31(img->y + (j*block - 1)*edged_width + i*block,
+								  img->y + (j*block + 0)*edged_width + i*block, nblocks);
+			}
 		}
-	}
 
-	for (j = 1; j < mb_height; j++)	// chroma
-	for (i = 0; i < mb_width; i++)
-	{
-		if (mbs[(j-1)*mb_stride + i].mode != MODE_NOT_CODED || 
-			mbs[(j+0)*mb_stride + i].mode != MODE_NOT_CODED)
+		for (j = 0; j < mb_height*2; j++)		/* vertical deblocking */
+		for (i = 1; i < mb_width*2; i++)
 		{
-			hfilter_31(img->u + (j*16 - 1)*edged_width2 + i*16,
-					   img->u + (j*16 + 0)*edged_width2 + i*16, 2);
-			hfilter_31(img->v + (j*16 - 1)*edged_width2 + i*16,
-					   img->v + (j*16 + 0)*edged_width2 + i*16, 2);
+			if (mbs[(j/2)*mb_stride + (i-1)/2].mode != MODE_NOT_CODED ||
+				mbs[(j/2)*mb_stride + (i+0)/2].mode != MODE_NOT_CODED)
+			{
+				vfilter_31(img->y + (j*block)*edged_width + i*block - 1,
+						   img->y + (j*block)*edged_width + i*block + 0,
+						   edged_width, nblocks);
+			}
 		}
 	}
 
-	/* vertical deblocking */
 
-	for (j = 0; j < mb_height*2; j++)		// luma: i,j in block units
-	for (i = 1; i < mb_width*2; i++)
+	/* chroma */
+	if ((flags & XVID_DEC_DEBLOCKUV))
 	{
-		if (mbs[(j/2)*mb_stride + (i-1)/2].mode != MODE_NOT_CODED ||
-			mbs[(j/2)*mb_stride + (i+0)/2].mode != MODE_NOT_CODED)
+		for (j = 1; j < mb_height; j++)		/* horizontal deblocking */
+		for (i = 0; i < mb_width; i++)
 		{
-			vfilter_31(img->y + (j*16)*edged_width + i*16 - 1,
-					   img->y + (j*16)*edged_width + i*16 + 0,
-					   edged_width, 2);
+			if (mbs[(j-1)*mb_stride + i].mode != MODE_NOT_CODED || 
+				mbs[(j+0)*mb_stride + i].mode != MODE_NOT_CODED)
+			{
+				hfilter_31(img->u + (j*block - 1)*edged_width2 + i*block,
+						   img->u + (j*block + 0)*edged_width2 + i*block, nblocks);
+				hfilter_31(img->v + (j*block - 1)*edged_width2 + i*block,
+						   img->v + (j*block + 0)*edged_width2 + i*block, nblocks);
+			}
 		}
-	}
 
-	for (j = 0; j < mb_height; j++)	// chroma
-	for (i = 1; i < mb_width; i++)
-	{
-		if (mbs[j*mb_stride + i - 1].mode != MODE_NOT_CODED ||
-			mbs[j*mb_stride + i + 0].mode != MODE_NOT_CODED) 
+		for (j = 0; j < mb_height; j++)		/* vertical deblocking */	
+		for (i = 1; i < mb_width; i++)
 		{
-			vfilter_31(img->u + (j*16)*edged_width2 + i*16 - 1,
-					   img->u + (j*16)*edged_width2 + i*16 + 0,
-					   edged_width2, 2);
-			vfilter_31(img->v + (j*16)*edged_width2 + i*16 - 1,
-					   img->v + (j*16)*edged_width2 + i*16 + 0,
-					   edged_width2, 2);
+			if (mbs[j*mb_stride + i - 1].mode != MODE_NOT_CODED ||
+				mbs[j*mb_stride + i + 0].mode != MODE_NOT_CODED) 
+			{
+				vfilter_31(img->u + (j*block)*edged_width2 + i*block - 1,
+						   img->u + (j*block)*edged_width2 + i*block + 0,
+						   edged_width2, nblocks);
+				vfilter_31(img->v + (j*block)*edged_width2 + i*block - 1,
+						   img->v + (j*block)*edged_width2 + i*block + 0,
+						   edged_width2, nblocks);
+			}
 		}
 	}
+
 }
+