--- branches/dev-api-3/xvidcore/src/image/image.c	2002/09/08 09:49:55	449
+++ branches/dev-api-3/xvidcore/src/image/image.c	2002/11/27 14:29:34	681
@@ -3,6 +3,15 @@
  *	XVID MPEG-4 VIDEO CODEC
  *	image stuff
  *
+ *	This program is an implementation of a part of one or more MPEG-4
+ *	Video tools as specified in ISO/IEC 14496-2 standard.  Those intending
+ *	to use this software module in hardware or software products are
+ *	advised that its use may infringe existing patents or copyrights, and
+ *	any such use would be at such party's own risk.  The original
+ *	developer of this software module and his/her company, and subsequent
+ *	editors and their companies, will have no liability for use of this
+ *	software or modifications or derivatives thereof.
+ *
  *	This program is free software; you can redistribute it and/or modify
  *	it under the terms of the GNU General Public License as published by
  *	the Free Software Foundation; either version 2 of the License, or
@@ -23,11 +32,12 @@
  *
  *	History:
  *
+ *  05.10.2002	support for interpolated images in qpel mode - Isibaar
  *	01.05.2002	BFRAME image-based u,v interpolation
  *  22.04.2002  added some B-frame support
  *	14.04.2002	added image_dump_yuvpgm(), added image_mad()
  *              XVID_CSP_USER input support
- *  09.04.2002  PSNR calculations
+ *  09.04.2002  PSNR calculations - Isibaar
  *	06.04.2002	removed interlaced edging from U,V blocks (as per spec)
  *  26.03.2002  interlacing support (field-based edging in set_edges)
  *	26.01.2002	rgb555, rgb565
@@ -44,6 +54,7 @@
 #include <math.h>
 
 #include "../portab.h"
+#include "../global.h"			// XVID_CSP_XXX's
 #include "../xvid.h"			// XVID_CSP_XXX's
 #include "image.h"
 #include "colorspace.h"
@@ -51,6 +62,8 @@
 #include "../divx4.h"
 #include "../utils/mem_align.h"
 
+#include "font.h"		// XXX: remove later
+
 #define SAFETY	64
 #define EDGE_SIZE2  (EDGE_SIZE/2)
 
@@ -151,8 +164,7 @@
 			   uint32_t edged_width,
 			   uint32_t edged_height,
 			   uint32_t width,
-			   uint32_t height,
-			   uint32_t interlacing)
+			   uint32_t height)
 {
 	const uint32_t edged_width2 = edged_width / 2;
 	const uint32_t width2 = width / 2;
@@ -165,18 +177,10 @@
 	src = image->y;
 
 	for (i = 0; i < EDGE_SIZE; i++) {
-/*		// if interlacing, edges contain top-most data from each field
-		if (interlacing && (i & 1)) {
-			memset(dst, *(src + edged_width), EDGE_SIZE);
-			memcpy(dst + EDGE_SIZE, src + edged_width, width);
-			memset(dst + edged_width - EDGE_SIZE,
-				   *(src + edged_width + width - 1), EDGE_SIZE);
-		} else {*/
-			memset(dst, *src, EDGE_SIZE);
-			memcpy(dst + EDGE_SIZE, src, width);
-			memset(dst + edged_width - EDGE_SIZE, *(src + width - 1),
-				   EDGE_SIZE);
-		/*}*/
+		memset(dst, *src, EDGE_SIZE);
+		memcpy(dst + EDGE_SIZE, src, width);
+		memset(dst + edged_width - EDGE_SIZE, *(src + width - 1),
+			   EDGE_SIZE);
 		dst += edged_width;
 	}
 
@@ -189,18 +193,10 @@
 
 	src -= edged_width;
 	for (i = 0; i < EDGE_SIZE; i++) {
-/*		// if interlacing, edges contain bottom-most data from each field
-		if (interlacing && !(i & 1)) {
-			memset(dst, *(src - edged_width), EDGE_SIZE);
-			memcpy(dst + EDGE_SIZE, src - edged_width, width);
-			memset(dst + edged_width - EDGE_SIZE,
-				   *(src - edged_width + width - 1), EDGE_SIZE);
-		} else {*/
-			memset(dst, *src, EDGE_SIZE);
-			memcpy(dst + EDGE_SIZE, src, width);
-			memset(dst + edged_width - EDGE_SIZE, *(src + width - 1),
+		memset(dst, *src, EDGE_SIZE);
+		memcpy(dst + EDGE_SIZE, src, width);
+		memset(dst + edged_width - EDGE_SIZE, *(src + width - 1),
 				   EDGE_SIZE);
-		/*}*/
 		dst += edged_width;
 	}
 
@@ -261,8 +257,7 @@
 	}
 }
 
-
-// image-based y,u,v interpolation
+// bframe encoding requires image-based u,v interpolation
 void
 image_interpolate(const IMAGE * refn,
 				  IMAGE * refh,
@@ -270,18 +265,19 @@
 				  IMAGE * refhv,
 				  uint32_t edged_width,
 				  uint32_t edged_height,
+				  uint32_t quarterpel,
 				  uint32_t rounding)
 {
-	const uint32_t offset = EDGE_SIZE * (edged_width + 1);
+	const uint32_t offset = EDGE_SIZE2 * (edged_width + 1); // we only interpolate half of the edge area
 	const uint32_t stride_add = 7 * edged_width;
-
-	/* --- u,v-image-based interpolation ---
+/*
+#ifdef BFRAMES
 	const uint32_t edged_width2 = edged_width / 2;
 	const uint32_t edged_height2 = edged_height / 2;
 	const uint32_t offset2 = EDGE_SIZE2 * (edged_width2 + 1);
 	const uint32_t stride_add2 = 7 * edged_width2;
-	*/
-
+#endif
+*/
 	uint8_t *n_ptr, *h_ptr, *v_ptr, *hv_ptr;
 	uint32_t x, y;
 
@@ -296,24 +292,71 @@
 	v_ptr -= offset;
 	hv_ptr -= offset;
 
-	for (y = 0; y < edged_height; y = y + 8) {
-		for (x = 0; x < edged_width; x = x + 8) {
-			interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width, rounding);
-			interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width, rounding);
-			interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width, rounding);
+	if(quarterpel) {
+		
+		for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) {
+			for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) {
+				interpolate8x8_6tap_lowpass_h(h_ptr, n_ptr, edged_width, rounding);
+				interpolate8x8_6tap_lowpass_v(v_ptr, n_ptr, edged_width, rounding);
+
+				n_ptr += 8;
+				h_ptr += 8;
+				v_ptr += 8;
+			}
+			
+			n_ptr += EDGE_SIZE;
+			h_ptr += EDGE_SIZE;
+			v_ptr += EDGE_SIZE;
+
+			h_ptr += stride_add;
+			v_ptr += stride_add;
+			n_ptr += stride_add;
+		}
 
-			n_ptr += 8;
-			h_ptr += 8;
-			v_ptr += 8;
-			hv_ptr += 8;
+		h_ptr = refh->y;
+		h_ptr -= offset;
+
+		for (y = 0; y < (edged_height - EDGE_SIZE); y = y + 8) {
+			for (x = 0; x < (edged_width - EDGE_SIZE); x = x + 8) {
+				interpolate8x8_6tap_lowpass_v(hv_ptr, h_ptr, edged_width, rounding);
+				hv_ptr += 8;
+				h_ptr += 8;
+			}
+
+			hv_ptr += EDGE_SIZE;
+			h_ptr += EDGE_SIZE;
+
+			hv_ptr += stride_add;
+			h_ptr += stride_add;
 		}
-		h_ptr += stride_add;
-		v_ptr += stride_add;
-		hv_ptr += stride_add;
-		n_ptr += stride_add;
 	}
+	else {
 
-/* --- u,v-image-based interpolation ---
+		for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) {
+			for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) {
+				interpolate8x8_halfpel_h(h_ptr, n_ptr, edged_width, rounding);
+				interpolate8x8_halfpel_v(v_ptr, n_ptr, edged_width, rounding);
+				interpolate8x8_halfpel_hv(hv_ptr, n_ptr, edged_width, rounding);
+
+				n_ptr += 8;
+				h_ptr += 8;
+				v_ptr += 8;
+				hv_ptr += 8;
+			}
+			
+			h_ptr += EDGE_SIZE;
+			v_ptr += EDGE_SIZE;
+			hv_ptr += EDGE_SIZE;
+			n_ptr += EDGE_SIZE;
+
+			h_ptr += stride_add;
+			v_ptr += stride_add;
+			hv_ptr += stride_add;
+			n_ptr += stride_add;
+		}
+	}
+/*
+#ifdef BFRAMES
 	n_ptr = refn->u;
 	h_ptr = refh->u;
 	v_ptr = refv->u;
@@ -324,8 +367,8 @@
 	v_ptr -= offset2;
 	hv_ptr -= offset2;
 
-	for (y = 0; y < edged_height2; y = y + 8) {
-		for (x = 0; x < edged_width2; x = x + 8) {
+	for (y = 0; y < edged_height2; y += 8) {
+		for (x = 0; x < edged_width2; x += 8) {
 			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);
@@ -367,83 +410,303 @@
 		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);
+	 */
+}
+
+
+
+/*
+  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);
+
+	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,
-			int csp)
+			int src_stride,
+			int csp,
+			int interlacing)
 {
+	const int edged_width2 = edged_width/2;
+	const int width2 = width/2;
+	const int height2 = height/2;
+	//const int height_signed = (csp & XVID_CSP_VFLIP) ? -height : height;
+
+	
+	//	int src_stride = width;
 
-/*	if (csp & XVID_CSP_VFLIP)
+	// --- xvid 2.1 compatiblity patch ---
+	// --- remove when xvid_dec_frame->stride equals real stride
+	/*
+	if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB555 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB565 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_YUY2 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_YVYU ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_UYVY)
 	{
-		height = -height;
+		src_stride *= 2;
+	} 
+	else if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB24)
+	{
+		src_stride *= 3;
 	}
-*/
+	else if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB32 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_ABGR ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGBA)
+	{
+		src_stride *= 4;
+	}
+	*/
+	// ^--- xvid 2.1 compatiblity fix ---^
 
 	switch (csp & ~XVID_CSP_VFLIP) {
 	case XVID_CSP_RGB555:
-		rgb555_to_yv12(image->y, image->u, image->v, src, width, height,
-					   edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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:
-		rgb565_to_yv12(image->y, image->u, image->v, src, width, height,
-					   edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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_RGB24:
-		rgb24_to_yv12(image->y, image->u, image->v, src, width, height,
-					  edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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_RGB32:
-		rgb32_to_yv12(image->y, image->u, image->v, src, width, height,
-					  edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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_I420:
-		yuv_to_yv12(image->y, image->u, image->v, src, width, height,
-					edged_width);
-		return 0;
+	case XVID_CSP_ABGR :
+		safe_packed_conv(
+			src, src_stride, 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_YV12:		/* u/v swapped */
-		yuv_to_yv12(image->y, image->v, image->u, src, width, height,
-					edged_width);
-		return 0;
+	case XVID_CSP_RGBA :
+		safe_packed_conv(
+			src, src_stride, 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_YUY2:
-		yuyv_to_yv12(image->y, image->u, image->v, src, width, height,
-					 edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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 */
-		yuyv_to_yv12(image->y, image->v, image->u, src, width, height,
-					 edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, image->y, image->v, image->y, 
+			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:
-		uyvy_to_yv12(image->y, image->u, image->v, src, width, height,
-					 edged_width);
-		return 0;
+		safe_packed_conv(
+			src, src_stride, 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:
+		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));
+		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));
+		break;
 
 	case XVID_CSP_USER:
-		user_to_yuv_c(image->y, image->u, image->v, edged_width,
-					  (DEC_PICTURE *) src, width, height);
-		return 0;
+		{
+			DEC_PICTURE * pic = (DEC_PICTURE*)src;
+			yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2,
+				pic->y, pic->u, pic->v, pic->stride_y, pic->stride_y,
+				width, height, (csp & XVID_CSP_VFLIP));
+		}
+		break;
 
 	case XVID_CSP_NULL:
 		break;
 
+	default :
+		return -1;
 	}
 
-	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;
 }
 
 
@@ -455,64 +718,141 @@
 			 uint32_t edged_width,
 			 uint8_t * dst,
 			 uint32_t dst_stride,
-			 int csp)
+			 int csp,
+			 int interlacing)
 {
-	if (csp & XVID_CSP_VFLIP) {
-		height = -height;
+	const int edged_width2 = edged_width/2;
+	int width2 = 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");
+*/
+
+
+	// --- xvid 2.1 compatiblity patch ---
+	// --- remove when xvid_dec_frame->stride equals real stride
+	/*
+	if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB555 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB565 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_YUY2 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_YVYU ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_UYVY)
+	{
+		dst_stride *= 2;
+	} 
+	else if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB24)
+	{
+		dst_stride *= 3;
 	}
+	else if ((csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGB32 ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_ABGR ||
+		(csp & ~XVID_CSP_VFLIP) == XVID_CSP_RGBA)
+	{
+		dst_stride *= 4;
+	}
+	*/
+	// ^--- xvid 2.1 compatiblity fix ---^
 
+	
 	switch (csp & ~XVID_CSP_VFLIP) {
 	case XVID_CSP_RGB555:
-		yv12_to_rgb555(dst, dst_stride, image->y, image->u, image->v,
-					   edged_width, edged_width / 2, width, height);
+		safe_packed_conv(
+			dst, dst_stride, 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:
-		yv12_to_rgb565(dst, dst_stride, image->y, image->u, image->v,
-					   edged_width, edged_width / 2, width, height);
+		safe_packed_conv(
+			dst, dst_stride, 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_RGB24:
-		yv12_to_rgb24(dst, dst_stride, image->y, image->u, image->v,
-					  edged_width, edged_width / 2, width, height);
+		safe_packed_conv(
+			dst, dst_stride, 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_RGB32:
-		yv12_to_rgb32(dst, dst_stride, image->y, image->u, image->v,
-					  edged_width, edged_width / 2, width, height);
+		safe_packed_conv(
+			dst, dst_stride, 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, dst_stride, 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(
+			dst, dst_stride, image->y, image->u, image->v,
+			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
+			interlacing?yv12_to_rgbai  :yv12_to_rgba,
+			interlacing?yv12_to_rgbai_c:yv12_to_rgba_c, 4);
 		return 0;
 
-	case XVID_CSP_I420:
-		yv12_to_yuv(dst, dst_stride, image->y, image->u, image->v, edged_width,
-					edged_width / 2, width, height);
+	case XVID_CSP_YUY2:
+		safe_packed_conv(
+			dst, dst_stride, image->y, image->u, image->v,
+			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
+			interlacing?yv12_to_yuyvi  :yv12_to_yuyv,
+			interlacing?yv12_to_yuyvi_c:yv12_to_yuyv_c, 2);
 		return 0;
 
-	case XVID_CSP_YV12:		// u,v swapped
-		yv12_to_yuv(dst, dst_stride, image->y, image->v, image->u, edged_width,
-					edged_width / 2, width, height);
+	case XVID_CSP_YVYU:		// u,v swapped
+		safe_packed_conv(
+			dst, dst_stride, image->y, image->v, image->u,
+			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
+			interlacing?yv12_to_yuyvi  :yv12_to_yuyv,
+			interlacing?yv12_to_yuyvi_c:yv12_to_yuyv_c, 2);
 		return 0;
 
-	case XVID_CSP_YUY2:
-		yv12_to_yuyv(dst, dst_stride, image->y, image->u, image->v,
-					 edged_width, edged_width / 2, width, height);
+	case XVID_CSP_UYVY:
+		safe_packed_conv(
+			dst, dst_stride, image->y, image->u, image->v,
+			edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP),
+			interlacing?yv12_to_uyvyi  :yv12_to_uyvy,
+			interlacing?yv12_to_uyvyi_c:yv12_to_uyvy_c, 2);
 		return 0;
 
-	case XVID_CSP_YVYU:		// u,v swapped
-		yv12_to_yuyv(dst, dst_stride, image->y, image->v, image->u,
-					 edged_width, edged_width / 2, width, height);
+	case XVID_CSP_I420:
+		yv12_to_yv12(dst, dst + width*height, dst + width*height + width2*height2,
+			width, width2,
+			image->y, image->u, image->v, edged_width, edged_width2,
+			width, height, (csp & XVID_CSP_VFLIP));
 		return 0;
 
-	case XVID_CSP_UYVY:
-		yv12_to_uyvy(dst, dst_stride, image->y, image->u, image->v,
-					 edged_width, edged_width / 2, width, height);
+	case XVID_CSP_YV12:		// u,v swapped
+		yv12_to_yv12(dst, dst + width*height, dst + width*height + width2*height2,
+			width, width2,
+			image->y, image->v, image->u, edged_width, edged_width2,
+			width, height, (csp & XVID_CSP_VFLIP));
 		return 0;
 
 	case XVID_CSP_USER:
-		((DEC_PICTURE *) dst)->y = image->y;
-		((DEC_PICTURE *) dst)->u = image->u;
-		((DEC_PICTURE *) dst)->v = image->v;
-		((DEC_PICTURE *) dst)->stride_y = edged_width;
-		((DEC_PICTURE *) dst)->stride_uv = edged_width / 2;
+		{
+			DEC_PICTURE * pic = (DEC_PICTURE*)dst;
+			pic->y = image->y;
+			pic->u = image->u;
+			pic->v = image->v;
+			pic->stride_y = edged_width;
+			pic->stride_uv = edged_width / 2;
+		}
 		return 0;
 
 	case XVID_CSP_NULL: