diff --git a/plugins/zzogl-pg/opengl/HostMemory.cpp b/plugins/zzogl-pg/opengl/HostMemory.cpp
index 5aaa23ed89..2f45eab73d 100644
--- a/plugins/zzogl-pg/opengl/HostMemory.cpp
+++ b/plugins/zzogl-pg/opengl/HostMemory.cpp
@@ -327,6 +327,7 @@
 
 		if (gs.imageY >= gs.imageEndY)
 		{
+			ZZLog::Error_Log("gs.imageY >= gs.imageEndY!");
 			assert(gs.imageY == gs.imageEndY);
 			gs.imageTransfer = -1;
 		}
@@ -335,38 +336,35 @@
 __forceinline void _TransferLocalLocal()
 {
 	//ZZLog::Error_Log("TransferLocalLocal(0x%x, 0x%x)", gs.srcbuf.psm, gs.dstbuf.psm);
-	_writePixel_0 wp = writePixelFun_0[gs.srcbuf.psm]; //writePixelFunction_0(gs.srcbuf.psm);
-	_readPixel_0 rp = readPixelFun_0[gs.dstbuf.psm]; //readPixelFunction_0(gs.dstbuf.psm);
+	_writePixel_0 wp = writePixelFun_0[gs.srcbuf.psm];
+	_readPixel_0 rp = readPixelFun_0[gs.dstbuf.psm];
 	u8* pSrcBuf = g_pbyGSMemory + gs.srcbuf.bp * 256;
 	u8* pDstBuf = g_pbyGSMemory + gs.dstbuf.bp * 256;
 	u32 widthlimit = 4;
+	u32 maxX = gs.trxpos.sx + gs.imageWnew;
+	u32 maxY = gs.trxpos.sy + gs.imageHnew;
 	
 	if (PSMT_BITMODE(gs.srcbuf.psm) == 0) widthlimit = 2;
 	if ((gs.imageWnew & widthlimit) != 0) return;
 	
-	for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy+gs.imageHnew; i++, i2++) 
+	for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < maxY; i++, i2++) 
 	{
-		for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx+gs.imageWnew; j+=widthlimit, j2+=widthlimit)
+		for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < maxX; j += widthlimit, j2 += widthlimit)
 		{
 			wp(pDstBuf, j2%2048, i2%2048,
-					rp(pSrcBuf, j%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
+				rp(pSrcBuf, j%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
+			
+			wp(pDstBuf, (j2+1)%2048, i2%2048,
+				rp(pSrcBuf, (j+1)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
 				
-			//if (widthlimit > 1) 
+			if (widthlimit > 2) 
 			{
-				wp(pDstBuf, (j2+1)%2048, i2%2048,
-					rp(pSrcBuf, (j+1)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
-				
-				if (widthlimit > 2) 
-				{
-					wp(pDstBuf, (j2+2)%2048, i2%2048,
-						rp(pSrcBuf, (j+2)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
-				
-					//if (widthlimit > 3) 
-					{
-						wp(pDstBuf, (j2+3)%2048, i2%2048,
-							rp(pSrcBuf, (j+3)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
-					}
-				}
+				// Then widthlimit == 4.
+				wp(pDstBuf, (j2+2)%2048, i2%2048,
+					rp(pSrcBuf, (j+2)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
+						
+				wp(pDstBuf, (j2+3)%2048, i2%2048,
+					rp(pSrcBuf, (j+3)%2048, i%2048, gs.srcbuf.bw), gs.dstbuf.bw);
 			}
 		}
 	}
@@ -375,16 +373,18 @@ __forceinline void _TransferLocalLocal()
 __forceinline void _TransferLocalLocal_4()
 {
 	//ZZLog::Error_Log("TransferLocalLocal_4(0x%x, 0x%x)", gs.srcbuf.psm, gs.dstbuf.psm);
-	_getPixelAddress_0 gsp = getPixelFun_0[gs.srcbuf.psm]; //getPixelFunction_0(gs.srcbuf.psm);
-	_getPixelAddress_0 gdp = getPixelFun_0[gs.dstbuf.psm]; //getPixelFunction_0(gs.dstbuf.psm);
+	_getPixelAddress_0 gsp = getPixelFun_0[gs.srcbuf.psm];
+	_getPixelAddress_0 gdp = getPixelFun_0[gs.dstbuf.psm];
 	u8* pSrcBuf = g_pbyGSMemory + gs.srcbuf.bp * 256;
 	u8* pDstBuf = g_pbyGSMemory + gs.dstbuf.bp * 256;
+	u32 maxX = gs.trxpos.sx + gs.imageWnew;
+	u32 maxY = gs.trxpos.sy + gs.imageHnew;
 
 	assert((gs.imageWnew % 8) == 0);
 		
-	for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < gs.trxpos.sy + gs.imageHnew; ++i, ++i2) 
+	for(int i = gs.trxpos.sy, i2 = gs.trxpos.dy; i < maxY; ++i, ++i2) 
 	{
-		for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < gs.trxpos.sx + gs.imageWnew; j += 8, j2 += 8) 
+		for(int j = gs.trxpos.sx, j2 = gs.trxpos.dx; j < maxX; j += 8, j2 += 8) 
 		{
 			/* NOTE: the 2 conseq 4bit values are in NOT in the same byte */
 			u32 read = gsp(j%2048, i%2048, gs.srcbuf.bw);
diff --git a/plugins/zzogl-pg/opengl/Mem.h b/plugins/zzogl-pg/opengl/Mem.h
index 17cd34a1fb..7b347078fc 100644
--- a/plugins/zzogl-pg/opengl/Mem.h
+++ b/plugins/zzogl-pg/opengl/Mem.h
@@ -270,33 +270,9 @@ static __forceinline u32 getPixelAddress16SZ_0(int x, int y, u32 bw)
 	return word;
 }
 
-static __forceinline _getPixelAddress_0 getPixelFunction_0(u32 psm)
-{
-	return getPixelFun_0[psm];
-	/*switch(psm)
-	{
-		case PSMCT32: return getPixelAddress32_0;
-		case PSMCT24: return getPixelAddress24_0;
-		case PSMCT16: return getPixelAddress16_0;
-		case PSMCT16S: return getPixelAddress16S_0;
-		case PSMT8: return getPixelAddress8_0;
-		case PSMT4: return getPixelAddress4_0;
-		case PSMT8H: return getPixelAddress8H_0;
-		case PSMT4HL: return getPixelAddress4HL_0;
-		case PSMT4HH: return getPixelAddress4HH_0;
-		case PSMT32Z: return getPixelAddress32Z_0;
-		case PSMT24Z: return getPixelAddress24Z_0;
-		case PSMT16Z: return getPixelAddress16Z_0;
-		case PSMT16SZ: return getPixelAddress16SZ_0;
-		default:  return getPixelAddress32_0;
-	}*/
-}
-
 #define getPixelAddress_0(psm,x,y,bw) getPixelAddress##psm##_0(x,y,bw)
 #define getPixelAddress(psm,x,y,bp,bw) getPixelAddress##psm##(x,y,bp,bw)
 
-
-
 static __forceinline void writePixel32(void* pmem, int x, int y, u32 pixel, u32 bp, u32 bw)
 {
 	((u32*)pmem)[getPixelAddress32(x, y, bp, bw)] = pixel;
@@ -537,29 +513,6 @@ static __forceinline void writePixel16SZ_0(void* pmem, int x, int y, u32 pixel,
 	((u16*)pmem)[getPixelAddress16SZ_0(x, y, bw)] = pixel;
 }
 
-
-static __forceinline _writePixel_0 writePixelFunction_0(u32 psm)
-{
-	return writePixelFun_0[psm];
-	/*switch(psm)
-	{
-		case PSMCT32: return writePixel32_0;
-		case PSMCT24: return writePixel24_0;
-		case PSMCT16: return writePixel16_0;
-		case PSMCT16S: return writePixel16S_0;
-		case PSMT8: return writePixel8_0;
-		case PSMT4: return writePixel4_0;
-		case PSMT8H: return writePixel8H_0;
-		case PSMT4HL: return writePixel4HL_0;
-		case PSMT4HH: return writePixel4HH_0;
-		case PSMT32Z: return writePixel32Z_0;
-		case PSMT24Z: return writePixel24Z_0;
-		case PSMT16Z: return writePixel16Z_0;
-		case PSMT16SZ: return writePixel16SZ_0;
-		default:  return writePixel32_0;
-	}*/
-}
-
 ///////////////
 
 static __forceinline u32 readPixel32_0(const void* pmem, int x, int y, u32 bw)
@@ -637,25 +590,4 @@ static __forceinline u32 readPixel16SZ_0(const void* pmem, int x, int y, u32 bw)
 	return ((const u16*)pmem)[getPixelAddress16SZ_0(x, y, bw)];
 }
 
-static __forceinline _readPixel_0 readPixelFunction_0(u32 psm)
-{
-	return readPixelFun_0[psm];
-	/*switch(psm)
-	{
-		case PSMCT32: return readPixel32_0;
-		case PSMCT24: return readPixel24_0;
-		case PSMCT16: return readPixel16_0;
-		case PSMCT16S: return readPixel16S_0;
-		case PSMT8: return readPixel8_0;
-		case PSMT4: return readPixel4_0;
-		case PSMT8H: return readPixel8H_0;
-		case PSMT4HL: return readPixel4HL_0;
-		case PSMT4HH: return readPixel4HH_0;
-		case PSMT32Z: return readPixel32Z_0;
-		case PSMT24Z: return readPixel24Z_0;
-		case PSMT16Z: return readPixel16Z_0;
-		case PSMT16SZ: return readPixel16SZ_0;
-		default:  return readPixel32_0;
-	}*/
-}
 #endif /* __MEM_H__ */
diff --git a/plugins/zzogl-pg/opengl/targets.cpp b/plugins/zzogl-pg/opengl/targets.cpp
index 391452608c..d24376eb38 100644
--- a/plugins/zzogl-pg/opengl/targets.cpp
+++ b/plugins/zzogl-pg/opengl/targets.cpp
@@ -2286,10 +2286,12 @@ ZeroGS::CMemoryTarget* ZeroGS::CMemoryTargetMngr::GetMemoryTarget(const tex0Info
 			{
 				// This is not unusual situation, when vector<u8> does not 16bit alignment, that is destructive for SSE2
 				// instruction movdqa [%eax], xmm0
-				// The idea would be resise vector to 15 elements, that set ptxedata to aligned position.
+				// The idea would be resize vector to 15 elements, that set ptxedata to aligned position.
 				// Later we would move eax by 16, so only we should verify is first element align
 				// FIXME. As I see, texdata used only once here, it does not have any impact on other code.
 				// Probably, usage of _aligned_maloc() would be preferable.
+				
+				// Note: this often happens when changing AA.
 				int disalignment = 16 - ((u32)(uptr)dst) % 16;		// This is value of shift. It could be 0 < disalignment <= 15
 				ptexdata = &texdata[disalignment];			// Set pointer to aligned element
 				dst = (u16*)ptexdata;
diff --git a/plugins/zzogl-pg/opengl/zerogs.cpp b/plugins/zzogl-pg/opengl/zerogs.cpp
index 2d04ed3e27..c311e7771b 100644
--- a/plugins/zzogl-pg/opengl/zerogs.cpp
+++ b/plugins/zzogl-pg/opengl/zerogs.cpp
@@ -803,15 +803,10 @@ void ZeroGS::KickSprite()
 
 	int next = (gs.primIndex + 1) % ARRAY_SIZE(gs.gsvertex);
 	int last = (gs.primIndex + 2) % ARRAY_SIZE(gs.gsvertex);
-
-	// sprite is too small and AA shows lines (tek4)
-
-	if (s_AAx)
-	{
-		gs.gsvertex[last].x += 4;
-
-		if (s_AAy) gs.gsvertex[last].y += 4;
-	}
+	
+	// sprite is too small and AA shows lines (tek4, Mana Khemia)
+	gs.gsvertex[last].x += (4*s_AAx);
+	gs.gsvertex[last].y += (4*s_AAy);
 
 	// might be bad sprite (KH dialog text)
 	//if( gs.gsvertex[next].x == gs.gsvertex[last].x || gs.gsvertex[next].y == gs.gsvertex[last].y )