diff --git a/GPU2D.cpp b/GPU2D.cpp
index b7a07807..9a2c2e96 100644
--- a/GPU2D.cpp
+++ b/GPU2D.cpp
@@ -1200,11 +1200,12 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
     u32 wmask = width - 8; // really ((width - 1) & ~0x7)
 
     u32 xoff;
+    u32 xend = width;
     if (xpos >= 0)
     {
         xoff = 0;
-        if ((xpos+width) > 256)
-            width = 256-xpos;
+        if ((xpos+xend) > 256)
+            xend = 256-xpos;
     }
     else
     {
@@ -1249,7 +1250,7 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
         u32 pixelsaddr = (Num ? 0x06600000 : 0x06400000) + tilenum;
         pixelsaddr += (xoff << 1);
 
-        for (; xoff < width;)
+        for (; xoff < xend;)
         {
             u16 color = GPU::ReadVRAM_OBJ<u16>(pixelsaddr);
             pixelsaddr += 2;
@@ -1291,7 +1292,7 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
                 pixelsaddr += (((width-1 - xoff) & wmask) << 3);
                 pixelsaddr += ((width-1 - xoff) & 0x7);
 
-                for (; xoff < width;)
+                for (; xoff < xend;)
                 {
                     u8 color = GPU::ReadVRAM_OBJ<u8>(pixelsaddr);
                     pixelsaddr--;
@@ -1309,7 +1310,7 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
                 pixelsaddr += ((xoff & wmask) << 3);
                 pixelsaddr += (xoff & 0x7);
 
-                for (; xoff < width;)
+                for (; xoff < xend;)
                 {
                     u8 color = GPU::ReadVRAM_OBJ<u8>(pixelsaddr);
                     pixelsaddr++;
@@ -1338,7 +1339,7 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
                 pixelsaddr += (((width-1 - xoff) & wmask) << 2);
                 pixelsaddr += (((width-1 - xoff) & 0x7) >> 1);
 
-                for (; xoff < width;)
+                for (; xoff < xend;)
                 {
                     u8 color;
                     if (xoff & 0x1)
@@ -1364,7 +1365,7 @@ void GPU2D::DrawSprite_Normal(u16* attrib, u32 width, s32 xpos, u32 ypos, u32* d
                 pixelsaddr += ((xoff & wmask) << 2);
                 pixelsaddr += ((xoff & 0x7) >> 1);
 
-                for (; xoff < width;)
+                for (; xoff < xend;)
                 {
                     u8 color;
                     if (xoff & 0x1)
diff --git a/GPU3D_Soft.cpp b/GPU3D_Soft.cpp
index 54189801..7c7613ca 100644
--- a/GPU3D_Soft.cpp
+++ b/GPU3D_Soft.cpp
@@ -94,6 +94,106 @@ void TextureLookup(u32 texparam, u32 texpal, s16 s, s16 t, u8* r, u8* g, u8* b)
         }
         break;
 
+    case 5: // compressed
+        {
+            vramaddr += ((t & 0x3FC) * (width>>2)) + (s & 0x3FC);
+            vramaddr += (t & 0x3);
+
+            u32 slot1addr = 0x20000 + ((vramaddr & 0x1FFFC) >> 1);
+            if (vramaddr >= 0x40000)
+                slot1addr += 0x10000;
+
+            u8 val = GPU::ReadVRAM_Texture<u8>(vramaddr);
+            val >>= (2 * (s & 0x3));
+
+            u16 palinfo = GPU::ReadVRAM_Texture<u16>(slot1addr);
+            u32 paloffset = (palinfo & 0x3FFF) << 2;
+            texpal <<= 4;
+
+            u16 color;
+            switch (val & 0x3)
+            {
+            case 0:
+                color = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset);
+                break;
+
+            case 1:
+                color = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 2);
+                break;
+
+            case 2:
+                if ((palinfo >> 14) == 1)
+                {
+                    u16 color0 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset);
+                    u16 color1 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 2);
+
+                    u32 r0 = color0 & 0x001F;
+                    u32 g0 = color0 & 0x03E0;
+                    u32 b0 = color0 & 0x7C00;
+                    u32 r1 = color1 & 0x001F;
+                    u32 g1 = color1 & 0x03E0;
+                    u32 b1 = color1 & 0x7C00;
+
+                    u32 r = (r0 + r1) >> 1;
+                    u32 g = ((g0 + g1) >> 1) & 0x03E0;
+                    u32 b = ((b0 + b1) >> 1) & 0x7C00;
+
+                    color = r | g | b;
+                }
+                else if ((palinfo >> 14) == 3)
+                {
+                    u16 color0 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset);
+                    u16 color1 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 2);
+
+                    u32 r0 = color0 & 0x001F;
+                    u32 g0 = color0 & 0x03E0;
+                    u32 b0 = color0 & 0x7C00;
+                    u32 r1 = color1 & 0x001F;
+                    u32 g1 = color1 & 0x03E0;
+                    u32 b1 = color1 & 0x7C00;
+
+                    u32 r = (r0*5 + r1*3) >> 3;
+                    u32 g = ((g0*5 + g1*3) >> 3) & 0x03E0;
+                    u32 b = ((b0*5 + b1*3) >> 3) & 0x7C00;
+
+                    color = r | g | b;
+                }
+                else
+                    color = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 4);
+                break;
+
+            case 3:
+                if ((palinfo >> 14) == 2)
+                    color = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 6);
+                else if ((palinfo >> 14) == 3)
+                {
+                    u16 color0 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset);
+                    u16 color1 = GPU::ReadVRAM_TexPal<u16>(texpal + paloffset + 2);
+
+                    u32 r0 = color0 & 0x001F;
+                    u32 g0 = color0 & 0x03E0;
+                    u32 b0 = color0 & 0x7C00;
+                    u32 r1 = color1 & 0x001F;
+                    u32 g1 = color1 & 0x03E0;
+                    u32 b1 = color1 & 0x7C00;
+
+                    u32 r = (r0*3 + r1*5) >> 3;
+                    u32 g = ((g0*3 + g1*5) >> 3) & 0x03E0;
+                    u32 b = ((b0*3 + b1*5) >> 3) & 0x7C00;
+
+                    color = r | g | b;
+                }
+                else
+                    color = 0; // TODO transparent!
+                break;
+            }
+
+            *r = (color << 1) & 0x3E; if (*r) *r++;
+            *g = (color >> 4) & 0x3E; if (*g) *g++;
+            *b = (color >> 9) & 0x3E; if (*b) *b++;
+        }
+        break;
+
     default:
         *r = (s)&0x3F;
         *g = 0;
diff --git a/melonDS.depend b/melonDS.depend
index 67904b55..d1e6fb33 100644
--- a/melonDS.depend
+++ b/melonDS.depend
@@ -10,7 +10,7 @@
 
 1481161027 c:\documents\sources\melonds\types.h
 
-1488411597 source:c:\documents\sources\melonds\nds.cpp
+1488415700 source:c:\documents\sources\melonds\nds.cpp
 	<stdio.h>
 	<string.h>
 	"NDS.h"
@@ -87,7 +87,7 @@
 	"NDS.h"
 	"SPI.h"
 
-1488405741 source:c:\documents\sources\melonds\gpu2d.cpp
+1488415860 source:c:\documents\sources\melonds\gpu2d.cpp
 	<stdio.h>
 	<string.h>
 	"NDS.h"
@@ -155,7 +155,7 @@
 	"GPU.h"
 	"FIFO.h"
 
-1488412022 source:c:\documents\sources\melonds\gpu3d_soft.cpp
+1488415378 source:c:\documents\sources\melonds\gpu3d_soft.cpp
 	<stdio.h>
 	<string.h>
 	"NDS.h"