diff --git a/arm9/source/common/screenshot.c b/arm9/source/common/screenshot.c
index 0661cf1..93b0dc9 100644
--- a/arm9/source/common/screenshot.c
+++ b/arm9/source/common/screenshot.c
@@ -4,34 +4,24 @@
 #include "vff.h"
 #include "png.h"
 
-static void Screenshot_CvtAndTranspose(u8 *dest, u16 *fb, u32 w, u32 stride)
+static void Screenshot_Transpose(u16 *dest, const u16 *fb, u32 w, u32 stride)
 {
     for (u32 y = 0; y < SCREEN_HEIGHT; y++) {
-        for (u32 x = 0; x < w; x++) {
-            u8 r, g, b;
-            u16 rgb_s = GetColor(fb, x, y);
-
-            r = ((rgb_s >> 11) & 0x1F) << 3;
-            g = ((rgb_s >> 5) & 0x3F) << 2;
-            b = (rgb_s & 0x1F) << 3;
-
-            *(dest++) = r;
-            *(dest++) = g;
-            *(dest++) = b;
-        }
-
+        for (u32 x = 0; x < w; x++)
+            *(dest++) = GetColor(fb, x, y);
         dest += stride;
     }
 }
 
 void CreateScreenshot(void) {
+    u8 *png;
+    u16 *buffer;
     DsTime dstime;
     size_t png_size;
-    u8 *png, *buffer;
     char filename[64];
     u32 snapbuf_size, snap_w, snap_h, bot_offset;
 
-    snapbuf_size = (SCREEN_WIDTH_TOP * SCREEN_HEIGHT * 3) * 2;
+    snapbuf_size = (SCREEN_WIDTH_TOP * SCREEN_HEIGHT * BYTES_PER_PIXEL) * 2;
     snap_w = SCREEN_WIDTH_TOP;
     snap_h = SCREEN_HEIGHT * 2;
 
@@ -45,12 +35,13 @@ void CreateScreenshot(void) {
     buffer = malloc(snapbuf_size);
     if (!buffer) return;
 
-    memset(buffer, 0x1F, snapbuf_size); // gray background
+    for (unsigned i = snapbuf_size/4; i < snapbuf_size/2; i++)
+        buffer[i] = RGB(0x1F, 0x1F, 0x1F); // gray background
 
-    bot_offset = ((400 * SCREEN_HEIGHT) + 40) * 3;
+    bot_offset = (SCREEN_WIDTH_TOP * SCREEN_HEIGHT) + 40;
 
-    Screenshot_CvtAndTranspose(buffer, TOP_SCREEN, SCREEN_WIDTH_TOP, 0);
-    Screenshot_CvtAndTranspose(buffer + bot_offset, BOT_SCREEN, SCREEN_WIDTH_BOT, 80 * 3);
+    Screenshot_Transpose(buffer, TOP_SCREEN, SCREEN_WIDTH_TOP, 0);
+    Screenshot_Transpose(buffer + bot_offset, BOT_SCREEN, SCREEN_WIDTH_BOT, 80);
 
     png = PNG_Compress(buffer, snap_w, snap_h, &png_size);
 
diff --git a/arm9/source/common/ui.c b/arm9/source/common/ui.c
index 9896ace..e7a25e9 100644
--- a/arm9/source/common/ui.c
+++ b/arm9/source/common/ui.c
@@ -151,7 +151,7 @@ void ClearScreenF(bool clear_main, bool clear_alt, u32 color)
     if (clear_alt) ClearScreen(ALT_SCREEN, color);
 }
 
-u16 GetColor(u16 *screen, int x, int y)
+u16 GetColor(const u16 *screen, int x, int y)
 {
     return screen[PIXEL_OFFSET(x, y)];
 }
@@ -171,7 +171,7 @@ void DrawRectangle(u16 *screen, int x, int y, u32 width, u32 height, u32 color)
     }
 }
 
-void DrawBitmap(u16 *screen, int x, int y, u32 w, u32 h, const u8* bitmap)
+void DrawBitmap(u16 *screen, int x, int y, u32 w, u32 h, const u16* bitmap)
 {
     // on negative values: center the bitmap
     if (x < 0) x = (SCREEN_WIDTH(screen) - w) >> 1;
@@ -183,10 +183,8 @@ void DrawBitmap(u16 *screen, int x, int y, u32 w, u32 h, const u8* bitmap)
 
     screen += PIXEL_OFFSET(x, y);
     while(h--) {
-        for (u32 i = 0; i < w; i++) {
-            screen[i * SCREEN_HEIGHT] = RGB(bitmap[2], bitmap[1], bitmap[0]);
-            bitmap += 3;
-        }
+        for (u32 i = 0; i < w; i++)
+            screen[i * SCREEN_HEIGHT] = *(bitmap++);
         screen--;
     }
 }
@@ -392,7 +390,7 @@ void ShowString(const char *format, ...)
     } else ClearScreenF(true, false, COLOR_STD_BG);
 }
 
-void ShowIconString(u8* icon, int w, int h, const char *format, ...)
+void ShowIconString(u16* icon, int w, int h, const char *format, ...)
 {
     static const u32 icon_offset = 10;
     u32 str_width, str_height, tot_height;
diff --git a/arm9/source/common/ui.h b/arm9/source/common/ui.h
index dacfc86..15f5b83 100644
--- a/arm9/source/common/ui.h
+++ b/arm9/source/common/ui.h
@@ -48,13 +48,13 @@ bool ShowUnlockSequence(u32 seqlvl, const char *format, ...);
 u8* GetFontFromPbm(const void* pbm, const u32 pbm_size, u32* w, u32* h);
 bool SetFontFromPbm(const void* pbm, const u32 pbm_size);
 
-u16 GetColor(u16 *screen, int x, int y);
+u16 GetColor(const u16 *screen, int x, int y);
 
 void ClearScreen(u16 *screen, u32 color);
 void ClearScreenF(bool clear_main, bool clear_alt, u32 color);
 void DrawPixel(u16 *screen, int x, int y, u32 color);
 void DrawRectangle(u16 *screen, int x, int y, u32 width, u32 height, u32 color);
-void DrawBitmap(u16 *screen, int x, int y, u32 w, u32 h, const u8* bitmap);
+void DrawBitmap(u16 *screen, int x, int y, u32 w, u32 h, const u16* bitmap);
 void DrawQrCode(u16 *screen, const u8* qrcode);
 
 void DrawCharacter(u16 *screen, int character, int x, int y, u32 color, u32 bgcolor);
@@ -74,7 +74,7 @@ void FormatNumber(char* str, u64 number);
 void FormatBytes(char* str, u64 bytes);
 
 void ShowString(const char *format, ...);
-void ShowIconString(u8* icon, int w, int h, const char *format, ...);
+void ShowIconString(u16* icon, int w, int h, const char *format, ...);
 bool ShowPrompt(bool ask, const char *format, ...);
 u32 ShowSelectPrompt(u32 n, const char** options, const char *format, ...);
 u32 ShowFileScrollPrompt(u32 n, const DirEntry** entries, bool hide_ext, const char *format, ...);
diff --git a/arm9/source/game/nds.c b/arm9/source/game/nds.c
index ec1d966..b2f669a 100644
--- a/arm9/source/game/nds.c
+++ b/arm9/source/game/nds.c
@@ -52,7 +52,7 @@ u32 GetTwlTitle(char* desc, const TwlIconData* twl_icon) {
 }
 
 // TWL icon: 32x32 pixel, 8x8 tiles
-u32 GetTwlIcon(u8* icon, const TwlIconData* twl_icon) {
+u32 GetTwlIcon(u16* icon, const TwlIconData* twl_icon) {
     const u32 h = TWLICON_DIM_ICON; // fixed size
     const u32 w = TWLICON_DIM_ICON; // fixed size
     const u16* palette = twl_icon->palette;
@@ -60,13 +60,17 @@ u32 GetTwlIcon(u8* icon, const TwlIconData* twl_icon) {
     for (u32 y = 0; y < h; y += 8) {
         for (u32 x = 0; x < w; x += 8) {
             for (u32 i = 0; i < 8*8; i++) {
+                u16 pix555;
+                u8 r, g, b;
                 u32 ix = x + (i & 0x7);
                 u32 iy = y + (i >> 3);
-                u16 pix555 = palette[((i%2) ? (*pix4 >> 4) : *pix4) & 0xF];
-                u8* pix888 = icon + ((iy * w) + ix) * 3;
-                *(pix888++) = ((pix555 >> 10) & 0x1F) << 3; // B
-                *(pix888++) = ((pix555 >>  5) & 0x1F) << 3; // G
-                *(pix888++) = ((pix555 >>  0) & 0x1F) << 3; // R
+
+                pix555 = palette[((i%2) ? (*pix4 >> 4) : *pix4) & 0xF];
+                r = pix555 & 0x1F;
+                g = ((pix555 >> 5) & 0x1F) << 1;
+                g |= (g >> 1) & 1;
+                b = (pix555 >> 10) & 0x1F;
+                icon[(iy * w) + ix] = (r << 11) | (g << 5) | b;
                 if (i % 2) pix4++;
             }
         }
diff --git a/arm9/source/game/nds.h b/arm9/source/game/nds.h
index 69d1516..abb133f 100644
--- a/arm9/source/game/nds.h
+++ b/arm9/source/game/nds.h
@@ -13,7 +13,7 @@
                               (v == 0x0003) ? 0x1240 : (v == 0x0103) ? 0x23C0 : 0x0000)
 #define TWLICON_SIZE_DESC   128
 #define TWLICON_DIM_ICON    32
-#define TWLICON_SIZE_ICON   (TWLICON_DIM_ICON * TWLICON_DIM_ICON * 3) // w * h * bpp (rgb888)
+#define TWLICON_SIZE_ICON   (TWLICON_DIM_ICON * TWLICON_DIM_ICON * 2) // w * h * bpp (rgb565)
 #define NDS_LOGO_CRC16      0xCF56
 
 #define TWL_UNITCODE_NTR    0x00
@@ -126,7 +126,7 @@ typedef struct {
 u32 ValidateTwlHeader(TwlHeader* twl);
 u32 LoadTwlMetaData(const char* path, TwlHeader* hdr, TwlIconData* icon);
 u32 GetTwlTitle(char* desc, const TwlIconData* twl_icon);
-u32 GetTwlIcon(u8* icon, const TwlIconData* twl_icon);
+u32 GetTwlIcon(u16* icon, const TwlIconData* twl_icon);
 
 u32 FindNitroRomDir(u32 dirid, u32* fileid, u8** fnt_entry, TwlHeader* hdr, u8* fnt, u8* fat);
 u32 NextNitroRomEntry(u32* fileid, u8** fnt_entry);
diff --git a/arm9/source/game/smdh.c b/arm9/source/game/smdh.c
index ea860c5..8ad0acd 100644
--- a/arm9/source/game/smdh.c
+++ b/arm9/source/game/smdh.c
@@ -8,7 +8,7 @@
                      32, 33, 40, 41, 34, 35, 42, 43, 48, 49, 56, 57, 50, 51, 58, 59, \
                      36, 37, 44, 45, 38, 39, 46, 47, 52, 53, 60, 61, 54, 55, 62, 63
 
-u32 ConvertSmdhIcon(u8* icon, const u16* smdh_icon, u32 w, u32 h) {
+u32 ConvertSmdhIcon(u16* icon, const u16* smdh_icon, u32 w, u32 h) {
     const u32 lut[8*8] = { SMDH_LUT };
     u16* pix565 = (u16*) smdh_icon;
     for (u32 y = 0; y < h; y += 8) {
@@ -16,11 +16,7 @@ u32 ConvertSmdhIcon(u8* icon, const u16* smdh_icon, u32 w, u32 h) {
             for (u32 i = 0; i < 8*8; i++) {
                 u32 ix = x + (lut[i] & 0x7);
                 u32 iy = y + (lut[i] >> 3);
-                u8* pix888 = icon + ((iy * w) + ix) * 3;
-                *(pix888++) = ((*pix565 >>  0) & 0x1F) << 3; // B
-                *(pix888++) = ((*pix565 >>  5) & 0x3F) << 2; // G
-                *(pix888++) = ((*pix565 >> 11) & 0x1F) << 3; // R
-                pix565++;
+                icon[(iy * w) + ix] = *(pix565++);
             }
         }
     }
@@ -52,11 +48,11 @@ u32 GetSmdhPublisher(char* pub, const Smdh* smdh) {
 }
 
 // small icons are 24x24 => 0x6C0 byte in RGB888
-u32 GetSmdhIconSmall(u8* icon, const Smdh* smdh) {
+u32 GetSmdhIconSmall(u16* icon, const Smdh* smdh) {
     return ConvertSmdhIcon(icon, smdh->icon_small, SMDH_DIM_ICON_SMALL, SMDH_DIM_ICON_SMALL);
 }
 
 // big icons are 48x48 => 0x1B00 byte in RGB888
-u32 GetSmdhIconBig(u8* icon, const Smdh* smdh) {
+u32 GetSmdhIconBig(u16* icon, const Smdh* smdh) {
     return ConvertSmdhIcon(icon, smdh->icon_big, SMDH_DIM_ICON_BIG, SMDH_DIM_ICON_BIG);
 }
diff --git a/arm9/source/game/smdh.h b/arm9/source/game/smdh.h
index e1ce89e..1588218 100644
--- a/arm9/source/game/smdh.h
+++ b/arm9/source/game/smdh.h
@@ -9,8 +9,8 @@
 #define SMDH_SIZE_PUBLISHER     64
 #define SMDH_DIM_ICON_SMALL     24
 #define SMDH_DIM_ICON_BIG       48
-#define SMDH_SIZE_ICON_SMALL    (SMDH_DIM_ICON_SMALL * SMDH_DIM_ICON_SMALL * 3) // w * h * bpp (rgb888)
-#define SMDH_SIZE_ICON_BIG      (SMDH_DIM_ICON_BIG * SMDH_DIM_ICON_BIG * 3) // w * h * bpp (rgb888)
+#define SMDH_SIZE_ICON_SMALL    (SMDH_DIM_ICON_SMALL * SMDH_DIM_ICON_SMALL * 2) // w * h * bpp (rgb565)
+#define SMDH_SIZE_ICON_BIG      (SMDH_DIM_ICON_BIG * SMDH_DIM_ICON_BIG * 2) // w * h * bpp (rgb565)
 
 // see: https://www.3dbrew.org/wiki/SMDH#Application_Titles
 typedef struct {
@@ -42,5 +42,5 @@ typedef struct {
 u32 GetSmdhDescShort(char* desc, const Smdh* smdh);
 u32 GetSmdhDescLong(char* desc, const Smdh* smdh);
 u32 GetSmdhPublisher(char* pub, const Smdh* smdh);
-u32 GetSmdhIconSmall(u8* icon, const Smdh* smdh);
-u32 GetSmdhIconBig(u8* icon, const Smdh* smdh);
+u32 GetSmdhIconSmall(u16* icon, const Smdh* smdh);
+u32 GetSmdhIconBig(u16* icon, const Smdh* smdh);
diff --git a/arm9/source/godmode.c b/arm9/source/godmode.c
index 786c1a8..5ca3dbf 100644
--- a/arm9/source/godmode.c
+++ b/arm9/source/godmode.c
@@ -115,10 +115,10 @@ u32 BootFirmHandler(const char* bootpath, bool verbose, bool delete) {
 }
 
 u32 SplashInit(const char* modestr) {
+    u16* bitmap;
     u64 splash_size;
     u32 splash_width, splash_height;
     u8* splash = FindVTarFileInfo(VRAM0_SPLASH_PNG, &splash_size);
-    u8* bitmap = NULL;
     const char* namestr = FLAVOR " " VERSION;
     const char* loadstr = "booting...";
     const u32 pos_xb = 10;
@@ -131,7 +131,10 @@ u32 SplashInit(const char* modestr) {
     if (splash) {
         bitmap = PNG_Decompress(splash, splash_size, &splash_width, &splash_height);
         if (bitmap) DrawBitmap(TOP_SCREEN, -1, -1, splash_width, splash_height, bitmap);
-    } else DrawStringF(TOP_SCREEN, 10, 10, COLOR_STD_FONT, COLOR_TRANSPARENT, "(" VRAM0_SPLASH_PNG " not found)");
+    } else {
+        DrawStringF(TOP_SCREEN, 10, 10, COLOR_STD_FONT, COLOR_TRANSPARENT, "(" VRAM0_SPLASH_PNG " not found)");
+        bitmap = NULL;
+    }
 
     if (modestr) DrawStringF(TOP_SCREEN, SCREEN_WIDTH_TOP - 10 - GetDrawStringWidth(modestr),
         SCREEN_HEIGHT - 10 - GetDrawStringHeight(modestr), COLOR_STD_FONT, COLOR_TRANSPARENT, modestr);
@@ -144,7 +147,7 @@ u32 SplashInit(const char* modestr) {
     DrawStringF(BOT_SCREEN, pos_xu, pos_yu, COLOR_STD_FONT, COLOR_STD_BG, loadstr);
     DrawStringF(BOT_SCREEN, pos_xb, pos_yu, COLOR_STD_FONT, COLOR_STD_BG, "built: " DBUILTL);
 
-    if (bitmap) free(bitmap);
+    free(bitmap);
     return 0;
 }
 
@@ -494,7 +497,7 @@ u32 SdFormatMenu(const char* slabel) {
 u32 FileGraphicsViewer(const char* path) {
     const u32 max_size = SCREEN_SIZE(ALT_SCREEN);
     u64 filetype = IdentifyFileType(path);
-    u8* bitmap = NULL;
+    u16* bitmap = NULL;
     u8* input = (u8*)malloc(max_size);
     u32 w = 0;
     u32 h = 0;
@@ -519,8 +522,8 @@ u32 FileGraphicsViewer(const char* path) {
         ClearScreenF(true, true, COLOR_STD_BG);
     } else ret = 1;
 
-    if (bitmap) free(bitmap);
-    if (input) free(input);
+    free(bitmap);
+    free(input);
     return ret;
 }
 
diff --git a/arm9/source/system/png.c b/arm9/source/system/png.c
index c606e3c..da4b4e9 100644
--- a/arm9/source/system/png.c
+++ b/arm9/source/system/png.c
@@ -5,43 +5,68 @@
 #include "png.h"
 
 #ifndef MONITOR_HEAP
-static inline void _rgb_swap(u8 *img, size_t sz)
+// dest and src can overlap
+static inline void _rgb24_to_rgb565(u16 *dest, const u8 *src, size_t dim)
 {
-	// maybe process in batches of 3 pixels / 12 bytes at a time?
-	for (size_t i = 0; i < sz; i+=3) {
-		u8 c = img[i];
-		img[i] = img[i + 2];
-		img[i + 2] = c;
+	for (size_t i = 0; i < dim; i++) {
+		u8 r, g, b;
+
+		r = *(src++) >> 3;
+		g = *(src++) >> 2;
+		b = *(src++) >> 3;
+		*(dest++) = r << 11 | g << 5 | b;
 	}
 }
 
-u8 *PNG_Decompress(const u8 *png, size_t png_len, u32 *w, u32 *h)
+// dest and src CAN NOT overlap
+static inline void _rgb565_to_rgb24(u8 *dest, const u16 *src, size_t dim)
 {
-	u8 *img;
-	u32 res;
-	size_t w_, h_;
+	for (size_t i = 0; i < dim; i++) {
+		u16 rgb = *(src++);
+
+		*(dest++) = (rgb >> 11) << 3;
+		*(dest++) = ((rgb >> 5) & 0x3F) << 2;
+		*(dest++) = (rgb & 0x1F) << 3;
+	}
+}
+
+u16 *PNG_Decompress(const u8 *png, size_t png_len, u32 *w, u32 *h)
+{
+	u16 *img;
+	unsigned res;
+	size_t width, height;
 
 	img = NULL;
-	res = lodepng_decode24(&img, &w_, &h_, png, png_len);
+	res = lodepng_decode24((u8**)&img, &width, &height, png, png_len);
 	if (res) {
 		free(img);
 		return NULL;
 	}
-	_rgb_swap(img, w_ * h_ * 3);
-	if (w) *w = w_;
-	if (h) *h = h_;
 
-	return img;
+	_rgb24_to_rgb565(img, (const u8*)img, width * height);
+	if (w) *w = width;
+	if (h) *h = height;
+
+	// the allocated buffer will be w*h*3 bytes long, but only w*h*2 bytes will be used
+	// however, this is not a problem and it'll all be freed with a regular free() call
+	return (u16*)img;
 }
 
-u8 *PNG_Compress(const u8 *fb, u32 w, u32 h, size_t *png_sz)
+u8 *PNG_Compress(const u16 *fb, u32 w, u32 h, size_t *png_sz)
 {
-	u8 *img;
-	u32 res;
+	u8 *img, *buf;
+	unsigned res;
 	size_t png_size;
 
 	img = NULL;
-	res = lodepng_encode24(&img, &png_size, fb, w, h);
+
+	buf = malloc(w * h * 3);
+	if (!buf) return NULL;
+
+	_rgb565_to_rgb24(buf, fb, w * h);
+	res = lodepng_encode24(&img, &png_size, buf, w, h);
+	free(buf);
+
 	if (res) {
 		free(img);
 		return NULL;
diff --git a/arm9/source/system/png.h b/arm9/source/system/png.h
index 06a02a5..a10562c 100644
--- a/arm9/source/system/png.h
+++ b/arm9/source/system/png.h
@@ -6,5 +6,5 @@
 
 #define PNG_MAGIC   0x89, 0x50, 0x4E, 0x47, 0x0D, 0x0A, 0x1A, 0x0A
 
-u8 *PNG_Decompress(const u8 *png, size_t png_len, u32 *w, u32 *h);
-u8 *PNG_Compress(const u8 *fb, u32 w, u32 h, size_t *png_sz);
+u16 *PNG_Decompress(const u8 *png, size_t png_len, u32 *w, u32 *h);
+u8 *PNG_Compress(const u16 *fb, u32 w, u32 h, size_t *png_sz);
diff --git a/arm9/source/utils/gameutil.c b/arm9/source/utils/gameutil.c
index 8aff4ec..512ebde 100644
--- a/arm9/source/utils/gameutil.c
+++ b/arm9/source/utils/gameutil.c
@@ -1971,7 +1971,7 @@ u32 LoadSmdhFromGameFile(const char* path, Smdh* smdh) {
 u32 ShowSmdhTitleInfo(Smdh* smdh) {
     const u8 smdh_magic[] = { SMDH_MAGIC };
     const u32 lwrap = 24;
-    u8 icon[SMDH_SIZE_ICON_BIG];
+    u16 icon[SMDH_SIZE_ICON_BIG];
     char desc_l[SMDH_SIZE_DESC_LONG+1];
     char desc_s[SMDH_SIZE_DESC_SHORT+1];
     char pub[SMDH_SIZE_PUBLISHER+1];
@@ -1992,7 +1992,7 @@ u32 ShowSmdhTitleInfo(Smdh* smdh) {
 
 u32 ShowTwlIconTitleInfo(TwlIconData* twl_icon) {
     const u32 lwrap = 24;
-    u8 icon[TWLICON_SIZE_ICON];
+    u16 icon[TWLICON_SIZE_ICON];
     char desc[TWLICON_SIZE_DESC+1];
     if ((GetTwlIcon(icon, twl_icon) != 0) ||
         (GetTwlTitle(desc, twl_icon) != 0))
diff --git a/arm9/source/utils/scripting.c b/arm9/source/utils/scripting.c
index e71edc0..bdda90e 100644
--- a/arm9/source/utils/scripting.c
+++ b/arm9/source/utils/scripting.c
@@ -1840,7 +1840,7 @@ bool ExecuteGM9Script(const char* path_script) {
                 if (preview_mode > 2) {
                     char* preview_str = get_var("PREVIEW_MODE", NULL);
                     u32 bitmap_width, bitmap_height;
-                    u8* bitmap = NULL;
+                    u16* bitmap = NULL;
 
                     u8* png = (u8*) malloc(SCREEN_SIZE_TOP);
                     if (png) {