From 96078b99ac738b1bedf2832a13c7909f614d7efb Mon Sep 17 00:00:00 2001
From: zeromus <zeromus@users.sf.net>
Date: Sat, 12 Sep 2015 03:44:08 +0000
Subject: [PATCH] support 512 color palettes, add support for proper deemph
 emulation on some codepaths. need to fix png saving and maybe other stuff

---
 trunk/src/driver.h                   |   6 +-
 trunk/src/drivers/common/vidblit.cpp |  55 +++++--
 trunk/src/drivers/win/common.h       |   2 +-
 trunk/src/drivers/win/config.cpp     |   4 +-
 trunk/src/drivers/win/main.cpp       |   9 +-
 trunk/src/drivers/win/main.h         |   5 +-
 trunk/src/drivers/win/palette.cpp    |  30 ++--
 trunk/src/drivers/win/video.cpp      |  26 +---
 trunk/src/fceu.cpp                   |   2 +-
 trunk/src/fceu.h                     |   4 +-
 trunk/src/palette.cpp                | 207 +++++++++++++++++----------
 trunk/src/palettes/palettes.h        |  25 +++-
 trunk/src/ppu.cpp                    |  23 ++-
 trunk/src/video.cpp                  |  63 ++++----
 trunk/src/video.h                    |   2 +
 trunk/src/vsuni.cpp                  |   2 +-
 16 files changed, 291 insertions(+), 174 deletions(-)

diff --git a/trunk/src/driver.h b/trunk/src/driver.h
index db120671..85ca097a 100644
--- a/trunk/src/driver.h
+++ b/trunk/src/driver.h
@@ -68,7 +68,7 @@ void FCEUI_NTSCSELTINT(void);
 void FCEUI_NTSCDEC(void);
 void FCEUI_NTSCINC(void);
 void FCEUI_GetNTSCTH(int *tint, int *hue);
-void FCEUI_SetNTSCTH(int n, int tint, int hue);
+void FCEUI_SetNTSCTH(bool en, int tint, int hue);
 
 void FCEUI_SetInput(int port, ESI type, void *ptr, int attrib);
 void FCEUI_SetInputFC(ESIFC type, void *ptr, int attrib);
@@ -141,9 +141,7 @@ void FCEUI_SetRenderedLines(int ntscf, int ntscl, int palf, int pall);
 //Sets the base directory(save states, snapshots, etc. are saved in directories below this directory.
 void FCEUI_SetBaseDirectory(std::string const & dir);
 
-//Tells FCE Ultra to copy the palette data pointed to by pal and use it.
-//Data pointed to by pal needs to be 64*3 bytes in length.
-void FCEUI_SetPaletteArray(uint8 *pal);
+void FCEUI_SetUserPalette(uint8 *pal, int nEntries);
 
 //Sets up sound code to render sound at the specified rate, in samples
 //per second.  Only sample rates of 44100, 48000, and 96000 are currently supported.
diff --git a/trunk/src/drivers/common/vidblit.cpp b/trunk/src/drivers/common/vidblit.cpp
index 73b6427c..037dace2 100644
--- a/trunk/src/drivers/common/vidblit.cpp
+++ b/trunk/src/drivers/common/vidblit.cpp
@@ -25,9 +25,16 @@
 #include "hq3x.h"
 
 #include "../../types.h"
+#include "../../palette.h"
 #include "../../utils/memory.h"
 #include "nes_ntsc.h"
 
+extern u8 *XBuf;
+extern u8 *XBackBuf;
+extern u8 *XDBuf;
+extern u8 *XDBackBuf;
+extern pal *palo;
+
 nes_ntsc_t* nes_ntsc;
 uint8 burst_phase = 0;
 
@@ -203,17 +210,17 @@ int InitBlitToHigh(int b, uint32 rmask, uint32 gmask, uint32 bmask, int efx, int
 	
 	if(efx&FVB_BLUR)
 	{
-		if(Bpp==2)
-			palettetranslate=(uint32 *)FCEU_dmalloc(65536*4);
-		else if(Bpp>=3)
+		//if(Bpp==2)
+		//	palettetranslate=(uint32 *)FCEU_dmalloc(65536*4);
+		//else if(Bpp>=3)
 			palettetranslate=(uint32 *)FCEU_dmalloc(65536*4);
 	}
 	else
 	{
-		if(Bpp==2)
-			palettetranslate=(uint32*)FCEU_dmalloc(65536*4);
-		else if(Bpp>=3)
-			palettetranslate=(uint32*)FCEU_dmalloc(256*4);
+		//if(Bpp==2)
+		//	palettetranslate=(uint32*)FCEU_dmalloc(65536*4);
+		//else if(Bpp>=3)
+			palettetranslate=(uint32*)FCEU_dmalloc(256*4 + 512*4);
 	}
 	
 	if(!palettetranslate)
@@ -340,6 +347,7 @@ void SetPaletteBlitToHigh(uint8 *src)
 				palettetranslate[x]=(r<<cshiftl[0])|(g<<cshiftl[1])|(b<<cshiftl[2]);
 			}
 			else	
+			{
 				for(y=0;y<256;y++)
 				{
 					r=src[x<<2]*(100-BLUR_RED);
@@ -359,6 +367,17 @@ void SetPaletteBlitToHigh(uint8 *src)
 					
 					palettetranslate[x|(y<<8)]=(r<<cshiftl[0])|(g<<cshiftl[1])|(b<<cshiftl[2]);
 				}
+			}
+		}
+		//full size deemph palette (breaks blur shit)
+		for(x=0;x<512;x++)
+		{
+			uint32 r,g,b;
+
+			r=palo[x].r;
+			g=palo[x].g;
+			b=palo[x].b;
+			palettetranslate[256+x]=(r<<cshiftl[0])|(g<<cshiftl[1])|(b<<cshiftl[2]);
 		}
 		break;
 	}
@@ -507,6 +526,25 @@ void Blit8To8(uint8 *src, uint8 *dest, int xr, int yr, int pitch, int xscale, in
 
 /* Todo:  Make sure 24bpp code works right with big-endian cpus */
 
+//takes a pointer to XBuf and applies fully modern deemph palettizing
+u32 ModernDeemphColorMap(u8* src)
+{
+	u8 pixel = *src;
+	
+	//look up the legacy translation
+	u32 color = palettetranslate[pixel];
+	
+	//find out which deemph bitplane value we're on
+	int ofs = src-XBuf;
+	uint8 deemph = XDBuf[ofs];
+
+	//if it was a deemph'd value, grab it from the deemph palette
+	if(deemph != 0)
+		color = palettetranslate[256+(pixel&0x3F)+deemph*64];
+
+	return color;
+}
+
 void Blit8ToHigh(uint8 *src, uint8 *dest, int xr, int yr, int pitch, int xscale, int yscale)
 {
 	int x,y;
@@ -1036,7 +1074,8 @@ void Blit8ToHigh(uint8 *src, uint8 *dest, int xr, int yr, int pitch, int xscale,
 				{
 					for(x=xr;x;x--)
 					{
-						*(uint32 *)dest=palettetranslate[(uint32)*src];
+						//THE MAIN BLITTING CODEPATH (there may be others that are important)
+						*(uint32 *)dest = ModernDeemphColorMap(src);
 						dest+=4;
 						src++;
 					}
diff --git a/trunk/src/drivers/win/common.h b/trunk/src/drivers/win/common.h
index 191c117f..6ac1fd8e 100644
--- a/trunk/src/drivers/win/common.h
+++ b/trunk/src/drivers/win/common.h
@@ -35,7 +35,7 @@ extern int eoptions;
 
 #define EO_BGRUN              1
 
-#define EO_CPALETTE           4
+#define EO_CPALETTE           4 //this is bad, it is desynchronized from other state management in palette.cpp
 #define EO_NOSPRLIM           8
 #define EO_FSAFTERLOAD       32
 #define EO_FOAFTERSTART      64
diff --git a/trunk/src/drivers/win/config.cpp b/trunk/src/drivers/win/config.cpp
index fc371848..87cbd7d4 100644
--- a/trunk/src/drivers/win/config.cpp
+++ b/trunk/src/drivers/win/config.cpp
@@ -174,7 +174,7 @@ static CFGSTRUCT fceuconfig[] =
 	ACS(ResumeROM),
 
 	AC(gNoBGFillColor),
-	AC(ntsccol),AC(ntsctint),AC(ntschue),
+	AC(ntsccol_enable),AC(ntsctint),AC(ntschue),
 	AC(force_grayscale),
 	AC(dendy),
 	AC(extrascanlines),
@@ -470,7 +470,7 @@ void LoadConfig(const char *filename)
 
 	LoadFCEUConfig(filename, fceuconfig);
 
-	FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+	FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 
 	//adelikat:Hacky fix for Ram Watch recent menu
 	for (int x = 0; x < 5; x++)
diff --git a/trunk/src/drivers/win/main.cpp b/trunk/src/drivers/win/main.cpp
index aa19fd8a..5a5286b7 100644
--- a/trunk/src/drivers/win/main.cpp
+++ b/trunk/src/drivers/win/main.cpp
@@ -683,10 +683,11 @@ int main(int argc,char *argv[])
 	DoVideoConfigFix();
 	DoTimingConfigFix();
 
-	if(eoptions & EO_CPALETTE)
-	{
-		FCEUI_SetPaletteArray(cpalette);
-	}
+	//TODO - how can this do anything? we havent set a palette yet
+	//if(eoptions & EO_CPALETTE)
+	//{
+	//	FCEUI_SetUserPalette(cpalette);
+	//}
 
 	if(!t)
 	{
diff --git a/trunk/src/drivers/win/main.h b/trunk/src/drivers/win/main.h
index 51e932f5..c8c40e9e 100644
--- a/trunk/src/drivers/win/main.h
+++ b/trunk/src/drivers/win/main.h
@@ -107,13 +107,14 @@ extern int soundquality;
 extern bool muteTurbo;
 extern bool swapDuty;
 
-extern uint8 cpalette[192];
+extern uint8 cpalette[64*8*3];
 extern int srendlinen;
 extern int erendlinen;
 extern int srendlinep;
 extern int erendlinep;
 
-extern int ntsccol, ntsctint, ntschue;
+extern int ntsctint, ntschue;
+extern bool ntsccol_enable;
 extern bool force_grayscale;
 
 //mbg merge 7/17/06 did these have to be unsigned?
diff --git a/trunk/src/drivers/win/palette.cpp b/trunk/src/drivers/win/palette.cpp
index d13d3f9a..726be71d 100644
--- a/trunk/src/drivers/win/palette.cpp
+++ b/trunk/src/drivers/win/palette.cpp
@@ -4,7 +4,7 @@
 #include "window.h"
 #include "gui.h"
 
-uint8 cpalette[192] = {0};
+u8 cpalette[64*8*3] = {0};
 extern int  palhue;
 extern bool palhdtv;
 extern bool palprecision;
@@ -15,9 +15,9 @@ bool SetPalette(const char* nameo)
 	FILE *fp;
 	if((fp = FCEUD_UTF8fopen(nameo, "rb")))
 	{
-		fread(cpalette, 1, 192, fp);
+		int readed = fread(cpalette, 1, sizeof(cpalette), fp);
 		fclose(fp);
-		FCEUI_SetPaletteArray(cpalette);
+		FCEUI_SetUserPalette(cpalette,readed/3);
 		eoptions |= EO_CPALETTE;
 		return true;
 	}
@@ -71,7 +71,7 @@ BOOL CALLBACK PaletteConCallB(HWND hwndDlg, UINT uMsg, WPARAM wParam, LPARAM lPa
 	{
 		case WM_INITDIALOG:
 
-			if(ntsccol)
+			if(ntsccol_enable)
 				CheckDlgButton(hwndDlg, CHECK_PALETTE_ENABLED, BST_CHECKED);
 
 			SendDlgItemMessage(hwndDlg, CTL_TINT_TRACKBAR,   TBM_SETRANGE, 1, MAKELONG(0, 128));
@@ -106,7 +106,7 @@ BOOL CALLBACK PaletteConCallB(HWND hwndDlg, UINT uMsg, WPARAM wParam, LPARAM lPa
 			ntsctint = SendDlgItemMessage(hwndDlg, CTL_TINT_TRACKBAR,   TBM_GETPOS, 0, (LPARAM)(LPSTR)0);
 			ntschue  = SendDlgItemMessage(hwndDlg, CTL_HUE_TRACKBAR,    TBM_GETPOS, 0, (LPARAM)(LPSTR)0);
 			palhue   = SendDlgItemMessage(hwndDlg, CTL_PALHUE_TRACKBAR, TBM_GETPOS, 0, (LPARAM)(LPSTR)0);
-			FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+			FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 			sprintf(text, "Hue: %d%%", palhue);
 			SendDlgItemMessage(hwndDlg, STATIC_HUEVALUE, WM_SETTEXT, 0, (LPARAM) text);
 			break;
@@ -121,13 +121,13 @@ BOOL CALLBACK PaletteConCallB(HWND hwndDlg, UINT uMsg, WPARAM wParam, LPARAM lPa
 				switch(wParam&0xFFFF)
 				{
 					case CHECK_PALETTE_ENABLED:
-						ntsccol ^= 1;
-						FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue); // it recalculates everything, use it for PAL block too!
+						ntsccol_enable ^= true;
+						FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue); // it recalculates everything, use it for PAL block too!
 						break;
 
 					case CHECK_PALETTE_GRAYSCALE:
 						force_grayscale ^= 1;
-						FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+						FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 						break;
 
 					case BTN_PALETTE_LOAD:
@@ -140,30 +140,30 @@ BOOL CALLBACK PaletteConCallB(HWND hwndDlg, UINT uMsg, WPARAM wParam, LPARAM lPa
 						SendDlgItemMessage(hwndDlg, CTL_PALHUE_TRACKBAR, TBM_SETPOS, 1, palhue);
 						sprintf(text, "Hue: %d%%", palhue);
 						SendDlgItemMessage(hwndDlg, STATIC_HUEVALUE, WM_SETTEXT, 0, (LPARAM) text);
-						FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+						FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 						break;
 
 					case CHECK_PALETTE_MONOCHROME:
 						palmonochrome ^= 1;
-						FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+						FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 						break;		
 					
 					case CHECK_PALETTE_HDTV:
 						palhdtv ^= 1;
-						FCEUI_SetNTSCTH(ntsccol, ntsctint, ntschue);
+						FCEUI_SetNTSCTH(ntsccol_enable, ntsctint, ntschue);
 						break;
 
 					case CHECK_PALETTE_CUSTOM:
 					{
 						if (eoptions & EO_CPALETTE)
 						{
-							// switch back to default palette
-							FCEUI_SetPaletteArray(0);
+							//disable user palette
+							FCEUI_SetUserPalette(0,0);
 							eoptions &= ~EO_CPALETTE;
 						} else
 						{
-							// switch to custom, even if it isn't loaded yet
-							FCEUI_SetPaletteArray(cpalette);
+							//switch to user palette (even if it isn't loaded yet!?)
+							FCEUI_SetUserPalette(cpalette,64); //just have to guess the size I guess
 							eoptions |= EO_CPALETTE;
 						}
 						break;
diff --git a/trunk/src/drivers/win/video.cpp b/trunk/src/drivers/win/video.cpp
index 796755fe..9f9a5ce8 100644
--- a/trunk/src/drivers/win/video.cpp
+++ b/trunk/src/drivers/win/video.cpp
@@ -227,27 +227,8 @@ static int InitBPPStuff(int fs)
 		break;
 	}
 
-	if(bpp >= 16)
-	{
-		InitBlitToHigh(bpp >> 3, CBM[0], CBM[1], CBM[2], 0, fs?vmodes[vmod].special:winspecial,specfilteropt);
-	}
-	else if(bpp==8)
-	{
-		ddrval=IDirectDraw7_CreatePalette( lpDD7, DDPCAPS_8BIT|DDPCAPS_ALLOW256|DDPCAPS_INITIALIZE,color_palette,&lpddpal,NULL);
-		if (ddrval != DD_OK)
-		{
-			//ShowDDErr("Error creating palette object.");
-			FCEU_printf("Error creating palette object.\n");
-			return 0;
-		}
-		ddrval=IDirectDrawSurface7_SetPalette(lpDDSPrimary, lpddpal);
-		if (ddrval != DD_OK)
-		{
-			//ShowDDErr("Error setting palette object.");
-			FCEU_printf("Error setting palette object.\n");
-			return 0;
-		}
-	}
+	InitBlitToHigh(bpp >> 3, CBM[0], CBM[1], CBM[2], 0, fs?vmodes[vmod].special:winspecial,specfilteropt);
+	
 	return 1;
 }
 
@@ -394,7 +375,8 @@ int SetVideoMode(int fs)
 		if(!GetBPP())
 			return 0;
 
-		if(bpp!=16 && bpp!=24 && bpp!=32)
+		//only 24 and 32bpp are supported now
+		if(bpp!=24 && bpp!=32)
 		{
 			//ShowDDErr("Current bit depth not supported!");
 			FCEU_printf("Current bit depth not supported!\n");
diff --git a/trunk/src/fceu.cpp b/trunk/src/fceu.cpp
index 17ba346d..51ba848c 100644
--- a/trunk/src/fceu.cpp
+++ b/trunk/src/fceu.cpp
@@ -395,7 +395,7 @@ void ResetGameLoaded(void) {
 	MMC5Hack = 0;
 	PEC586Hack = 0;
 	PAL &= 1;
-	pale = 0;
+	default_palette_selection = 0;
 }
 
 int UNIFLoad(const char *name, FCEUFILE *fp);
diff --git a/trunk/src/fceu.h b/trunk/src/fceu.h
index b8046a96..d1aa887b 100644
--- a/trunk/src/fceu.h
+++ b/trunk/src/fceu.h
@@ -129,7 +129,7 @@ void FCEU_DispMessage(char *format, int disppos, ...);
 void FCEU_DispMessageOnMovie(char *format, ...);
 void FCEU_TogglePPU();
 
-void SetNESDeemph(uint8 d, int force);
+void SetNESDeemph_OldHacky(uint8 d, int force);
 void DrawTextTrans(uint8 *dest, uint32 width, uint8 *textmsg, uint8 fgcolor);
 void FCEU_PutImage(void);
 #ifdef FRAMESKIP
@@ -142,7 +142,7 @@ extern void PushCurrentVideoSettings();
 #endif
 
 extern uint8 Exit;
-extern uint8 pale;
+extern int default_palette_selection;
 extern uint8 vsdip;
 
 //#define FCEUDEF_DEBUGGER //mbg merge 7/17/06 - cleaning out conditional compiles
diff --git a/trunk/src/palette.cpp b/trunk/src/palette.cpp
index eba1a7b2..29fc80ca 100644
--- a/trunk/src/palette.cpp
+++ b/trunk/src/palette.cpp
@@ -31,8 +31,6 @@
 #include "palette.h"
 #include "palettes/palettes.h"
 
-
-
 #ifndef M_PI
 #define M_PI 3.14159265358979323846
 #endif
@@ -42,24 +40,25 @@
 #include <cmath>
 #include <cstring>
 
-static int ntsccol=0;
-static int ntsctint=46+10;
-static int ntschue=72;
-
 bool force_grayscale = false;
 
-/* These are dynamically filled/generated palettes: */
-pal palettei[64];       // Custom palette for an individual game.
-pal palettec[64];       // Custom "global" palette.
-pal paletten[64];       // Mathematically generated palette.
+pal palette_game[64*8]; //custom palette for an individual game. (formerly palettei)
+pal palette_user[64*8]; //user's overridden palette (formerly palettec)
+pal palette_ntsc[64*8]; //mathematically generated NTSC palette (formerly paletten)
 
-static void CalculatePalette(void);
-static void ChoosePalette(void);
-static void WritePalette(void);
-uint8 pale=0;
+static bool palette_game_available; //whether palette_game is available
+static bool palette_user_available; //whether palette_user is available
 
-pal *palo;
-static pal *palpoint[8]=
+//ntsc parameters:
+bool ntsccol_enable = false; //whether NTSC palette is selected
+static int ntsctint = 46+10;
+static int ntschue = 72;
+
+//the default basic palette
+int default_palette_selection = 0;
+
+//library of default palettes
+static pal *default_palette[8]=
 {
 	palette,
 	rp2c04001,
@@ -68,35 +67,99 @@ static pal *palpoint[8]=
 	rp2c05004,
 };
 
-void FCEUI_SetPaletteArray(uint8 *pal)
+static void CalculatePalette(void);
+static void ChoosePalette(void);
+static void WritePalette(void);
+
+//points to the actually selected current palette
+pal *palo;
+
+static float rtmul[] = { 1.239f, 0.794f, 1.019f, 0.905f, 1.023f, 0.741f, 0.75f };
+static float gtmul[] = { 0.915f, 1.086f, 0.98f, 1.026f, 0.908f, 0.987f, 0.75f };
+static float btmul[] = { 0.743f, 0.882f, 0.653f, 1.277f, 0.979f, 0.101f, 0.75f };
+
+static void ApplyDeemphasis(u8& r, u8& g, u8& b, int deemph_bits)
+{
+	//DEEMPH BITS MAY BE ORDERED WRONG. PLEASE CHECK
+	if (deemph_bits == 0) return;
+	int d = deemph_bits - 1;
+	int nr = (int)(r * rtmul[d]);
+	int ng = (int)(g * gtmul[d]);
+	int nb = (int)(b * btmul[d]);
+	if (nr > 0xFF) nr = 0xFF;
+	if (ng > 0xFF) ng = 0xFF;
+	if (nb > 0xFF) nb = 0xFF;
+	r = (u8)nr;
+	g = (u8)ng;
+	b = (u8)nb;
+}
+
+static void ApplyDeemphasisComplete(pal* pal512)
 {
-	if(!pal)
-		palpoint[0]=palette;
-	else
-	{
-		int x;
-		palpoint[0]=palettec;
-		for(x=0;x<64;x++)
-		{
-			palpoint[0][x].r=*((uint8 *)pal+x+x+x);
-			palpoint[0][x].g=*((uint8 *)pal+x+x+x+1);
-			palpoint[0][x].b=*((uint8 *)pal+x+x+x+2);
+	//for each deemph level beyond 0
+	for(int i=1,idx=64;i<8;i++)
+	{
+		//for each palette entry
+		for(int p=0;p<64;p++,idx++)
+		{
+			pal512[idx] = pal512[p];
+			ApplyDeemphasis(pal512[idx].r,pal512[idx].g,pal512[idx].b,i);
 		}
 	}
+}
+
+void FCEUI_SetUserPalette(uint8 *pal, int nEntries)
+{
+	if(!pal)
+	{
+		palette_user_available = false;
+		return;
+	}
+
+	palette_user_available = true;
+	memcpy(palette_user,pal,nEntries*3);
+
+	//if palette is incomplete, generate deemph entries
+	if(nEntries != 512)
+		ApplyDeemphasisComplete(palette_user);
+
 	FCEU_ResetPalette();
 }
 
+void FCEU_LoadGamePalette(void)
+{
+	palette_game_available = false;
+	std::string path = FCEU_MakeFName(FCEUMKF_PALETTE,0,0);
+	FILE* fp = FCEUD_UTF8fopen(path,"rb");
+	if(fp)
+	{
+		int readed = fread(palette_game,1,64*8*3,fp);
+		int nEntries = readed/3;
+		fclose(fp);
 
-void FCEUI_SetNTSCTH(int n, int tint, int hue)
+		//if palette is incomplete, generate deemph entries
+		if(nEntries != 512)
+			ApplyDeemphasisComplete(palette_game);
+
+		palette_game_available = true;
+	}
+
+	//not sure whether this is needed
+	FCEU_ResetPalette();
+}
+
+void FCEUI_SetNTSCTH(bool en, int tint, int hue)
 {
 	ntsctint=tint;
 	ntschue=hue;
-	ntsccol=n;
+	ntsccol_enable = en;
 	FCEU_ResetPalette();
 }
 
+//this prepares the 'deemph' palette which was a horrible idea to jam a single deemph palette into 0xC0-0xFF of the 8bpp palette.
+//its needed for GUI and lua and stuff, so we're leaving it, despite having a newer codepath for applying deemph
 static uint8 lastd=0;
-void SetNESDeemph(uint8 d, int force)
+void SetNESDeemph_OldHacky(uint8 d, int force)
 {
 	static uint16 rtmul[]={
         static_cast<uint16>(32768*1.239),
@@ -194,6 +257,10 @@ void SetNESDeemph(uint8 d, int force)
 // Converted from Kevin Horton's qbasic palette generator.
 static void CalculatePalette(void)
 {
+	//PRECONDITION: ntsc palette is enabled
+ 	if(!ntsccol_enable)
+		return;
+
 	int x,z;
 	int r,g,b;
 	double s,luma,theta;
@@ -229,39 +296,14 @@ static void CalculatePalette(void)
 			if(g<0) g=0;
 			if(b<0) b=0;
 
-			paletten[(x<<4)+z].r=r;
-			paletten[(x<<4)+z].g=g;
-			paletten[(x<<4)+z].b=b;
+			palette_ntsc[(x<<4)+z].r=r;
+			palette_ntsc[(x<<4)+z].g=g;
+			palette_ntsc[(x<<4)+z].b=b;
 		}
-		WritePalette();
-}
 
-static int ipalette=0;
-
-void FCEU_LoadGamePalette(void)
-{
-	uint8 ptmp[192];
-	FILE *fp;
-	char *fn;
-
-	ipalette=0;
-
-	fn=strdup(FCEU_MakeFName(FCEUMKF_PALETTE,0,0).c_str());
-
-	if((fp=FCEUD_UTF8fopen(fn,"rb")))
-	{
-		int x;
-		fread(ptmp,1,192,fp);
-		fclose(fp);
-		for(x=0;x<64;x++)
-		{
-			palettei[x].r=ptmp[x+x+x];
-			palettei[x].g=ptmp[x+x+x+1];
-			palettei[x].b=ptmp[x+x+x+2];
-		}
-		ipalette=1;
-	}
-	free(fn);
+	//can't call FCEU_ResetPalette(), it would be re-entrant
+	//see precondition for this function
+	WritePalette();
 }
 
 void FCEU_ResetPalette(void)
@@ -275,38 +317,51 @@ void FCEU_ResetPalette(void)
 
 static void ChoosePalette(void)
 {
+	//if it's an NSF, there's NO palette? that's right, only the 'unvarying' palette will get used
 	if(GameInfo->type==GIT_NSF)
-		palo=0;
-	else if(ipalette)
-		palo=palettei;
-	else if(ntsccol && !PAL && GameInfo->type!=GIT_VSUNI)
+		palo = NULL;
+	//user palette takes priority over others
+	else if(palette_user_available)
+		palo = palette_user;
+	//NTSC takes priority next, if it's appropriate
+	else if(ntsccol_enable && !PAL && GameInfo->type!=GIT_VSUNI)
 	{
-		palo=paletten;
+		//for NTSC games, we can actually use the NTSC palette
+		palo = palette_ntsc;
 		CalculatePalette();
 	}
+	//select the game's overridden palette if available
+	else if(palette_game_available)
+		palo = palette_game;
+	//finally, use a default built-in palette
 	else
-		palo=palpoint[pale];
+	{
+		palo = default_palette[default_palette_selection];
+		//need to calcualte a deemph on the fly.. sorry. maybe support otherwise later
+		ApplyDeemphasisComplete(palo);
+	}
 }
 
 void WritePalette(void)
 {
 	int x;
 
+	//set the 'unvarying' palettes to low < 128 palette entries
 	for(x=0;x<7;x++)
-		FCEUD_SetPalette(x,unvpalette[x].r,unvpalette[x].g,unvpalette[x].b);
+		FCEUD_SetPalette(x,palette_unvarying[x].r,palette_unvarying[x].g,palette_unvarying[x].b);
+
 	if(GameInfo->type==GIT_NSF)
 	{
 		#ifdef _S9XLUA_H
 		FCEU_LuaUpdatePalette();
 		#endif
-		//for(x=0;x<128;x++)
-		// FCEUD_SetPalette(x,x,0,x);
 	}
 	else
 	{
+		//sets palette entries >= 128 with the 64 selected main colors
 		for(x=0;x<64;x++)
 			FCEUD_SetPalette(128+x,palo[x].r,palo[x].g,palo[x].b);
-		SetNESDeemph(lastd,1);
+		SetNESDeemph_OldHacky(lastd,1);
 	}
 }
 
@@ -321,7 +376,7 @@ static int controllength=0;
 
 void FCEUI_NTSCDEC(void)
 {
-	if(ntsccol && GameInfo->type!=GIT_VSUNI &&!PAL && GameInfo->type!=GIT_NSF)
+	if(ntsccol_enable && GameInfo->type!=GIT_VSUNI &&!PAL && GameInfo->type!=GIT_NSF)
 	{
 		int which;
 		if(controlselect)
@@ -343,7 +398,7 @@ void FCEUI_NTSCDEC(void)
 
 void FCEUI_NTSCINC(void)
 {
-	if(ntsccol && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF)
+	if(ntsccol_enable && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF)
 		if(controlselect)
 		{
 			if(controllength)
@@ -366,12 +421,12 @@ void FCEUI_NTSCINC(void)
 
 void FCEUI_NTSCSELHUE(void)
 {
-	if(ntsccol && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF){controlselect=1;controllength=360;}
+	if(ntsccol_enable && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF){controlselect=1;controllength=360;}
 }
 
 void FCEUI_NTSCSELTINT(void)
 {
-	if(ntsccol && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF){controlselect=2;controllength=360;}
+	if(ntsccol_enable && GameInfo->type!=GIT_VSUNI && !PAL && GameInfo->type!=GIT_NSF){controlselect=2;controllength=360;}
 }
 
 void FCEU_DrawNTSCControlBars(uint8 *XBuf)
diff --git a/trunk/src/palettes/palettes.h b/trunk/src/palettes/palettes.h
index 3149254a..f604538a 100644
--- a/trunk/src/palettes/palettes.h
+++ b/trunk/src/palettes/palettes.h
@@ -1,19 +1,30 @@
-pal rp2c04001[64] = {
+#define EMPTY_PALETTE_1 {0,0,0},
+#define EMPTY_PALETTE_4 EMPTY_PALETTE_1 EMPTY_PALETTE_1 EMPTY_PALETTE_1 EMPTY_PALETTE_1
+#define EMPTY_PALETTE_16 EMPTY_PALETTE_4 EMPTY_PALETTE_4 EMPTY_PALETTE_4 EMPTY_PALETTE_4
+#define EMPTY_PALETTE_64 EMPTY_PALETTE_16 EMPTY_PALETTE_16 EMPTY_PALETTE_16 EMPTY_PALETTE_16
+#define EMPTY_PALETTE_DEEMPH_X_7 EMPTY_PALETTE_64 EMPTY_PALETTE_64 EMPTY_PALETTE_64 EMPTY_PALETTE_64 EMPTY_PALETTE_64 EMPTY_PALETTE_64 EMPTY_PALETTE_64
+
+pal rp2c04001[512] = {
  #include "rp2c04001.h"
+	EMPTY_PALETTE_DEEMPH_X_7
 };
 
-pal rp2c04002[64] = {
+pal rp2c04002[512] = {
  #include "rp2c04002.h"
+	EMPTY_PALETTE_DEEMPH_X_7
 };
 
-pal rp2c04003[64] = {
+pal rp2c04003[512] = {
  #include "rp2c04003.h"
+	EMPTY_PALETTE_DEEMPH_X_7
 };
-pal rp2c05004[64] = {
+pal rp2c05004[512] = {
  #include "rp2c05004.h"
+	EMPTY_PALETTE_DEEMPH_X_7
 };
 
-pal unvpalette[7] = {
+//I think these are used for 'unvarying' palette entries, for the gui
+pal palette_unvarying[7] = {
 { 0x00<<2,0x00<<2,0x00<<2}, // Black
 { 0x3F<<2,0x3F<<2,0x34<<2}, // White
 { 0x00<<2,0x00<<2,0x00<<2}, // Black
@@ -25,7 +36,7 @@ pal unvpalette[7] = {
 
 
 // Default palette 
-pal palette[64] = {
+pal palette[512] = {
 
         { 0x1D<<2, 0x1D<<2, 0x1D<<2 }, /* Value 0 */
         { 0x09<<2, 0x06<<2, 0x23<<2 }, /* Value 1 */
@@ -292,4 +303,6 @@ slightly better but too dark:
 
 */
 
+EMPTY_PALETTE_DEEMPH_X_7
+
 };
diff --git a/trunk/src/ppu.cpp b/trunk/src/ppu.cpp
index 25cf94dc..bffb54c6 100644
--- a/trunk/src/ppu.cpp
+++ b/trunk/src/ppu.cpp
@@ -1223,6 +1223,7 @@ static void DoLine(void) {
 	// send all future writes to the invisible sanline. the easiest way to "skip" them altogether in old ppu
 	// todo: figure out what exactly should be skipped. it's known that there's no activity on PPU bus
 	uint8 *target = XBuf + ((scanline < 240 ? scanline : 240) << 8);
+	u8* dtarget = XDBuf + ((scanline < 240 ? scanline : 240) << 8);
 
 	if (MMC5Hack) MMC5_hb(scanline);
 
@@ -1236,19 +1237,23 @@ static void DoLine(void) {
 			col = Pal[0];
 		else col = gNoBGFillColor;
 		tem = col | (col << 8) | (col << 16) | (col << 24);
-		tem |= 0x40404040;
+		tem |= 0x40404040; 
 		FCEU_dwmemset(target, tem, 256);
 	}
 
 	if (SpriteON)
 		CopySprites(target);
 
-	if (ScreenON || SpriteON) {	// Yes, very el-cheapo.
+	//greyscale handling (mask some bits off the color) ? ? ?
+	if (ScreenON || SpriteON)
+	{
 		if (PPU[1] & 0x01) {
 			for (x = 63; x >= 0; x--)
 				*(uint32*)&target[x << 2] = (*(uint32*)&target[x << 2]) & 0x30303030;
 		}
 	}
+
+	//some pathetic attempts at deemph
 	if ((PPU[1] >> 5) == 0x7) {
 		for (x = 63; x >= 0; x--)
 			*(uint32*)&target[x << 2] = ((*(uint32*)&target[x << 2]) & 0x3f3f3f3f) | 0xc0c0c0c0;
@@ -1259,6 +1264,10 @@ static void DoLine(void) {
 		for (x = 63; x >= 0; x--)
 			*(uint32*)&target[x << 2] = ((*(uint32*)&target[x << 2]) & 0x3f3f3f3f) | 0x80808080;
 
+	//write the actual deemph
+	for (x = 63; x >= 0; x--)
+		*(uint32*)&dtarget[x << 2] = ((PPU[1]>>5)<<0)|((PPU[1]>>5)<<8)|((PPU[1]>>5)<<16)|((PPU[1]>>5)<<24);
+
 	sphitx = 0x100;
 
 	if (ScreenON || SpriteON)
@@ -1785,7 +1794,6 @@ int FCEUPPU_Loop(int skip) {
 		}
 		#endif
 		else {
-			int x, max, maxref;
 			deemp = PPU[1] >> 5;
 
 			// manual samples can't play correctly with overclocking
@@ -1803,14 +1811,16 @@ int FCEUPPU_Loop(int skip) {
 			DMC_7bit = 0;
 
 			if (MMC5Hack) MMC5_hb(scanline);
-			for (x = 1, max = 0, maxref = 0; x < 7; x++) {
+
+			//deemph nonsense, kept for complicated reasons (see SetNESDeemph_OldHacky implementation)
+			for (int x = 1, max = 0, maxref = 0; x < 7; x++) {
 				if (deempcnt[x] > max) {
 					max = deempcnt[x];
 					maxref = x;
 				}
 				deempcnt[x] = 0;
 			}
-			SetNESDeemph(maxref, 0);
+			SetNESDeemph_OldHacky(maxref, 0);
 		}
 	}	//else... to if(ppudead)
 
@@ -2079,7 +2089,9 @@ int FCEUX_PPU_Loop(int skip) {
 					int xstart = xt << 3;
 					oamcount = oamcounts[renderslot];
 					uint8 * const target = XBuf + (yp << 8) + xstart;
+					uint8 * const dtarget = XDBuf + (yp << 8) + xstart;
 					uint8 *ptr = target;
+					uint8 *dptr = dtarget;
 					int rasterpos = xstart;
 
 					//check all the conditions that can cause things to render in these 8px
@@ -2147,6 +2159,7 @@ int FCEUX_PPU_Loop(int skip) {
 						}
 
 						*ptr++ = PaletteAdjustPixel(pixelcolor);
+						*dptr++= PPU[1]>>5; //grap deemph
 					}
 				}
 			}
diff --git a/trunk/src/video.cpp b/trunk/src/video.cpp
index ac831fa8..76a5cce4 100644
--- a/trunk/src/video.cpp
+++ b/trunk/src/video.cpp
@@ -56,10 +56,17 @@
 #include <cstdarg>
 #include <zlib.h>
 
-uint8 *XBuf=NULL;
-uint8 *XBackBuf=NULL;
+//XBuf:
+//0-63 is reserved for 7 special colours used by FCEUX (overlay, etc.)
+//64-127 is the most-used emphasis setting per frame
+//128-195 is the palette with no emphasis
+//196-255 is the palette with all emphasis bits on
+u8 *XBuf=NULL; //used for current display
+u8 *XBackBuf=NULL; //ppu output is stashed here before drawing happens
+u8 *XDBuf=NULL; //corresponding to XBuf but with deemph bits
+u8 *XDBackBuf=NULL; //corresponding to XBackBuf but with deemph bits
 int ClipSidesOffset=0;	//Used to move displayed messages when Clips left and right sides is checked
-static uint8 *xbsave=NULL;
+static u8 *xbsave=NULL;
 
 GUIMESSAGE guiMessage;
 GUIMESSAGE subtitleMessage;
@@ -105,29 +112,35 @@ void FCEU_KillVirtualVideo(void)
 **/
 int FCEU_InitVirtualVideo(void)
 {
-	if(!XBuf)		/* Some driver code may allocate XBuf externally. */
-		/* 256 bytes per scanline, * 240 scanline maximum, +16 for alignment,
-		*/
-
-		if(!(XBuf= (uint8*) (FCEU_malloc(256 * 256 + 16))) ||
-			!(XBackBuf= (uint8*) (FCEU_malloc(256 * 256 + 16))))
-		{
-			return 0;
-		}
-
-		xbsave = XBuf;
-
-		if( sizeof(uint8*) == 4 )
-		{
-			uintptr_t m = (uintptr_t)XBuf;
-			m = ( 8 - m) & 7;
-			XBuf+=m;
-		}
-
-		memset(XBuf,128,256*256); //*240);
-		memset(XBackBuf,128,256*256);
-
+	//Some driver code may allocate XBuf externally.
+	//256 bytes per scanline, * 240 scanline maximum, +16 for alignment,
+	if(XBuf)
 		return 1;
+	
+	XBuf = (u8*)FCEU_malloc(256 * 256 + 16);
+	XBackBuf = (u8*)FCEU_malloc(256 * 256 + 16);
+	XDBuf = (u8*)FCEU_malloc(256 * 256 + 16);
+	XDBackBuf = (u8*)FCEU_malloc(256 * 256 + 16);
+	if(!XBuf || !XBackBuf || !XDBuf || !XDBackBuf)
+	{
+		return 0;
+	}
+
+	xbsave = XBuf;
+
+	if( sizeof(uint8*) == 4 )
+	{
+		uintptr_t m = (uintptr_t)XBuf;
+		m = ( 8 - m) & 7;
+		XBuf+=m;
+	}
+
+	memset(XBuf,128,256*256);
+	memset(XBackBuf,128,256*256);
+	memset(XBuf,128,256*256);
+	memset(XBackBuf,128,256*256);
+
+	return 1;
 }
 
 #ifdef FRAMESKIP
diff --git a/trunk/src/video.h b/trunk/src/video.h
index 167a75d1..79c6fec3 100644
--- a/trunk/src/video.h
+++ b/trunk/src/video.h
@@ -9,6 +9,8 @@ uint32 GetScreenPixel(int x, int y, bool usebackup);
 int GetScreenPixelPalette(int x, int y, bool usebackup);
 extern uint8 *XBuf;
 extern uint8 *XBackBuf;
+extern uint8 *XDBuf;
+extern uint8 *XDBackBuf;
 extern int ClipSidesOffset;
 extern struct GUIMESSAGE
 {
diff --git a/trunk/src/vsuni.cpp b/trunk/src/vsuni.cpp
index e7e718a3..5755474b 100644
--- a/trunk/src/vsuni.cpp
+++ b/trunk/src/vsuni.cpp
@@ -301,7 +301,7 @@ void FCEU_VSUniCheck(uint64 md5partial, int *MapperNo, uint8 *Mirroring) {
 
 	while (vs->name) {
 		if (md5partial == vs->md5partial) {
-			if (vs->ppu < RCP2C03B) pale = vs->ppu;
+			if (vs->ppu < RCP2C03B) default_palette_selection = vs->ppu;
 			*MapperNo = vs->mapper;
 			*Mirroring = vs->mirroring;
 			GameInfo->type = GIT_VSUNI;