Adding new 3D engine by Ian Curtis. Old engine moved to Graphics/Legacy3D/.

Makefiles/Makefile.SDL.Win32.GCC

@@ -38,7 +38,7 @@ BITS = 64
 #
 # Include console-based debugger in emulator ('yes' or 'no')
 #
-ENABLE_DEBUGGER = no
+ENABLE_DEBUGGER = yes
 
 
 ###############################################################################
@@ -47,6 +47,11 @@ ENABLE_DEBUGGER = no
 # Edit these paths as necessary.
 ###############################################################################
 
+#
+# Boost 1.55
+#
+BOOST_INCLUDEPATH = /mingw64/boost_1_55_0
+
 #
 # SDL
 #
@@ -90,11 +95,11 @@ OUTFILE = $(BIN_DIR)\Supermodel.exe
 CC = gcc
 CXX = g++
 LD = g++
-COMPILER_FLAGS = -I$(SDL_INCLUDEPATH) -ISrc/ -ISrc/OSD/ -ISrc/OSD/SDL/ -ISrc/OSD/Windows/ -c -Wall -O3 -DSUPERMODEL_WIN32 -DGLEW_STATIC 
+COMPILER_FLAGS = -I$(SDL_INCLUDEPATH) -ISrc/ -ISrc/OSD/ -ISrc/OSD/SDL/ -ISrc/OSD/Windows/ -c -Wall -DSUPERMODEL_WIN32 -DGLEW_STATIC -O3
 CFLAGS = $(COMPILER_FLAGS)
-CPPFLAGS = $(COMPILER_FLAGS)
+CPPFLAGS = $(COMPILER_FLAGS) -I$(BOOST_INCLUDEPATH) -std=c++11
 #LFLAGS = -s -o $(OUTFILE) $(OBJ) -L$(SDL_LIBPATH) -lmingw32 -lSDLmain -lSDL -lopengl32 -lglu32 -ldinput8 -ldxguid -lole32 -loleaut32 -lz -l:$(WINSDK_LIBPATH)/WbemUuid.lib
-LFLAGS = -s -o $(OUTFILE) $(OBJ) -L$(SDL_LIBPATH) -lmingw32 -lSDLmain -lSDL -lopengl32 -lglu32 -ldinput8 -ldxguid -lole32 -loleaut32 -lz -l:$(WINSDK_LIBPATH)/WbemUuid.lib
+LFLAGS = -o $(OUTFILE) $(OBJ) -L$(SDL_LIBPATH) -lmingw32 -lSDLmain -lSDL -lopengl32 -lglu32 -ldinput8 -ldxguid -lole32 -loleaut32 -lz -l:$(WINSDK_LIBPATH)/WbemUuid.lib -s
 
 #
 # Build options...
@@ -115,7 +120,9 @@ endif
 #
 OBJ = $(OBJ_DIR)/PPCDisasm.o $(OBJ_DIR)/Games.o $(OBJ_DIR)/Config.o $(OBJ_DIR)/INIFile.o $(OBJ_DIR)/BlockFile.o $(OBJ_DIR)/93C46.o \
 	$(OBJ_DIR)/ROMLoad.o $(OBJ_DIR)/unzip.o $(OBJ_DIR)/ioapi.o $(OBJ_DIR)/Error.o $(OBJ_DIR)/glew.o $(OBJ_DIR)/Shader.o \
-	$(OBJ_DIR)/Real3D.o $(OBJ_DIR)/Render3D.o $(OBJ_DIR)/Models.o $(OBJ_DIR)/TextureRefs.o $(OBJ_DIR)/Render2D.o $(OBJ_DIR)/TileGen.o \
+	$(OBJ_DIR)/Real3D.o $(OBJ_DIR)/Legacy3D.o $(OBJ_DIR)/Models.o $(OBJ_DIR)/TextureRefs.o \
+	$(OBJ_DIR)/New3D.o $(OBJ_DIR)/Mat4.o $(OBJ_DIR)/Model.o $(OBJ_DIR)/PolyHeader.o $(OBJ_DIR)/Texture.o $(OBJ_DIR)/TextureSheet.o $(OBJ_DIR)/VBO.o $(OBJ_DIR)/Vec.o $(OBJ_DIR)/R3DShader.o \
+	$(OBJ_DIR)/Render2D.o $(OBJ_DIR)/TileGen.o \
 	$(OBJ_DIR)/Model3.o $(OBJ_DIR)/ppc.o $(OBJ_DIR)/Main.o $(OBJ_DIR)/Audio.o $(OBJ_DIR)/Thread.o $(OBJ_DIR)/SoundBoard.o \
 	$(OBJ_DIR)/SCSP.o $(OBJ_DIR)/SCSPDSP.o $(OBJ_DIR)/68K.o $(OBJ_DIR)/m68kcpu.o $(OBJ_DIR)/m68kopnz.o $(OBJ_DIR)/m68kopdm.o \
 	$(OBJ_DIR)/m68kopac.o $(OBJ_DIR)/m68kops.o $(OBJ_DIR)/DSB.o $(OBJ_DIR)/Z80.o \
@@ -143,6 +150,10 @@ endif
 all:	$(BIN_DIR) $(OBJ_DIR) $(OBJ)
 	$(LD) $(LFLAGS)
 
+ppcd: $(BIN_DIR) $(OBJ_DIR)
+	$(CXX) Src/CPU/PowerPC/PPCDisasm.cpp $(CPPFLAGS) -DSTANDALONE -o $(OBJ_DIR)/ppcd.o
+	$(LD) -o $(BIN_DIR)/ppcd.exe -mconsole $(OBJ_DIR)/ppcd.o
+
 $(BIN_DIR):
 	mkdir $(BIN_DIR)
 	
@@ -195,7 +206,13 @@ $(OBJ_DIR)/%.o:	Src/%.cpp
 $(OBJ_DIR)/%.o:	Src/Model3/%.cpp
 	$(CXX) $< $(CPPFLAGS) -o $(OBJ_DIR)/$(*F).o
 
-$(OBJ_DIR)/%.o:	Src/Graphics/%.cpp Src/Graphics/Shaders2D.h Src/Graphics/Shaders3D.h
+$(OBJ_DIR)/%.o:	Src/Graphics/%.cpp Src/Graphics/Shaders2D.h
+	$(CXX) $< $(CPPFLAGS) -o $(OBJ_DIR)/$(*F).o
+	
+$(OBJ_DIR)/%.o:	Src/Graphics/Legacy3D/%.cpp Src/Graphics/Legacy3D/Shaders3D.h
+	$(CXX) $< $(CPPFLAGS) -o $(OBJ_DIR)/$(*F).o
+
+$(OBJ_DIR)/%.o:	Src/Graphics/New3D/%.cpp Src/Graphics/New3D/%.h
 	$(CXX) $< $(CPPFLAGS) -o $(OBJ_DIR)/$(*F).o
 
 $(OBJ_DIR)/%.o:	Src/Sound/%.cpp

Src/Graphics/IRender3D.h

@@ -0,0 +1,27 @@
+#ifndef INCLUDED_IRENDER3D_H
+#define INCLUDED_IRENDER3D_H
+
+#include <cstdint>
+
+/*
+ * IRender3D:
+ *
+ * Interface (abstract base class) for Real3D rendering engine.
+ */
+class IRender3D
+{
+public:
+  virtual void RenderFrame(void) = 0;
+  virtual void BeginFrame(void) = 0;
+  virtual void EndFrame(void) = 0;
+  virtual void UploadTextures(unsigned x, unsigned y, unsigned width, unsigned height) = 0;
+  virtual void AttachMemory(const uint32_t *cullingRAMLoPtr, const uint32_t *cullingRAMHiPtr, const uint32_t *polyRAMPtr, const uint32_t *vromPtr, const uint16_t *textureRAMPtr) = 0;
+  virtual void SetStep(int stepID) = 0;
+  virtual bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes) = 0;
+  
+  virtual ~IRender3D()
+  {
+  }
+};
+
+#endif  // INCLUDED_IRENDER3D_H

Src/Graphics/Legacy3D/Error.cpp

@@ -0,0 +1,67 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Error.cpp
+ * 
+ * Error reporting for 3D renderer. There are some situations in which it is 
+ * appropriate to inform the user of an error during rendering. However, these 
+ * will frequently lead to an avalanche of error messages. Therefore, error 
+ * messages are managed by this interface, ensuring that they are displayed not
+ * more than once per frame.
+ *
+ * Error functions can always be counted on to return FAIL, like ErrorLog().
+ */
+
+#include "Supermodel.h"
+
+namespace Legacy3D {
+
+// Error bit flags (must not conflict)
+#define ERROR_LOCAL_VERTEX_OVERFLOW	0x1
+#define ERROR_UNABLE_TO_CACHE_MODEL	0x2
+
+
+// Overflow in the local vertex buffer, which holds one model
+bool CLegacy3D::ErrorLocalVertexOverflow(void)
+{
+	if ((errorMsgFlags&ERROR_LOCAL_VERTEX_OVERFLOW))
+		return FAIL;
+	errorMsgFlags |= ERROR_LOCAL_VERTEX_OVERFLOW;
+	return ErrorLog("Overflow in local vertex buffer!");
+}
+
+// Model could not be cached, even after dumping display list and re-caching
+bool CLegacy3D::ErrorUnableToCacheModel(UINT32 modelAddr)
+{
+	if ((errorMsgFlags&ERROR_UNABLE_TO_CACHE_MODEL))
+		return FAIL;
+	errorMsgFlags |= ERROR_UNABLE_TO_CACHE_MODEL;
+	return ErrorLog("Encountered a model that was too large to cache (at %08X)!", modelAddr);
+}
+	
+// Call this every frame to clear the error flag, allowing errors to be printed
+void CLegacy3D::ClearErrors(void)
+{
+	errorMsgFlags = 0;
+}
+
+} // Legacy3D

Src/Graphics/Legacy3D/Legacy3D.h

@@ -0,0 +1,485 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Legacy3D.h
+ * 
+ * Header file defining the CLegacy3D class: Supermodel's original OpenGL 
+ * Real3D graphics engine.
+ */
+
+#ifndef INCLUDED_LEGACY3D_H
+#define INCLUDED_LEGACY3D_H
+
+#include "Graphics/IRender3D.h"
+#include "Pkgs/glew.h"
+
+namespace Legacy3D {
+
+/******************************************************************************
+ Internal Definitions and Data Structures
+ 
+ NOTE: These should probably be moved inside the Legacy3D namespace at some
+ point.
+******************************************************************************/
+
+// Model caches sort models by alpha (translucency) state
+enum POLY_STATE
+{
+	POLY_STATE_NORMAL = 0,
+	POLY_STATE_ALPHA
+};
+
+struct Vertex
+{
+	GLfloat	x,y,z;	// vertex
+	GLfloat	n[3];	// normal X, Y, Z
+	GLfloat	u,v;	// texture U, V coordinates (in texels, relative to selected texture)
+};
+
+struct Poly
+{
+	Vertex			Vert[4];
+	GLfloat			n[3];		// polygon normal (used for backface culling)
+	POLY_STATE		state;		// alpha or normal?
+	unsigned		numVerts;	// triangle (3) or quad (4)
+	const UINT32	*header;	// pointer to Real3D 7-word polygon header
+};
+
+/*
+ * VBORef:
+ *
+ * Reference to model polygons stored in a VBO. Each reference has two sets of
+ * vertices: normal and alpha. Copies of the model with different texture
+ * offsets applied are searchable via the linked list of texture offset states.
+ */
+
+struct VBORef
+{
+	unsigned		index[2];		// index of model polygons in VBO
+	unsigned		numVerts[2];	// number of vertices
+	unsigned		lutIdx;			// LUT index associated with this model (for fast LUT clearing)
+	
+	struct VBORef	*nextTexOffset;	// linked list of models with different texture offset states
+	UINT16			texOffset;		// texture offset data for this model
+	
+	CTextureRefs	texRefs;		// unique texture references contained in this model
+	
+	/*
+	 * Clear():
+	 *
+	 * Clears the VBORef by setting all fields to 0 and clearing the texture
+	 * references.
+	 */
+	inline void Clear(void)
+	{
+		texRefs.Clear();
+		lutIdx = 0;
+		texOffset = 0;
+		nextTexOffset = NULL;
+		for (int i = 0; i < 2; i++)
+		{
+			index[i] = 0;
+			numVerts[i] = 0;
+		}
+	}
+};
+
+// Display list items: model instances and viewport settings
+struct DisplayList
+{
+	bool		isViewport;				// if true, this is a viewport node
+	
+	union
+	{
+		// Viewport data
+		struct
+		{
+			GLfloat		projectionMatrix[4*4];	// projection matrix
+			GLfloat		lightingParams[6];		// lighting parameters (see RenderViewport() and vertex shader)
+			GLfloat		spotEllipse[4];			// spotlight ellipse (see RenderViewport())
+			GLfloat		spotRange[2];			// Z range
+			GLfloat		spotColor[3];			// color
+			GLfloat		fogParams[5];			// fog parameters (...)
+			GLint		x, y;					// viewport coordinates (scaled and in OpenGL format)
+			GLint		width, height;			// viewport dimensions (scaled for display surface size)
+		} Viewport;
+		
+		// Model data
+		struct
+		{
+			GLfloat		modelViewMatrix[4*4];	// model-view matrix
+			unsigned	index;					// index in VBO
+			unsigned	numVerts;				// number of vertices
+			GLint		frontFace;				// GL_CW (default), GL_CCW, or -GL_CW to indicate no culling
+		} Model;
+	} Data;
+			
+	DisplayList	*next;					// next display list item with the same state (alpha or non-alpha)
+};
+
+/*
+ * ModelCache:
+ *
+ * A model cache tracks all models in a particular region (ie., VROM or polygon
+ * RAM). It contains a look-up table to quickly obtain VBO indices. Be careful
+ * when accessing the LUT, there are some special cases.
+ *
+ * If the model cache is marked dynamic, cached models may not necessarily be
+ * retained. Clearing the model cache is also much faster. The LUT entry for 
+ * the last model cached will be valid, but because the LUT may not be
+ * cleared, one cannot assume a model exists because there is a LUT entry
+ * pointing to it. Always use NeedToCache() to determine whether caching is
+ * necessary before reading the LUT!
+ */
+struct ModelCache
+{
+	// Cache type
+	bool		dynamic;
+	
+	// Vertex buffer object
+	unsigned	vboMaxOffset;	// size of VBO (in bytes)
+	unsigned	vboCurOffset;	// current offset in VBO (in bytes)
+	GLuint		vboID;			// OpenGL VBO handle
+	
+	// Local vertex buffers (enough for a single model)
+	unsigned	maxVertIdx;		// size of each local vertex buffer (in vertices)
+	unsigned	curVertIdx[2];	// current vertex index (in vertices)
+	GLfloat		*verts[2];
+	
+	// Array of cached models
+	unsigned	maxModels;	// maximum number of models
+	unsigned	numModels;	// current number stored
+	VBORef		*Models;
+	
+	/*
+	 * Look-Up Table:
+	 *
+	 * Can be accessed directly with a LUT index to determine the model index.
+	 * However, it should not be used to determine whether a model needs to be
+	 * cached. Use NeedToCache() instead. A valid index, for example, may still
+	 * have to be re-cached if the model cache is dynamic (polygon RAM).
+	 */
+	unsigned	lutSize;	// number of elements in LUT
+	INT16		*lut;		// stores indices into Models[] or -1 if not yet cached
+
+	// Display list
+	unsigned	maxListSize;	// maximum number of display list items
+	unsigned	listSize;		// number of items in display list
+	DisplayList	*List;			// holds all display list items
+	DisplayList	*ListHead[2];	// heads of linked lists for each state
+	DisplayList	*ListTail[2];	// current tail node for each state
+};
+
+struct TexSheet
+{
+	unsigned sheetNum;
+	unsigned mapNum;
+	unsigned xOffset;
+	unsigned yOffset;
+
+	/*
+	 * Texture Format Buffer
+	 *
+	 * Records the format that a texture (at a given location within the
+	 * texture sheet) is currently stored in. A negative value indicates the
+	 * texture has not been accessed and converted yet and non-negative values
+	 * correspond to the texture format bits in the polygon headers. They can
+	 * be used to determine whether a texture needs to be updated.
+	 */
+	int	 texWidth[2048/32][2048/32];
+	int	 texHeight[2048/32][2048/32];
+	INT8 texFormat[2048/32][2048/32];
+};
+
+/******************************************************************************
+ CLegacy3D Classes
+******************************************************************************/
+
+/*
+ * CLegacy3DConfig:
+ *
+ * Settings used by CLegacy3D.
+ */
+class CLegacy3DConfig
+{
+public:
+	string 		vertexShaderFile;	// path to vertex shader or "" to use internal shader
+	string 		fragmentShaderFile;	// fragment shader
+	unsigned	maxTexMaps;       	// maximum number of texture maps to use (1-9)
+	unsigned    maxTexMapExtent;    // maximum extent of texture maps (where num of tex sheets per map = extent ^ 2)
+	bool		multiTexture;		// if enabled and no external fragment shader, select internal shader w/ multiple texture sheet support
+
+	// Defaults
+	CLegacy3DConfig(void)
+	{
+		// strings will be clear to begin with
+		maxTexMaps = 9;
+		maxTexMapExtent = 4;
+		multiTexture = false;
+	}
+};
+
+/*
+ * CLegacy3D:
+ *
+ * 3D renderer. Lots of work to do here :)
+ */
+class CLegacy3D: public IRender3D
+{
+	friend class CTextureRefs;
+
+public:
+	/*
+	 * RenderFrame(void):
+	 *
+	 * Renders the complete scene database. Must be called between BeginFrame() and
+	 * EndFrame(). This function traverses the scene database and builds up display 
+	 * lists.
+	 */
+	void RenderFrame(void);
+	
+	/*
+	 * BeginFrame(void):
+	 *
+	 * Prepare to render a new frame. Must be called once per frame prior to
+	 * drawing anything.
+	 */
+	void BeginFrame(void);
+	
+	/*
+	 * EndFrame(void):
+	 *
+	 * Signals the end of rendering for this frame. Must be called last during
+	 * the frame.
+	 */
+	void EndFrame(void);
+	
+	/*
+	 * UploadTextures(x, y, width, height):
+	 *
+	 * Signals that a portion of texture RAM has been updated.
+	 *
+	 * Parameters:
+	 *		x		X position within texture RAM.
+	 *		y		Y position within texture RAM.
+	 *		width	Width of texture data in texels.
+	 *		height	Height.
+	 */
+	void UploadTextures(unsigned x, unsigned y, unsigned width, unsigned height);
+	
+	/*
+	 * AttachMemory(cullingRAMLoPtr, cullingRAMHiPtr, polyRAMPtr, vromPtr,
+	 * 				textureRAMPtr):
+	 *
+	 * Attaches RAM and ROM areas. This must be done prior to any rendering
+	 * otherwise the program may crash with an access violation.
+	 *
+	 * Parameters:
+	 *		cullingRAMLoPtr		Pointer to low culling RAM (4 MB).
+	 *		cullingRAMHiPtr		Pointer to high culling RAM (1 MB).
+	 *		polyRAMPtr			Pointer to polygon RAM (4 MB).
+	 *		vromPtr				Pointer to video ROM (64 MB).
+	 *		textureRAMPtr		Pointer to texture RAM (8 MB).
+	 */
+	void AttachMemory(const UINT32 *cullingRAMLoPtr,
+					  const UINT32 *cullingRAMHiPtr, const UINT32 *polyRAMPtr,
+					  const UINT32 *vromPtr, const UINT16 *textureRAMPtr);
+
+	/*
+	 * SetStep(stepID):
+	 *
+	 * Sets the Model 3 hardware stepping, which also determines the Real3D
+	 * functionality. The default is Step 1.0. This should be called prior to 
+	 * any other emulation functions and after Init().
+	 *
+	 * Parameters:
+	 *		stepID	0x10 for Step 1.0, 0x15 for Step 1.5, 0x20 for Step 2.0,
+	 *				or 0x21 for Step 2.1. Anything else defaults to 1.0.
+	 */
+	void SetStep(int stepID);
+	
+	/*
+	 * Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes):
+	 *
+	 * One-time initialization of the context. Must be called before any other
+	 * members (meaning it should be called even before being attached to any
+	 * other objects that want to use it).
+	 *
+	 * External shader files are loaded according to configuration settings.
+	 *
+	 * Parameters:
+	 *		xOffset		X offset of the viewable area within OpenGL display 
+	 *                  surface, in pixels.
+	 *		yOffset		Y offset.
+	 *		xRes		Horizontal resolution of the viewable area.
+	 *		yRes		Vertical resolution.
+	 *		totalXRes	Horizontal resolution of the complete display area.
+	 *		totalYRes	Vertical resolution.
+	 *
+	 * Returns:
+	 *		OKAY is successful, otherwise FAILED if a non-recoverable error
+	 *		occurred. Any allocated memory will not be freed until the
+	 *		destructor is called. Prints own error messages.
+	 */
+	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes);
+	 
+	/*
+	 * CLegacy3D(void):
+	 * ~CLegacy3D(void):
+	 *
+	 * Constructor and destructor.
+	 */
+	CLegacy3D(void);
+	~CLegacy3D(void);
+
+private:
+	/*
+	 * Private Members
+	 */
+	
+	// Real3D address translation
+	const UINT32 *TranslateCullingAddress(UINT32 addr);
+	const UINT32 *TranslateModelAddress(UINT32 addr);
+	
+	// Model caching and display list management
+	void 			DrawDisplayList(ModelCache *Cache, POLY_STATE state);
+	bool 			AppendDisplayList(ModelCache *Cache, bool isViewport, const struct VBORef *Model);
+	void 			ClearDisplayList(ModelCache *Cache);
+	bool 			InsertPolygon(ModelCache *cache, const Poly *p);
+	void 			InsertVertex(ModelCache *cache, const Vertex *v, const Poly *p, float normFlip);
+	struct VBORef	*BeginModel(ModelCache *cache);
+	void			EndModel(ModelCache *cache, struct VBORef *Model, int lutIdx, UINT16 texOffset);
+	struct VBORef	*CacheModel(ModelCache *cache, int lutIdx, UINT16 texOffset, const UINT32 *data);
+	struct VBORef	*LookUpModel(ModelCache *cache, int lutIdx, UINT16 texOffset);
+	void 			ClearModelCache(ModelCache *cache);
+	bool 			CreateModelCache(ModelCache *cache, unsigned vboMaxVerts, unsigned localMaxVerts, unsigned maxNumModels, unsigned numLUTEntries, unsigned displayListSize, bool isDynamic);
+	void 			DestroyModelCache(ModelCache *cache);
+	
+	// Texture management
+	void DecodeTexture(int format, int x, int y, int width, int height);
+	
+	// Matrix stack
+	void	MultMatrix(UINT32 matrixOffset);
+	void 	InitMatrixStack(UINT32 matrixBaseAddr);
+	
+	// Scene database traversal
+	bool DrawModel(UINT32 modelAddr);
+	void DescendCullingNode(UINT32 addr);
+	void DescendPointerList(UINT32 addr);
+	void DescendNodePtr(UINT32 nodeAddr);
+	void RenderViewport(UINT32 addr, int pri);
+	
+	// In-frame error reporting
+	bool ErrorLocalVertexOverflow(void);
+	bool ErrorUnableToCacheModel(UINT32 modelAddr);
+	void ClearErrors(void);
+	
+	/*
+	 * Data
+	 */
+	
+	// Stepping
+	int		step;
+	int		offset;			// offset to subtract for words 3 and higher of culling nodes
+	GLfloat	vertexFactor;	// fixed-point conversion factor for vertices
+	
+	// Memory (passed from outside)
+	const UINT32	*cullingRAMLo;	// 4 MB
+	const UINT32	*cullingRAMHi;	// 1 MB
+	const UINT32	*polyRAM;		// 4 MB
+	const UINT32	*vrom;			// 64 MB
+	const UINT16	*textureRAM;	// 8 MB
+	
+	// Error reporting
+	unsigned	errorMsgFlags;	// tracks which errors have been printed this frame
+	
+	// Real3D Base Matrix Pointer
+	const float	*matrixBasePtr;
+	
+	// Current viewport parameters (updated as viewports are traversed)
+	GLfloat	lightingParams[6];
+	GLfloat	fogParams[5];
+	GLfloat	spotEllipse[4];
+	GLfloat	spotRange[2];
+	GLfloat	spotColor[3];
+	GLint	viewportX, viewportY;
+	GLint	viewportWidth, viewportHeight;
+	
+	// Scene graph stack
+	int		listDepth;	// how many lists have we recursed into
+	int		stackDepth;	// for debugging and error handling purposes
+	
+	// Texture offset (during scene graph processing)
+	GLfloat	texOffsetXY[2];	// decoded X, Y offsets
+	UINT16	texOffset;		// raw texture offset data as it appears in culling node
+	
+	// Resolution and scaling factors (to support resolutions higher than 496x384) and offsets
+	GLfloat		xRatio, yRatio;
+	unsigned	xOffs, yOffs;
+	unsigned 	totalXRes, totalYRes;
+	
+	// Texture details
+	static int	defaultFmtToTexSheetNum[8];  // default mapping from Model3 texture format to texture sheet	
+	unsigned    numTexMaps;                  // total number of texture maps
+	GLuint		texMapIDs[9];                // GL texture IDs of texture maps
+	unsigned    numTexSheets;                // total number of texture sheets
+	TexSheet   *texSheets;                   // texture sheet objects
+	TexSheet   *fmtToTexSheet[8];            // final mapping from Model3 texture format to texture sheet
+	
+	// Shader programs and input data locations
+	GLuint	shaderProgram;			// shader program object
+	GLuint	vertexShader;			// vertex shader handle
+	GLuint	fragmentShader;			// fragment shader
+	GLint   textureMapLoc;          // location of "textureMap" uniform (if available)
+	GLint	textureMapLocs[8];		// location of "textureMap[0-7]" uniforms (if available)
+	GLint	modelViewMatrixLoc;		// uniform
+	GLint	projectionMatrixLoc;	// uniform
+	GLint	lightingLoc;			// uniform
+	GLint   mapSizeLoc;             // uniform
+	GLint	spotEllipseLoc;			// uniform
+	GLint	spotRangeLoc;			// uniform
+	GLint	spotColorLoc;			// uniform
+	GLint	subTextureLoc;			// attribute
+	GLint	texParamsLoc;			// attribute
+	GLint	texFormatLoc;			// attribute
+	GLint   texMapLoc;              // attribute
+	GLint	transLevelLoc;			// attribute
+	GLint	lightEnableLoc;			// attribute
+	GLint	shininessLoc;			// attribute
+	GLint	fogIntensityLoc;		// attribute
+	
+	// Model caching
+	ModelCache	VROMCache;	// VROM (static) models
+	ModelCache	PolyCache;	// polygon RAM (dynamic) models
+	
+	/*
+ 	 * Texture Decode Buffer
+ 	 *
+ 	 * Textures are decoded and copied from texture RAM into this temporary buffer
+ 	 * before being uploaded. Dimensions are 512x512.
+ 	 */
+	GLfloat	*textureBuffer;	// RGBA8 format
+};
+
+} // Legacy3D
+
+#endif	// INCLUDED_LEGACY3D_H

Src/Graphics/Legacy3D/Shaders/DIR.txt

@@ -0,0 +1,2 @@
+Shader source files go here. Completed versions of the shader files should be
+copied into Src/Graphics/Shaders3D.h and Src/Graphics/Shaders2D.h.

Src/Graphics/Legacy3D/Shaders/Fragment.glsl

@@ -0,0 +1,194 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Fragment.glsl
+ *
+ * Fragment shader for 3D rendering.
+ */
+
+#version 120
+
+// Global uniforms
+uniform sampler2D	textureMap;		// complete texture map, 2048x2048 texels
+uniform vec4		spotEllipse;	// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)
+uniform vec2		spotRange;		// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+uniform vec3		spotColor;		// spotlight RGB color
+uniform vec3		lighting[2];	// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)
+uniform float		mapSize;		// texture map size (2048,4096,6144 etc)
+
+// Inputs from vertex shader 
+varying vec4		fsSubTexture;	// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)
+varying vec4		fsTexParams;	// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode
+varying float		fsTexFormat;	// T1RGB5 contour texture (if > 0)
+varying float		fsTexMap;		// texture map number
+varying float		fsTransLevel;	// translucence level, 0.0 (transparent) to 1.0 (opaque)
+varying vec3		fsLightIntensity;	// lighting intensity 
+varying float		fsSpecularTerm;	// specular highlight
+varying float		fsFogFactor;	// fog factor
+varying float		fsViewZ;		// Z distance to fragment from viewpoint at origin
+
+/*
+ * WrapTexelCoords():
+ *
+ * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture
+ * sheet, taking into account wrapping behavior.
+ *
+ * Computing normalized OpenGL texture coordinates (0 to 1) within the 
+ * Real3D texture sheet:
+ *
+ * If the texture is not mirrored, we simply have to clamp the
+ * coordinates to fit within the texture dimensions, add the texture
+ * X, Y position to select the appropriate one, and normalize by 2048
+ * (the dimensions of the Real3D texture sheet).
+ *
+ *		= [(u,v)%(w,h)+(x,y)]/(2048,2048)
+ *
+ * If mirroring is enabled, textures are mirrored every odd multiple of
+ * the original texture. To detect whether we are in an odd multiple, 
+ * simply divide the coordinate by the texture dimension and check 
+ * whether the result is odd. Then, clamp the coordinates as before but
+ * subtract from the last texel to mirror them:
+ *
+ * 		= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)
+ *		where M is 1.0 if the texture must be mirrored.
+ *
+ * As an optimization, this function computes TWO texture coordinates
+ * simultaneously. The first is texCoord.xy, the second is in .zw. The other
+ * parameters must have .xy = .zw.
+ */
+vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)
+{
+	vec4	clampedCoord, mirror, glTexCoord;
+	
+	clampedCoord = mod(texCoord,texSize);						// clamp coordinates to within texture size
+	mirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);	// whether this texel needs to be mirrored
+
+	glTexCoord = (	mirror*(texSize-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+	return glTexCoord;
+}
+
+/*
+ * main():
+ *
+ * Fragment shader entry point.
+ */
+
+void main(void)
+{	
+	vec4	uv_top, uv_bot, c[4];
+	vec2	r;
+	vec4	fragColor;
+	vec2	ellipse;
+	vec3	lightIntensity;
+	float	insideSpot;
+	int		x;
+	
+	// Get polygon color for untextured polygons (textured polygons will overwrite)
+	if (fsTexParams.x < 0.5)
+		fragColor = gl_Color;		
+	else
+	// Textured polygons: set fragment color to texel value
+	{			
+		/*
+		 * Bilinear Filtering
+		 *
+		 * In order to get this working on ATI, the number of operations is
+		 * reduced by putting everything into vec4s. uv_top holds the UV 
+		 * coordinates for the top two texels (.xy=left, .zw=right) and uv_bot
+		 * is for the lower two.
+		 */
+
+		// Compute fractional blending factor, r, and lower left corner of texel 0
+		uv_bot.xy = gl_TexCoord[0].st-vec2(0.5,0.5);	// move into the lower left blending texel 
+		r = uv_bot.xy-floor(uv_bot.xy);					// fractional part
+		uv_bot.xy = floor(uv_bot.xy);					// integral part
+		
+		// Compute texel coordinates
+		uv_bot.xy += vec2(0.5,0.5);	// offset to center of pixel (should not be needed but it fixes a lot of glitches, esp. on Nvidia)
+		uv_bot.zw = uv_bot.xy + vec2(1.0,0.0);			// compute coordinates of the other three neighbors
+		uv_top = uv_bot + vec4(0.0,1.0,0.0,1.0);
+
+		// Compute the properly wrapped texel coordinates
+		uv_top = WrapTexelCoords(uv_top,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+		uv_bot = WrapTexelCoords(uv_bot,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+
+		// Fetch the texels
+		c[0]=texture2D(textureMap,uv_bot.xy);	// bottom-left (base texel)
+		c[1]=texture2D(textureMap,uv_bot.zw);	// bottom-right
+		c[2]=texture2D(textureMap,uv_top.xy);	// top-left
+		c[3]=texture2D(textureMap,uv_top.zw);	// top-right		
+
+		// Interpolate texels and blend result with material color to determine final (unlit) fragment color
+		// fragColor = (c[0]*(1.0-r.s)*(1.0-r.t) + c[1]*r.s*(1.0-r.t) + c[2]*(1.0-r.s)*r.t + c[3]*r.s*r.t);
+		// Faster method:
+		c[0] += (c[1]-c[0])*r.s;			// 2 alu
+		c[2] += (c[3]-c[2])*r.s;			// 2 alu
+		fragColor = c[0]+(c[2]-c[0])*r.t;	//2 alu
+	
+		/*
+		 * T1RGB5:
+		 *
+		 * The transparency bit determines whether to discard pixels (if set).
+		 * What is unknown is how this bit behaves when interpolated. OpenGL
+		 * processes it as an alpha value, so it might concievably be blended
+		 * with neighbors. Here, an arbitrary threshold is chosen.
+		 *
+		 * To-do: blending could probably enabled and this would work even
+		 * better with a hard threshold.
+		 *
+		 * Countour processing also seems to be enabled for RGBA4 textures.
+		 * When the alpha value is 0.0 (or close), pixels are discarded 
+		 * entirely.
+		 */
+		if (fsTexParams.y > 0.5)	// contour processing enabled
+		{
+			if (fragColor.a < 0.01)	// discard anything with alpha == 0
+				discard;
+		}
+		
+		// If contour texture and not discarded, force alpha to 1.0 because will later be modified by polygon translucency
+		if (fsTexFormat < 0.5)		// contour (T1RGB5) texture
+			fragColor.a = 1.0;
+	}
+
+	// Compute spotlight and apply lighting
+	ellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;
+	insideSpot = dot(ellipse,ellipse);
+	if ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))
+		lightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;
+	else
+		lightIntensity = fsLightIntensity;
+	fragColor.rgb *= lightIntensity;
+	fragColor.rgb += vec3(fsSpecularTerm,fsSpecularTerm,fsSpecularTerm);
+	
+	// Translucency (modulates existing alpha channel for RGBA4 texels)
+	fragColor.a *= fsTransLevel;
+
+	// Apply fog under the control of fog factor setting from polygon header
+	fragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor);
+
+	// Store final color
+	gl_FragColor = fragColor;
+}

Src/Graphics/Legacy3D/Shaders/Fragment2D.glsl

@@ -0,0 +1,42 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Fragment2D.glsl
+ *
+ * Fragment shader for 2D tilemap rendering.
+ */
+
+#version 120
+
+// Global uniforms
+uniform sampler2D	textureMap;		// 512x512 layer surface
+
+/*
+ * main():
+ *
+ * Fragment shader entry point.
+ */
+
+void main(void)
+{	
+	gl_FragColor = texture2D(textureMap, gl_TexCoord[0].st);
+}

Src/Graphics/Legacy3D/Shaders/Fragment_Flat.glsl

@@ -0,0 +1,142 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Fragment_NoSpotlight.glsl
+ *
+ * Fragment shader for 3D rendering. Spotlight effect removed. Fixes fragment
+ * shader link errors on older ATI Radeon GPUs.
+ *
+ * To load external fragment shaders, use the -frag-shader=<file> option when
+ * starting Supermodel.
+ */
+
+#version 120
+
+// Global uniforms
+uniform sampler2D	textureMap;		// complete texture map, 2048x2048 texels
+uniform vec4	spotEllipse;		// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)
+uniform vec2	spotRange;			// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+uniform vec3	spotColor;			// spotlight RGB color
+uniform float		mapSize;		// texture map size (2048,4096,6144 etc)
+
+// Inputs from vertex shader 
+varying vec4		fsSubTexture;	// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)
+varying vec4		fsTexParams;	// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode
+varying float		fsTexFormat;	// .x=T1RGB5 contour texture (if > 0)
+varying float		fsTransLevel;	// translucence level, 0.0 (transparent) to 1.0 (opaque)
+varying vec3		fsLightIntensity;	// lighting intensity 
+varying float		fsFogFactor;		// fog factor
+varying float		fsViewZ;		// Z distance to fragment from viewpoint at origin
+
+/*
+ * WrapTexelCoords():
+ *
+ * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture
+ * sheet, taking into account wrapping behavior.
+ *
+ * Computing normalized OpenGL texture coordinates (0 to 1) within the 
+ * Real3D texture sheet:
+ *
+ * If the texture is not mirrored, we simply have to clamp the
+ * coordinates to fit within the texture dimensions, add the texture
+ * X, Y position to select the appropriate one, and normalize by 2048
+ * (the dimensions of the Real3D texture sheet).
+ *
+ *		= [(u,v)%(w,h)+(x,y)]/(2048,2048)
+ *
+ * If mirroring is enabled, textures are mirrored every odd multiple of
+ * the original texture. To detect whether we are in an odd multiple, 
+ * simply divide the coordinate by the texture dimension and check 
+ * whether the result is odd. Then, clamp the coordinates as before but
+ * subtract from the last texel to mirror them:
+ *
+ * 		= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)
+ *		where M is 1.0 if the texture must be mirrored.
+ *
+ * As an optimization, this function computes TWO texture coordinates
+ * simultaneously. The first is texCoord.xy, the second is in .zw. The other
+ * parameters must have .xy = .zw.
+ */
+vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)
+{
+	vec4	clampedCoord, mirror, glTexCoord;
+	
+	clampedCoord = mod(texCoord,texSize);						// clamp coordinates to within texture size
+	mirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);	// whether this texel needs to be mirrored
+
+	glTexCoord = (	mirror*(texSize-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+/*
+	glTexCoord = (	mirror*(texSize-vec4(1.0,1.0,1.0,1.0)-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+*/
+	return glTexCoord;
+}
+
+/*
+ * main():
+ *
+ * Fragment shader entry point.
+ */
+
+void main(void)
+{	
+	vec4	uv_top, uv_bot, c[4];
+	vec2	r;
+	vec4	fragColor;
+	vec2	ellipse;
+	vec3	lightIntensity;
+	float	insideSpot;
+	
+	// Get polygon color for untextured polygons (textured polygons will overwrite)
+	if (fsTexParams.x < 0.5)
+		fragColor = gl_Color;
+	else
+	// Textured polygons: set fragment color to texel value
+	{			
+		fragColor = texture2D(textureMap,(fsSubTexture.xy+fsSubTexture.zw/2.0)/mapSize);
+		//fragColor += texture2D(textureMap,(fsSubTexture.xy+fsSubTexture.zw)/mapSize);
+	
+	}
+
+	// Compute spotlight and apply lighting
+	ellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;
+	insideSpot = dot(ellipse,ellipse);
+	if ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))
+		lightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;
+	else
+		lightIntensity = fsLightIntensity;
+	fragColor.rgb *= lightIntensity;
+		
+	// Translucency (modulates existing alpha channel for RGBA4 texels)
+	fragColor.a *= fsTransLevel;
+
+	// Apply fog under the control of fog factor setting from polygon header
+	fragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor );
+
+	// Store final color
+	gl_FragColor = fragColor;
+}

Src/Graphics/Legacy3D/Shaders/Fragment_MultiSheet.glsl

@@ -0,0 +1,239 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Fragment_MultiSheet.glsl
+ *
+ * Fragment shader for 3D rendering. Uses 8 texture sheets to decode the 
+ * different possible formats.
+ */
+
+#version 120
+
+// Global uniforms
+uniform sampler2D	textureMap0;		// complete texture map (fmt 0), 2048x2048 texels
+uniform sampler2D	textureMap1;		// complete texture map (fmt 1), 2048x2048 texels
+uniform sampler2D	textureMap2;		// complete texture map (fmt 2), 2048x2048 texels
+uniform sampler2D	textureMap3;		// complete texture map (fmt 3), 2048x2048 texels
+uniform sampler2D	textureMap4;		// complete texture map (fmt 4), 2048x2048 texels
+uniform sampler2D	textureMap5;		// complete texture map (fmt 5), 2048x2048 texels
+uniform sampler2D	textureMap6;		// complete texture map (fmt 6), 2048x2048 texels
+uniform sampler2D	textureMap7;		// complete texture map (fmt 7), 2048x2048 texels
+uniform vec4		spotEllipse;		// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)
+uniform vec2		spotRange;			// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+uniform vec3		spotColor;			// spotlight RGB color
+uniform vec3		lighting[2];		// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)
+uniform float		mapSize;			// texture map size (2048,4096,6144 etc)
+
+// Inputs from vertex shader 
+varying vec4		fsSubTexture;	// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)
+varying vec4		fsTexParams;	// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode
+varying float		fsTexFormat;	// T1RGB5 contour texture (if > 0)
+varying float		fsTexMap;		// texture map number
+varying float		fsTransLevel;	// translucence level, 0.0 (transparent) to 1.0 (opaque)
+varying vec3		fsLightIntensity;	// lighting intensity 
+varying float		fsSpecularTerm;	// specular highlight
+varying float		fsFogFactor;	// fog factor
+varying float		fsViewZ;		// Z distance to fragment from viewpoint at origin
+
+/*
+ * WrapTexelCoords():
+ *
+ * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture
+ * sheet, taking into account wrapping behavior.
+ *
+ * Computing normalized OpenGL texture coordinates (0 to 1) within the 
+ * Real3D texture sheet:
+ *
+ * If the texture is not mirrored, we simply have to clamp the
+ * coordinates to fit within the texture dimensions, add the texture
+ * X, Y position to select the appropriate one, and normalize by 2048
+ * (the dimensions of the Real3D texture sheet).
+ *
+ *		= [(u,v)%(w,h)+(x,y)]/(2048,2048)
+ *
+ * If mirroring is enabled, textures are mirrored every odd multiple of
+ * the original texture. To detect whether we are in an odd multiple, 
+ * simply divide the coordinate by the texture dimension and check 
+ * whether the result is odd. Then, clamp the coordinates as before but
+ * subtract from the last texel to mirror them:
+ *
+ * 		= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)
+ *		where M is 1.0 if the texture must be mirrored.
+ *
+ * As an optimization, this function computes TWO texture coordinates
+ * simultaneously. The first is texCoord.xy, the second is in .zw. The other
+ * parameters must have .xy = .zw.
+ */
+vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)
+{
+	vec4	clampedCoord, mirror, glTexCoord;
+	
+	clampedCoord = mod(texCoord,texSize);						// clamp coordinates to within texture size
+	mirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);	// whether this texel needs to be mirrored
+
+	glTexCoord = (	mirror*(texSize-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+	return glTexCoord;
+}
+
+/*
+ * main():
+ *
+ * Fragment shader entry point.
+ */
+
+void main(void)
+{	
+	vec4	uv_top, uv_bot, c[4];
+	vec2	r;
+	vec4	fragColor;
+	vec2	ellipse;
+	vec3	lightIntensity;
+	float	insideSpot;
+	int		x;
+	
+	// Get polygon color for untextured polygons (textured polygons will overwrite)
+	if (fsTexParams.x < 0.5)
+		fragColor = gl_Color;		
+	else
+	// Textured polygons: set fragment color to texel value
+	{			
+		/*
+		 * Bilinear Filtering
+		 *
+		 * In order to get this working on ATI, the number of operations is
+		 * reduced by putting everything into vec4s. uv_top holds the UV 
+		 * coordinates for the top two texels (.xy=left, .zw=right) and uv_bot
+		 * is for the lower two.
+		 */
+
+		// Compute fractional blending factor, r, and lower left corner of texel 0
+		uv_bot.xy = gl_TexCoord[0].st-vec2(0.5,0.5);	// move into the lower left blending texel 
+		r = uv_bot.xy-floor(uv_bot.xy);					// fractional part
+		uv_bot.xy = floor(uv_bot.xy);					// integral part
+		
+		// Compute texel coordinates
+		uv_bot.xy += vec2(0.5,0.5);	// offset to center of pixel (should not be needed but it fixes a lot of glitches, esp. on Nvidia)
+		uv_bot.zw = uv_bot.xy + vec2(1.0,0.0);			// compute coordinates of the other three neighbors
+		uv_top = uv_bot + vec4(0.0,1.0,0.0,1.0);
+
+		// Compute the properly wrapped texel coordinates
+		uv_top = WrapTexelCoords(uv_top,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+		uv_bot = WrapTexelCoords(uv_bot,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+
+		// Fetch the texels from the given texture map
+		if (fsTexMap < 0.5f)	{
+			c[0]=texture2D(textureMap0, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap0, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap0, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap0, uv_top.zw);       // top-right
+		} else if (fsTexMap < 1.5f) {
+            c[0]=texture2D(textureMap1, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap1, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap1, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap1, uv_top.zw);       // top-right
+		} else if (fsTexMap < 2.5f) {
+            c[0]=texture2D(textureMap2, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap2, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap2, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap2, uv_top.zw);       // top-right
+		} else if (fsTexMap < 3.5f) {
+            c[0]=texture2D(textureMap3, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap3, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap3, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap3, uv_top.zw);       // top-right
+		} else if (fsTexMap < 4.5f) {
+            c[0]=texture2D(textureMap4, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap4, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap4, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap4, uv_top.zw);       // top-right
+		} else if (fsTexMap < 5.5f) {
+            c[0]=texture2D(textureMap5, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap5, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap5, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap5, uv_top.zw);       // top-right
+		} else if (fsTexMap < 6.5f) {
+			c[0]=texture2D(textureMap6, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap6, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap6, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap6, uv_top.zw);       // top-right
+		} else {
+            c[0]=texture2D(textureMap7, uv_bot.xy);       // bottom-left (base texel)
+			c[1]=texture2D(textureMap7, uv_bot.zw);       // bottom-right
+			c[2]=texture2D(textureMap7, uv_top.xy);       // top-left
+			c[3]=texture2D(textureMap7, uv_top.zw);       // top-right
+		}                      
+
+		// Interpolate texels and blend result with material color to determine final (unlit) fragment color
+		// fragColor = (c[0]*(1.0-r.s)*(1.0-r.t) + c[1]*r.s*(1.0-r.t) + c[2]*(1.0-r.s)*r.t + c[3]*r.s*r.t);
+		// Faster method:
+		c[0] += (c[1]-c[0])*r.s;			// 2 alu
+		c[2] += (c[3]-c[2])*r.s;			// 2 alu
+		fragColor = c[0]+(c[2]-c[0])*r.t;	// 2 alu
+	
+		/*
+		 * T1RGB5:
+		 *
+		 * The transparency bit determines whether to discard pixels (if set).
+		 * What is unknown is how this bit behaves when interpolated. OpenGL
+		 * processes it as an alpha value, so it might concievably be blended
+		 * with neighbors. Here, an arbitrary threshold is chosen.
+		 *
+		 * To-do: blending could probably enabled and this would work even
+		 * better with a hard threshold.
+		 *
+		 * Countour processing also seems to be enabled for RGBA4 textures.
+		 * When the alpha value is 0.0 (or close), pixels are discarded 
+		 * entirely.
+		 */
+		if (fsTexParams.y > 0.5)	// contour processing enabled
+		{
+			if (fragColor.a < 0.01)	// discard anything with alpha == 0
+				discard;
+		}
+		
+		// If contour texture and not discarded, force alpha to 1.0 because will later be modified by polygon translucency
+		if (fsTexFormat < 0.5)		// contour (T1RGB5) texture map
+			fragColor.a = 1.0;
+	}
+
+	// Compute spotlight and apply lighting
+	ellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;
+	insideSpot = dot(ellipse,ellipse);
+	if ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))
+		lightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;
+	else
+		lightIntensity = fsLightIntensity;
+	fragColor.rgb *= lightIntensity;
+	fragColor.rgb += vec3(fsSpecularTerm,fsSpecularTerm,fsSpecularTerm);
+	
+	// Translucency (modulates existing alpha channel for RGBA4 texels)
+	fragColor.a *= fsTransLevel;
+
+	// Apply fog under the control of fog factor setting from polygon header
+	fragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor);
+
+	// Store final color
+	gl_FragColor = fragColor;
+}

Src/Graphics/Legacy3D/Shaders/Fragment_NoSpotlight.glsl

@@ -0,0 +1,202 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Fragment_NoSpotlight.glsl
+ *
+ * Fragment shader for 3D rendering. Spotlight effect removed. Fixes fragment
+ * shader link errors on older ATI Radeon GPUs.
+ *
+ * To load external fragment shaders, use the -frag-shader=<file> option when
+ * starting Supermodel.
+ */
+
+#version 120
+
+// Global uniforms
+uniform sampler2D	textureMap;		// complete texture map, 2048x2048 texels
+uniform vec4	spotEllipse;		// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)
+uniform vec2	spotRange;			// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+uniform vec3	spotColor;			// spotlight RGB color
+uniform float		mapSize;		// texture map size (2048,4096,6144 etc)
+
+// Inputs from vertex shader 
+varying vec4		fsSubTexture;	// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)
+varying vec4		fsTexParams;	// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode
+varying float		fsTexFormat;	// .x=T1RGB5 contour texture (if > 0)
+varying float		fsTransLevel;	// translucence level, 0.0 (transparent) to 1.0 (opaque)
+varying vec3		fsLightIntensity;	// lighting intensity 
+varying float		fsFogFactor;		// fog factor
+varying float		fsViewZ;		// Z distance to fragment from viewpoint at origin
+
+/*
+ * WrapTexelCoords():
+ *
+ * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture
+ * sheet, taking into account wrapping behavior.
+ *
+ * Computing normalized OpenGL texture coordinates (0 to 1) within the 
+ * Real3D texture sheet:
+ *
+ * If the texture is not mirrored, we simply have to clamp the
+ * coordinates to fit within the texture dimensions, add the texture
+ * X, Y position to select the appropriate one, and normalize by 2048
+ * (the dimensions of the Real3D texture sheet).
+ *
+ *		= [(u,v)%(w,h)+(x,y)]/(2048,2048)
+ *
+ * If mirroring is enabled, textures are mirrored every odd multiple of
+ * the original texture. To detect whether we are in an odd multiple, 
+ * simply divide the coordinate by the texture dimension and check 
+ * whether the result is odd. Then, clamp the coordinates as before but
+ * subtract from the last texel to mirror them:
+ *
+ * 		= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)
+ *		where M is 1.0 if the texture must be mirrored.
+ *
+ * As an optimization, this function computes TWO texture coordinates
+ * simultaneously. The first is texCoord.xy, the second is in .zw. The other
+ * parameters must have .xy = .zw.
+ */
+vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)
+{
+	vec4	clampedCoord, mirror, glTexCoord;
+	
+	clampedCoord = mod(texCoord,texSize);						// clamp coordinates to within texture size
+	mirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);	// whether this texel needs to be mirrored
+
+	glTexCoord = (	mirror*(texSize-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+/*
+	glTexCoord = (	mirror*(texSize-vec4(1.0,1.0,1.0,1.0)-clampedCoord) +
+					(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +
+					texOffset
+				 ) / mapSize;
+*/
+	return glTexCoord;
+}
+
+/*
+ * main():
+ *
+ * Fragment shader entry point.
+ */
+
+void main(void)
+{	
+	vec4	uv_top, uv_bot, c[4];
+	vec2	r;
+	vec4	fragColor;
+	vec2	ellipse;
+	vec3	lightIntensity;
+	float	insideSpot;
+	
+	// Get polygon color for untextured polygons (textured polygons will overwrite)
+	if (fsTexParams.x < 0.5)
+		fragColor = gl_Color;
+	else
+	// Textured polygons: set fragment color to texel value
+	{			
+		/*
+		 * Bilinear Filtering
+		 *
+		 * In order to get this working on ATI, the number of operations is
+		 * reduced by putting everything into vec4s. uv_top holds the UV 
+		 * coordinates for the top two texels (.xy=left, .zw=right) and uv_bot
+		 * is for the lower two.
+		 */
+
+		// Compute fractional blending factor, r, and lower left corner of texel 0
+		uv_bot.xy = gl_TexCoord[0].st-vec2(0.5,0.5);	// move into the lower left blending texel 
+		r = uv_bot.xy-floor(uv_bot.xy);					// fractional part
+		uv_bot.xy = floor(uv_bot.xy);					// integral part
+		
+		// Compute texel coordinates
+		uv_bot.xy += vec2(0.5,0.5);	// offset to center of pixel (should not be needed but it fixes a lot of glitches, esp. on Nvidia)
+		uv_bot.zw = uv_bot.xy + vec2(1.0,0.0);			// compute coordinates of the other three neighbors
+		uv_top = uv_bot + vec4(0.0,1.0,0.0,1.0);
+
+		// Compute the properly wrapped texel coordinates
+		uv_top = WrapTexelCoords(uv_top,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+		uv_bot = WrapTexelCoords(uv_bot,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));
+
+		// Fetch the texels
+		c[0]=texture2D(textureMap,uv_bot.xy);	// bottom-left (base texel)
+		c[1]=texture2D(textureMap,uv_bot.zw);	// bottom-right
+		c[2]=texture2D(textureMap,uv_top.xy);	// top-left
+		c[3]=texture2D(textureMap,uv_top.zw);	// top-right		
+
+		// Interpolate texels and blend result with material color to determine final (unlit) fragment color
+		// fragColor = (c[0]*(1.0-r.s)*(1.0-r.t) + c[1]*r.s*(1.0-r.t) + c[2]*(1.0-r.s)*r.t + c[3]*r.s*r.t);
+		// Faster method:
+		c[0] += (c[1]-c[0])*r.s;			// 2 alu
+		c[2] += (c[3]-c[2])*r.s;			// 2 alu
+		fragColor = c[0]+(c[2]-c[0])*r.t;	//2 alu
+	
+		/*
+		 * T1RGB5:
+		 *
+		 * The transparency bit determines whether to discard pixels (if set).
+		 * What is unknown is how this bit behaves when interpolated. OpenGL
+		 * processes it as an alpha value, so it might concievably be blended
+		 * with neighbors. Here, an arbitrary threshold is chosen.
+		 *
+		 * To-do: blending could probably enabled and this would work even
+		 * better with a hard threshold.
+		 *
+		 * Countour processing also seems to be enabled for RGBA4 textures.
+		 * When the alpha value is 0.0 (or close), pixels are discarded 
+		 * entirely.
+		 */
+		if (fsTexParams.y > 0.5)	// contour processing enabled
+		{
+			if (fragColor.a < 0.01)	// discard anything with alpha == 0
+				discard;
+		}
+		
+		// If contour texture and not discarded, force alpha to 1.0 because will later be modified by polygon translucency
+		if (fsTexFormat < 0.5)		// contour (T1RGB5) texture map
+			fragColor.a = 1.0;
+	}
+
+	// Compute spotlight and apply lighting
+	/***
+	ellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;
+	insideSpot = dot(ellipse,ellipse);
+	if ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))
+		lightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;
+	else
+		lightIntensity = fsLightIntensity;
+	fragColor.rgb *= lightIntensity;
+	***/
+	fragColor.rgb *= fsLightIntensity;
+
+	// Translucency (modulates existing alpha channel for RGBA4 texels)
+	fragColor.a *= fsTransLevel;
+
+	// Apply fog under the control of fog factor setting from polygon header
+	fragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor );
+
+	// Store final color
+	gl_FragColor = fragColor;
+}

Src/Graphics/Legacy3D/Shaders/Vertex.glsl

@@ -0,0 +1,181 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+  
+/*
+ * Vertex.glsl
+ *
+ * Vertex shader for 3D rendering.
+ */
+ 
+#version 120
+
+// Global uniforms
+uniform mat4	modelViewMatrix;	// model -> view space matrix
+uniform mat4	projectionMatrix;	// view space -> screen space matrix
+uniform vec3	lighting[2];		// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)
+uniform vec4	spotEllipse;		// spotlight ellipse position: .x=X position (normalized device coordinates), .y=Y position, .z=half-width, .w=half-height)
+uniform vec2	spotRange;			// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+uniform vec3	spotColor;			// spotlight RGB color
+
+// Custom vertex attributes
+attribute vec4	subTexture;		// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)
+attribute vec4	texParams;		// .x=texture enable (if 1, else 0), .y=use transparency (if >=0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode
+attribute float	texFormat;		// T1RGB5 contour texture (if > 0)
+attribute float texMap;         // texture map number
+attribute float	transLevel;		// translucence level, 0.0 (transparent) to 1.0 (opaque). if less than 1.0, replace alpha value
+attribute float	lightEnable;	// lighting enabled (1.0) or luminous (0.0), drawn at full intensity
+attribute float	shininess;		// specular shininess (if >= 0.0) or disable specular lighting (negative)
+attribute float	fogIntensity;	// fog intensity (1.0, full fog effect, 0.0, no fog) 
+
+// Custom outputs to fragment shader
+varying vec4	fsSubTexture;
+varying vec4	fsTexParams;
+varying float	fsTexFormat;
+varying float	fsTexMap;
+varying float	fsTransLevel;
+varying vec3	fsLightIntensity;	// total light intensity for this vertex
+varying float	fsSpecularTerm;		// specular light term (additive)
+varying float	fsFogFactor;		// fog factor
+varying float	fsViewZ;
+
+// Gets the 3x3 matrix out of a 4x4 (because mat3(mat4matrix) does not work on ATI!)
+mat3 GetLinearPart( mat4 m )
+{
+	mat3 result;
+	
+	result[0][0] = m[0][0]; 
+	result[0][1] = m[0][1]; 
+	result[0][2] = m[0][2]; 
+
+	result[1][0] = m[1][0]; 
+	result[1][1] = m[1][1]; 
+	result[1][2] = m[1][2]; 
+	
+	result[2][0] = m[2][0]; 
+	result[2][1] = m[2][1]; 
+	result[2][2] = m[2][2]; 
+	
+	return result;
+}
+
+void main(void)
+{
+	vec3	viewVertex;		// vertex coordinates in view space
+	vec3	viewNormal;		// vertex normal in view space
+	vec3	sunVector;		// sun lighting vector (as reflecting away from vertex)
+	float	sunFactor;		// sun light projection along vertex normal (0.0 to 1.0)
+	vec3	halfway;
+	float	specFactor;
+	
+	// Transform vertex
+	gl_Position = projectionMatrix * modelViewMatrix * gl_Vertex;
+	viewVertex = vec3(modelViewMatrix * gl_Vertex);	
+	
+	/*
+	 * Modulation
+	 *
+ 	 * Polygon color serves as material color (modulating the light intensity)
+	 * for textured polygons. The fragment shader will ignore (overwrite) the
+	 * the color passed to it if the fragment is textured. 
+	 *
+	 * Untextured fragments must be set to the polygon color and the light
+	 * intensity is initialized to 1.0 here. Alpha must be set to 1.0 because
+	 * the fragment shader multiplies it by the polygon translucency setting. 
+	 *
+	 * TO-DO: Does OpenGL set alpha to 1.0 by default if no alpha is specified
+	 * for the vertex? If so, we can remove that line from here.
+	 */
+
+	gl_FrontColor = gl_Color;	// untextured polygons will use this
+	gl_FrontColor.a = 1.0;	
+	fsLightIntensity = vec3(1.0,1.0,1.0);
+	if (texParams.x > 0.5)		// textured
+		fsLightIntensity *= gl_Color.rgb;
+		
+	/*
+ 	 * Sun Light
+	 *
+	 * Parallel light source and ambient lighting are only applied for non-
+	 * luminous polygons.
+ 	 */
+	fsSpecularTerm = 0.0;
+ 	if (lightEnable > 0.5)	// not luminous
+	{
+		// Normal -> view space
+		viewNormal = normalize(GetLinearPart(modelViewMatrix)*gl_Normal);
+
+		// Real3D -> OpenGL view space convention (TO-DO: do this outside of shader)
+		sunVector = lighting[0]*vec3(1.0,-1.0,-1.0);
+		
+		// Compute diffuse factor for sunlight
+		sunFactor = max(dot(sunVector,viewNormal),0.0);
+		
+		// Total light intensity: sum of all components
+		fsLightIntensity *= (sunFactor*lighting[1].x+lighting[1].y);
+		
+		/*
+		 * Specular Lighting
+		 *
+		 * The specular term is treated similarly to the "separate specular
+		 * color" functionality of OpenGL: it is added as a highlight in the
+		 * fragment shader. This allows even black textures to be lit.
+		 *
+		 * TO-DO: Ambient intensity viewport parameter is known but what about
+		 * the intensity of the specular term? Always applied with full 
+		 * intensity here but this is unlikely to be correct.
+		 */
+  		if (shininess >= 0.0)
+  		{
+  			// Standard specular lighting equation
+  			vec3 V = normalize(-viewVertex);
+  			vec3 H = normalize(sunVector+V);	// halfway vector
+  			float s = max(10.0,64.0-shininess);	// seems to look nice, but probably not correct
+  			fsSpecularTerm = pow(max(dot(viewNormal,H),0.0),s);
+  			if (sunFactor <= 0.0) fsSpecularTerm = 0.0;
+  			
+  			// Faster approximation  			
+  			//float temp = max(dot(viewNormal,H),0.0);
+  			//float s = 64.0-shininess;
+  			//fsSpecularTerm = temp/(s-temp*s+temp);
+  			
+  			// Phong formula
+  			//vec3 R = normalize(2.0*dot(sunVector,viewNormal)*viewNormal - sunVector);
+  			//vec3 V = normalize(-viewVertex);
+  			//float s = max(2.0,64.0-shininess);
+  			//fsSpecularTerm = pow(max(dot(R,V),0.0),s);
+  		}
+	}
+	
+	// Fog
+	float z = length(viewVertex);
+	fsFogFactor = clamp(1.0-fogIntensity*(gl_Fog.start+z*gl_Fog.density), 0.0, 1.0);
+
+	// Pass viewspace Z coordinate (for spotlight)
+	fsViewZ = -viewVertex.z;	// convert Z from GL->Real3D convention (want +Z to be further into screen)
+
+	// Pass remaining parameters to fragment shader
+	gl_TexCoord[0] = gl_MultiTexCoord0;
+	fsSubTexture = subTexture;
+	fsTexParams = texParams;
+	fsTransLevel = transLevel;
+	fsTexFormat = texFormat;
+	fsTexMap = texMap;
+}

Src/Graphics/Legacy3D/Shaders/Vertex2D.glsl

@@ -0,0 +1,34 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+
+/*
+ * Vertex2D.glsl
+ *
+ * Vertex shader for 2D tilemap rendering.
+ */
+ 
+#version 120
+
+void main(void)
+{
+	gl_TexCoord[0] = gl_MultiTexCoord0;
+	gl_Position = gl_ModelViewProjectionMatrix*gl_Vertex;
+}

Src/Graphics/Legacy3D/Shaders3D.h

@@ -0,0 +1,666 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * Shaders3D.h
+ * 
+ * Header file containing the 3D vertex and fragment shaders.
+ */
+
+#ifndef INCLUDED_SHADERS3D_H
+#define INCLUDED_SHADERS3D_H
+
+namespace Legacy3D {
+
+// Vertex shader
+static const char vertexShaderSource[] =
+{
+"/**\n"
+" ** Supermodel\n"
+" ** A Sega Model 3 Arcade Emulator.\n"
+" ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson \n"
+" **\n"
+" ** This file is part of Supermodel.\n"
+" **\n"
+" ** Supermodel is free software: you can redistribute it and/or modify it under\n"
+" ** the terms of the GNU General Public License as published by the Free \n"
+" ** Software Foundation, either version 3 of the License, or (at your option)\n"
+" ** any later version.\n"
+" **\n"
+" ** Supermodel is distributed in the hope that it will be useful, but WITHOUT\n"
+" ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or\n"
+" ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for\n"
+" ** more details.\n"
+" **\n"
+" ** You should have received a copy of the GNU General Public License along\n"
+" ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.\n"
+" **/\n"
+"  \n"
+"/*\n"
+" * Vertex.glsl\n"
+" *\n"
+" * Vertex shader for 3D rendering.\n"
+" */\n"
+" \n"
+"#version 120\n"
+"\n"
+"// Global uniforms\n"
+"uniform mat4\tmodelViewMatrix;\t// model -> view space matrix\n"
+"uniform mat4\tprojectionMatrix;\t// view space -> screen space matrix\n"
+"uniform vec3\tlighting[2];\t\t// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)\n"
+"uniform vec4\tspotEllipse;\t\t// spotlight ellipse position: .x=X position (normalized device coordinates), .y=Y position, .z=half-width, .w=half-height)\n"
+"uniform vec2\tspotRange;\t\t\t// spotlight Z range: .x=start (viewspace coordinates), .y=limit\n"
+"uniform vec3\tspotColor;\t\t\t// spotlight RGB color\n"
+"\n"
+"// Custom vertex attributes\n"
+"attribute vec4\tsubTexture;\t\t// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)\n"
+"attribute vec4\ttexParams;\t\t// .x=texture enable (if 1, else 0), .y=use transparency (if >=0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode\n"
+"attribute float\ttexFormat;\t\t// T1RGB5 contour texture (if > 0)\n"
+"attribute float\ttexMap;\t\t// texture map number\n"
+"attribute float\ttransLevel;\t\t// translucence level, 0.0 (transparent) to 1.0 (opaque). if less than 1.0, replace alpha value\n"
+"attribute float\tlightEnable;\t// lighting enabled (1.0) or luminous (0.0), drawn at full intensity\n"
+"attribute float\tshininess;\t\t// specular shininess (if >= 0.0) or disable specular lighting (negative)\n"
+"attribute float\tfogIntensity;\t// fog intensity (1.0, full fog effect, 0.0, no fog) \n"
+"\n"
+"// Custom outputs to fragment shader\n"
+"varying vec4\tfsSubTexture;\n"
+"varying vec4\tfsTexParams;\n"
+"varying float\tfsTexFormat;\n"
+"varying float\tfsTexMap;\n"
+"varying float\tfsTransLevel;\n"
+"varying vec3\tfsLightIntensity;\t// total light intensity for this vertex\n"
+"varying float\tfsSpecularTerm;\t\t// specular light term (additive)\n"
+"varying float\tfsFogFactor;\t\t// fog factor\n"
+"varying float\tfsViewZ;\n"
+"\n"
+"// Gets the 3x3 matrix out of a 4x4 (because mat3(mat4matrix) does not work on ATI!)\n"
+"mat3 GetLinearPart( mat4 m )\n"
+"{\n"
+"\tmat3 result;\n"
+"\t\n"
+"\tresult[0][0] = m[0][0]; \n"
+"\tresult[0][1] = m[0][1]; \n"
+"\tresult[0][2] = m[0][2]; \n"
+"\n"
+"\tresult[1][0] = m[1][0]; \n"
+"\tresult[1][1] = m[1][1]; \n"
+"\tresult[1][2] = m[1][2]; \n"
+"\t\n"
+"\tresult[2][0] = m[2][0]; \n"
+"\tresult[2][1] = m[2][1]; \n"
+"\tresult[2][2] = m[2][2]; \n"
+"\t\n"
+"\treturn result;\n"
+"}\n"
+"\n"
+"void main(void)\n"
+"{\n"
+"\tvec3\tviewVertex;\t\t// vertex coordinates in view space\n"
+"\tvec3\tviewNormal;\t\t// vertex normal in view space\n"
+"\tvec3\tsunVector;\t\t// sun lighting vector (as reflecting away from vertex)\n"
+"\tfloat\tsunFactor;\t\t// sun light projection along vertex normal (0.0 to 1.0)\n"
+"\tvec3\thalfway;\n"
+"\tfloat\tspecFactor;\n"
+"\t\n"
+"\t// Transform vertex\n"
+"\tgl_Position = projectionMatrix * modelViewMatrix * gl_Vertex;\n"
+"\tviewVertex = vec3(modelViewMatrix * gl_Vertex);\t\n"
+"\t\n"
+"\t/*\n"
+"\t * Modulation\n"
+"\t *\n"
+" \t * Polygon color serves as material color (modulating the light intensity)\n"
+"\t * for textured polygons. The fragment shader will ignore (overwrite) the\n"
+"\t * the color passed to it if the fragment is textured. \n"
+"\t *\n"
+"\t * Untextured fragments must be set to the polygon color and the light\n"
+"\t * intensity is initialized to 1.0 here. Alpha must be set to 1.0 because\n"
+"\t * the fragment shader multiplies it by the polygon translucency setting. \n"
+"\t *\n"
+"\t * TO-DO: Does OpenGL set alpha to 1.0 by default if no alpha is specified\n"
+"\t * for the vertex? If so, we can remove that line from here.\n"
+"\t */\n"
+"\n"
+"\tgl_FrontColor = gl_Color;\t// untextured polygons will use this\n"
+"\tgl_FrontColor.a = 1.0;\t\n"
+"\tfsLightIntensity = vec3(1.0,1.0,1.0);\n"
+"\tif (texParams.x > 0.5)\t\t// textured\n"
+"\t\tfsLightIntensity *= gl_Color.rgb;\n"
+"\t\t\n"
+"\t/*\n"
+" \t * Sun Light\n"
+"\t *\n"
+"\t * Parallel light source and ambient lighting are only applied for non-\n"
+"\t * luminous polygons.\n"
+" \t */\n"
+"\tfsSpecularTerm = 0.0;\n"
+" \tif (lightEnable > 0.5)\t// not luminous\n"
+"\t{\n"
+"\t\t// Normal -> view space\n"
+"\t\tviewNormal = normalize(GetLinearPart(modelViewMatrix)*gl_Normal);\n"
+"\n"
+"\t\t// Real3D -> OpenGL view space convention (TO-DO: do this outside of shader)\n"
+"\t\tsunVector = lighting[0]*vec3(1.0,-1.0,-1.0);\n"
+"\t\t\n"
+"\t\t// Compute diffuse factor for sunlight\n"
+"\t\tsunFactor = max(dot(sunVector,viewNormal),0.0);\n"
+"\t\t\n"
+"\t\t// Total light intensity: sum of all components\n"
+"\t\tfsLightIntensity *= (sunFactor*lighting[1].x+lighting[1].y);\n"
+"\t\t\n"
+"\t\t/*\n"
+"\t\t * Specular Lighting\n"
+"\t\t *\n"
+"\t\t * The specular term is treated similarly to the \"separate specular\n"
+"\t\t * color\" functionality of OpenGL: it is added as a highlight in the\n"
+"\t\t * fragment shader. This allows even black textures to be lit.\n"
+"\t\t *\n"
+"\t\t * TO-DO: Ambient intensity viewport parameter is known but what about\n"
+"\t\t * the intensity of the specular term? Always applied with full \n"
+"\t\t * intensity here but this is unlikely to be correct.\n"
+"\t\t */\n"
+"  \t\tif (shininess >= 0.0)\n"
+"  \t\t{\n"
+"  \t\t\t// Standard specular lighting equation\n"
+"  \t\t\tvec3 V = normalize(-viewVertex);\n"
+"  \t\t\tvec3 H = normalize(sunVector+V);\t// halfway vector\n"
+"  \t\t\tfloat s = max(10.0,64.0-shininess);\t\t// seems to look nice, but probably not correct\n"
+"  \t\t\tfsSpecularTerm = pow(max(dot(viewNormal,H),0.0),s);\n"
+"  \t\t\tif (sunFactor <= 0.0) fsSpecularTerm = 0.0;\n"
+"  \t\t\t\n"
+"  \t\t\t// Faster approximation  \t\t\t\n"
+"  \t\t\t//float temp = max(dot(viewNormal,H),0.0);\n"
+"  \t\t\t//float s = 64.0-shininess;\n"
+"  \t\t\t//fsSpecularTerm = temp/(s-temp*s+temp);\n"
+"  \t\t\t\n"
+"  \t\t\t// Phong formula\n"
+"  \t\t\t//vec3 R = normalize(2.0*dot(sunVector,viewNormal)*viewNormal - sunVector);\n"
+"  \t\t\t//vec3 V = normalize(-viewVertex);\n"
+"  \t\t\t//float s = max(2.0,64.0-shininess);\n"
+"  \t\t\t//fsSpecularTerm = pow(max(dot(R,V),0.0),s);\n"
+"  \t\t}\n"
+"\t}\n"
+"\t\n"
+"\t// Fog\n"
+"\tfloat z = length(viewVertex);\n"
+"\tfsFogFactor = clamp(1.0-fogIntensity*(gl_Fog.start+z*gl_Fog.density), 0.0, 1.0);\n"
+"\n"
+"\t// Pass viewspace Z coordinate (for spotlight)\n"
+"\tfsViewZ = -viewVertex.z;\t// convert Z from GL->Real3D convention (want +Z to be further into screen)\n"
+"\n"
+"\t// Pass remaining parameters to fragment shader\n"
+"\tgl_TexCoord[0] = gl_MultiTexCoord0;\n"
+"\tfsSubTexture = subTexture;\n"
+"\tfsTexParams = texParams;\n"
+"\tfsTransLevel = transLevel;\n"
+"\tfsTexFormat = texFormat;\n"
+"\tfsTexMap = texMap;\n"
+"}\n"
+};
+	
+	
+// Fragment shader (single texture sheet)
+static const char fragmentShaderSingleSheetSource[] = 
+{
+"/**\n"
+" ** Supermodel\n"
+" ** A Sega Model 3 Arcade Emulator.\n"
+" ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson \n"
+" **\n"
+" ** This file is part of Supermodel.\n"
+" **\n"
+" ** Supermodel is free software: you can redistribute it and/or modify it under\n"
+" ** the terms of the GNU General Public License as published by the Free \n"
+" ** Software Foundation, either version 3 of the License, or (at your option)\n"
+" ** any later version.\n"
+" **\n"
+" ** Supermodel is distributed in the hope that it will be useful, but WITHOUT\n"
+" ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or\n"
+" ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for\n"
+" ** more details.\n"
+" **\n"
+" ** You should have received a copy of the GNU General Public License along\n"
+" ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.\n"
+" **/\n"
+" \n"
+"/*\n"
+" * Fragment.glsl\n"
+" *\n"
+" * Fragment shader for 3D rendering.\n"
+" */\n"
+"\n"
+"#version 120\n"
+"\n"
+"// Global uniforms\n"
+"uniform sampler2D\ttextureMap;\t\t// complete texture map, 2048x2048 texels\n"
+"uniform vec4\t\tspotEllipse;\t\t// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)\n"
+"uniform vec2\t\tspotRange;\t\t// spotlight Z range: .x=start (viewspace coordinates), .y=limit\n"
+"uniform vec3\t\tspotColor;\t\t// spotlight RGB color\n"
+"uniform vec3\t\tlighting[2];\t\t// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)\n"
+"uniform float\t\tmapSize;\t\t// texture map size (2048,4096,6144 etc)\n"
+"\n"
+"// Inputs from vertex shader \n"
+"varying vec4\t\tfsSubTexture;\t// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)\n"
+"varying vec4\t\tfsTexParams;\t// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode\n"
+"varying float\t\tfsTexFormat;\t// T1RGB5 contour texture (if > 0)\n"
+"varying float\t\tfsTexMap;\t\t// texture map number\n"
+"varying float\t\tfsTransLevel;\t// translucence level, 0.0 (transparent) to 1.0 (opaque)\n"
+"varying vec3\t\tfsLightIntensity;\t// lighting intensity \n"
+"varying float\t\tfsSpecularTerm;\t// specular highlight\n"
+"varying float\t\tfsFogFactor;\t// fog factor\n"
+"varying float\t\tfsViewZ;\t\t// Z distance to fragment from viewpoint at origin\n"
+"\n"
+"/*\n"
+" * WrapTexelCoords():\n"
+" *\n"
+" * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture\n"
+" * sheet, taking into account wrapping behavior.\n"
+" *\n"
+" * Computing normalized OpenGL texture coordinates (0 to 1) within the \n"
+" * Real3D texture sheet:\n"
+" *\n"
+" * If the texture is not mirrored, we simply have to clamp the\n"
+" * coordinates to fit within the texture dimensions, add the texture\n"
+" * X, Y position to select the appropriate one, and normalize by 2048\n"
+" * (the dimensions of the Real3D texture sheet).\n"
+" *\n"
+" *\t\t= [(u,v)%(w,h)+(x,y)]/(2048,2048)\n"
+" *\n"
+" * If mirroring is enabled, textures are mirrored every odd multiple of\n"
+" * the original texture. To detect whether we are in an odd multiple, \n"
+" * simply divide the coordinate by the texture dimension and check \n"
+" * whether the result is odd. Then, clamp the coordinates as before but\n"
+" * subtract from the last texel to mirror them:\n"
+" *\n"
+" * \t\t= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)\n"
+" *\t\twhere M is 1.0 if the texture must be mirrored.\n"
+" *\n"
+" * As an optimization, this function computes TWO texture coordinates\n"
+" * simultaneously. The first is texCoord.xy, the second is in .zw. The other\n"
+" * parameters must have .xy = .zw.\n"
+" */\n"
+"vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)\n"
+"{\n"
+"\tvec4\tclampedCoord, mirror, glTexCoord;\n"
+"\t\n"
+"\tclampedCoord = mod(texCoord,texSize);\t\t\t\t\t\t// clamp coordinates to within texture size\n"
+"\tmirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);\t// whether this texel needs to be mirrored\n"
+"\n"
+"\tglTexCoord = (\tmirror*(texSize-clampedCoord) +\n"
+"\t\t\t\t\t(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +\n"
+"\t\t\t\t\ttexOffset\n"
+"\t\t\t\t ) / mapSize;\n"
+"\treturn glTexCoord;\n"
+"}\n"
+"\n"
+"/*\n"
+" * main():\n"
+" *\n"
+" * Fragment shader entry point.\n"
+" */\n"
+"\n"
+"void main(void)\n"
+"{\t\n"
+"\tvec4\tuv_top, uv_bot, c[4];\n"
+"\tvec2\tr;\n"
+"\tvec4\tfragColor;\n"
+"\tvec2\tellipse;\n"
+"\tvec3\tlightIntensity;\n"
+"\tfloat\tinsideSpot;\n"
+"\tint\t\tx;\n"
+"\t\n"
+"\t// Get polygon color for untextured polygons (textured polygons will overwrite)\n"
+"\tif (fsTexParams.x < 0.5)\n"
+"\t\tfragColor = gl_Color;\t\t\n"
+"\telse\n"
+"\t// Textured polygons: set fragment color to texel value\n"
+"\t{\t\t\t\n"
+"\t\t/*\n"
+"\t\t * Bilinear Filtering\n"
+"\t\t *\n"
+"\t\t * In order to get this working on ATI, the number of operations is\n"
+"\t\t * reduced by putting everything into vec4s. uv_top holds the UV \n"
+"\t\t * coordinates for the top two texels (.xy=left, .zw=right) and uv_bot\n"
+"\t\t * is for the lower two.\n"
+"\t\t */\n"
+"\n"
+"\t\t// Compute fractional blending factor, r, and lower left corner of texel 0\n"
+"\t\tuv_bot.xy = gl_TexCoord[0].st-vec2(0.5,0.5);\t// move into the lower left blending texel \n"
+"\t\tr = uv_bot.xy-floor(uv_bot.xy);\t\t\t\t\t// fractional part\n"
+"\t\tuv_bot.xy = floor(uv_bot.xy);\t\t\t\t\t// integral part\n"
+"\t\t\n"
+"\t\t// Compute texel coordinates\n"
+"\t\tuv_bot.xy += vec2(0.5,0.5);\t// offset to center of pixel (should not be needed but it fixes a lot of glitches, esp. on Nvidia)\n"
+"\t\tuv_bot.zw = uv_bot.xy + vec2(1.0,0.0);\t\t\t// compute coordinates of the other three neighbors\n"
+"\t\tuv_top = uv_bot + vec4(0.0,1.0,0.0,1.0);\n"
+"\n"
+"\t\t// Compute the properly wrapped texel coordinates\n"
+"\t\tuv_top = WrapTexelCoords(uv_top,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));\n"
+"\t\tuv_bot = WrapTexelCoords(uv_bot,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));\n"
+"\n"
+"\t\t// Fetch the texels\n"
+"\t\tc[0]=texture2D(textureMap,uv_bot.xy);\t// bottom-left (base texel)\n"
+"\t\tc[1]=texture2D(textureMap,uv_bot.zw);\t// bottom-right\n"
+"\t\tc[2]=texture2D(textureMap,uv_top.xy);\t// top-left\n"
+"\t\tc[3]=texture2D(textureMap,uv_top.zw);\t// top-right\t\t\n"
+"\n"
+"\t\t// Interpolate texels and blend result with material color to determine final (unlit) fragment color\n"
+"\t\t// fragColor = (c[0]*(1.0-r.s)*(1.0-r.t) + c[1]*r.s*(1.0-r.t) + c[2]*(1.0-r.s)*r.t + c[3]*r.s*r.t);\n"
+"\t\t// Faster method:\n"
+"\t\tc[0] += (c[1]-c[0])*r.s;\t\t\t// 2 alu\n"
+"\t\tc[2] += (c[3]-c[2])*r.s;\t\t\t// 2 alu\n"
+"\t\tfragColor = c[0]+(c[2]-c[0])*r.t;\t//2 alu\n"
+"\t\n"
+"\t\t/*\n"
+"\t\t * T1RGB5:\n"
+"\t\t *\n"
+"\t\t * The transparency bit determines whether to discard pixels (if set).\n"
+"\t\t * What is unknown is how this bit behaves when interpolated. OpenGL\n"
+"\t\t * processes it as an alpha value, so it might concievably be blended\n"
+"\t\t * with neighbors. Here, an arbitrary threshold is chosen.\n"
+"\t\t *\n"
+"\t\t * To-do: blending could probably enabled and this would work even\n"
+"\t\t * better with a hard threshold.\n"
+"\t\t *\n"
+"\t\t * Countour processing also seems to be enabled for RGBA4 textures.\n"
+"\t\t * When the alpha value is 0.0 (or close), pixels are discarded \n"
+"\t\t * entirely.\n"
+"\t\t */\n"
+"\t\tif (fsTexParams.y > 0.5)\t// contour processing enabled\n"
+"\t\t{\n"
+"\t\t\tif (fragColor.a < 0.01)\t// discard anything with alpha == 0\n"
+"\t\t\t\tdiscard;\n"
+"\t\t}\n"
+"\t\t\n"
+"\t\t// If contour texture and not discarded, force alpha to 1.0 because will later be modified by polygon translucency\n"
+"\t\tif (fsTexFormat < 0.5)\t\t// contour (T1RGB5) texture\n"
+"\t\t\tfragColor.a = 1.0;\n"
+"\t}\n"
+"\n"
+"\t// Compute spotlight and apply lighting\n"
+"\tellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;\n"
+"\tinsideSpot = dot(ellipse,ellipse);\n"
+"\tif ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))\n"
+"\t\tlightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;\n"
+"\telse\n"
+"\t\tlightIntensity = fsLightIntensity;\n"
+"\tfragColor.rgb *= lightIntensity;\n"
+"\tfragColor.rgb += vec3(fsSpecularTerm,fsSpecularTerm,fsSpecularTerm);\n"
+"\t\n"
+"\t// Translucency (modulates existing alpha channel for RGBA4 texels)\n"
+"\tfragColor.a *= fsTransLevel;\n"
+"\n"
+"\t// Apply fog under the control of fog factor setting from polygon header\n"
+"\tfragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor);\n"
+"\n"
+"\t// Store final color\n"
+"\tgl_FragColor = fragColor;\n"
+"}\n"
+};
+
+
+// Fragment shader (8 texture sheets)
+static const char fragmentShaderMultiSheetSource[] = 
+{
+"/**\n"
+" ** Supermodel\n"
+" ** A Sega Model 3 Arcade Emulator.\n"
+" ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson \n"
+" **\n"
+" ** This file is part of Supermodel.\n"
+" **\n"
+" ** Supermodel is free software: you can redistribute it and/or modify it under\n"
+" ** the terms of the GNU General Public License as published by the Free \n"
+" ** Software Foundation, either version 3 of the License, or (at your option)\n"
+" ** any later version.\n"
+" **\n"
+" ** Supermodel is distributed in the hope that it will be useful, but WITHOUT\n"
+" ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or\n"
+" ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for\n"
+" ** more details.\n"
+" **\n"
+" ** You should have received a copy of the GNU General Public License along\n"
+" ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.\n"
+" **/\n"
+" \n"
+"/*\n"
+" * Fragment_MultiSheet.glsl\n"
+" *\n"
+" * Fragment shader for 3D rendering. Uses 8 texture sheets to decode the \n"
+" * different possible formats.\n"
+" */\n"
+"\n"
+"#version 120\n"
+"\n"
+"// Global uniforms\n"
+"uniform sampler2D\ttextureMap0;\t\t// complete texture map (fmt 0), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap1;\t\t// complete texture map (fmt 1), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap2;\t\t// complete texture map (fmt 2), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap3;\t\t// complete texture map (fmt 3), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap4;\t\t// complete texture map (fmt 4), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap5;\t\t// complete texture map (fmt 5), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap6;\t\t// complete texture map (fmt 6), 2048x2048 texels\n"
+"uniform sampler2D\ttextureMap7;\t\t// complete texture map (fmt 7), 2048x2048 texels\n"
+"uniform vec4\t\tspotEllipse;\t\t// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)\n"
+"uniform vec2\t\tspotRange;\t\t\t// spotlight Z range: .x=start (viewspace coordinates), .y=limit\n"
+"uniform vec3\t\tspotColor;\t\t\t// spotlight RGB color\n"
+"uniform vec3\t\tlighting[2];\t\t// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)\n"
+"uniform float\t\tmapSize;\t\t// texture map size (2048,4096,6144 etc)\n"
+"\n"
+"// Inputs from vertex shader \n"
+"varying vec4\t\tfsSubTexture;\t// .x=texture X, .y=texture Y, .z=texture width, .w=texture height (all in texels)\n"
+"varying vec4\t\tfsTexParams;\t// .x=texture enable (if 1, else 0), .y=use transparency (if > 0), .z=U wrap mode (1=mirror, 0=repeat), .w=V wrap mode\n"
+"varying float\t\tfsTexFormat;\t// T1RGB5 contour texture (if > 0)\n"
+"varying float\t\tfsTexMap;\t// texture map number\n"
+"varying float\t\tfsTransLevel;\t// translucence level, 0.0 (transparent) to 1.0 (opaque)\n"
+"varying vec3\t\tfsLightIntensity;\t// lighting intensity \n"
+"varying float\t\tfsSpecularTerm;\t// specular highlight\n"
+"varying float\t\tfsFogFactor;\t// fog factor\n"
+"varying float\t\tfsViewZ;\t\t// Z distance to fragment from viewpoint at origin\n"
+"\n"
+"/*\n"
+" * WrapTexelCoords():\n"
+" *\n"
+" * Computes the normalized OpenGL S,T coordinates within the 2048x2048 texture\n"
+" * sheet, taking into account wrapping behavior.\n"
+" *\n"
+" * Computing normalized OpenGL texture coordinates (0 to 1) within the \n"
+" * Real3D texture sheet:\n"
+" *\n"
+" * If the texture is not mirrored, we simply have to clamp the\n"
+" * coordinates to fit within the texture dimensions, add the texture\n"
+" * X, Y position to select the appropriate one, and normalize by 2048\n"
+" * (the dimensions of the Real3D texture sheet).\n"
+" *\n"
+" *\t\t= [(u,v)%(w,h)+(x,y)]/(2048,2048)\n"
+" *\n"
+" * If mirroring is enabled, textures are mirrored every odd multiple of\n"
+" * the original texture. To detect whether we are in an odd multiple, \n"
+" * simply divide the coordinate by the texture dimension and check \n"
+" * whether the result is odd. Then, clamp the coordinates as before but\n"
+" * subtract from the last texel to mirror them:\n"
+" *\n"
+" * \t\t= [M*((w-1,h-1)-(u,v)%(w,h)) + (1-M)*(u,v)%(w,h) + (x,y)]/(2048,2048)\n"
+" *\t\twhere M is 1.0 if the texture must be mirrored.\n"
+" *\n"
+" * As an optimization, this function computes TWO texture coordinates\n"
+" * simultaneously. The first is texCoord.xy, the second is in .zw. The other\n"
+" * parameters must have .xy = .zw.\n"
+" */\n"
+"vec4 WrapTexelCoords(vec4 texCoord, vec4 texOffset, vec4 texSize, vec4 mirrorEnable)\n"
+"{\n"
+"\tvec4\tclampedCoord, mirror, glTexCoord;\n"
+"\t\n"
+"\tclampedCoord = mod(texCoord,texSize);\t\t\t\t\t\t// clamp coordinates to within texture size\n"
+"\tmirror = mirrorEnable * mod(floor(texCoord/texSize),2.0);\t// whether this texel needs to be mirrored\n"
+"\n"
+"\tglTexCoord = (\tmirror*(texSize-clampedCoord) +\n"
+"\t\t\t\t\t(vec4(1.0,1.0,1.0,1.0)-mirror)*clampedCoord +\n"
+"\t\t\t\t\ttexOffset\n"
+"\t\t\t\t ) / mapSize;\n"
+"\treturn glTexCoord;\n"
+"}\n"
+"\n"
+"/*\n"
+" * main():\n"
+" *\n"
+" * Fragment shader entry point.\n"
+" */\n"
+"\n"
+"void main(void)\n"
+"{\t\n"
+"\tvec4\tuv_top, uv_bot, c[4];\n"
+"\tvec2\tr;\n"
+"\tvec4\tfragColor;\n"
+"\tvec2\tellipse;\n"
+"\tvec3\tlightIntensity;\n"
+"\tfloat\tinsideSpot;\n"
+"\tint\t\tx;\n"
+"\t\n"
+"\t// Get polygon color for untextured polygons (textured polygons will overwrite)\n"
+"\tif (fsTexParams.x < 0.5)\n"
+"\t\tfragColor = gl_Color;\t\t\n"
+"\telse\n"
+"\t// Textured polygons: set fragment color to texel value\n"
+"\t{\t\t\t\n"
+"\t\t/*\n"
+"\t\t * Bilinear Filtering\n"
+"\t\t *\n"
+"\t\t * In order to get this working on ATI, the number of operations is\n"
+"\t\t * reduced by putting everything into vec4s. uv_top holds the UV \n"
+"\t\t * coordinates for the top two texels (.xy=left, .zw=right) and uv_bot\n"
+"\t\t * is for the lower two.\n"
+"\t\t */\n"
+"\n"
+"\t\t// Compute fractional blending factor, r, and lower left corner of texel 0\n"
+"\t\tuv_bot.xy = gl_TexCoord[0].st-vec2(0.5,0.5);\t// move into the lower left blending texel \n"
+"\t\tr = uv_bot.xy-floor(uv_bot.xy);\t\t\t\t\t// fractional part\n"
+"\t\tuv_bot.xy = floor(uv_bot.xy);\t\t\t\t\t// integral part\n"
+"\t\t\n"
+"\t\t// Compute texel coordinates\n"
+"\t\tuv_bot.xy += vec2(0.5,0.5);\t// offset to center of pixel (should not be needed but it fixes a lot of glitches, esp. on Nvidia)\n"
+"\t\tuv_bot.zw = uv_bot.xy + vec2(1.0,0.0);\t\t\t// compute coordinates of the other three neighbors\n"
+"\t\tuv_top = uv_bot + vec4(0.0,1.0,0.0,1.0);\n"
+"\n"
+"\t\t// Compute the properly wrapped texel coordinates\n"
+"\t\tuv_top = WrapTexelCoords(uv_top,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));\n"
+"\t\tuv_bot = WrapTexelCoords(uv_bot,vec4(fsSubTexture.xy,fsSubTexture.xy),vec4(fsSubTexture.zw,fsSubTexture.zw), vec4(fsTexParams.zw,fsTexParams.zw));\n"
+"\n"
+"\t\t// Fetch the texels from the given texture map\n"
+"\t\tif (fsTexMap < 0.5f)\t{\n"
+"\t\t\tc[0]=texture2D(textureMap0, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap0, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap0, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap0, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 1.5f) {\n"
+"            c[0]=texture2D(textureMap1, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap1, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap1, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap1, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 2.5f) {\n"
+"            c[0]=texture2D(textureMap2, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap2, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap2, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap2, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 3.5f) {\n"
+"            c[0]=texture2D(textureMap3, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap3, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap3, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap3, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 4.5f) {\n"
+"            c[0]=texture2D(textureMap4, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap4, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap4, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap4, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 5.5f) {\n"
+"            c[0]=texture2D(textureMap5, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap5, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap5, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap5, uv_top.zw);       // top-right\n"
+"\t\t} else if (fsTexMap < 6.5f) {\n"
+"\t\t\tc[0]=texture2D(textureMap6, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap6, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap6, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap6, uv_top.zw);       // top-right\n"
+"\t\t} else {\n"
+"            c[0]=texture2D(textureMap7, uv_bot.xy);       // bottom-left (base texel)\n"
+"\t\t\tc[1]=texture2D(textureMap7, uv_bot.zw);       // bottom-right\n"
+"\t\t\tc[2]=texture2D(textureMap7, uv_top.xy);       // top-left\n"
+"\t\t\tc[3]=texture2D(textureMap7, uv_top.zw);       // top-right\n"
+"\t\t}                      \n"
+"\n"
+"\t\t// Interpolate texels and blend result with material color to determine final (unlit) fragment color\n"
+"\t\t// fragColor = (c[0]*(1.0-r.s)*(1.0-r.t) + c[1]*r.s*(1.0-r.t) + c[2]*(1.0-r.s)*r.t + c[3]*r.s*r.t);\n"
+"\t\t// Faster method:\n"
+"\t\tc[0] += (c[1]-c[0])*r.s;\t\t\t// 2 alu\n"
+"\t\tc[2] += (c[3]-c[2])*r.s;\t\t\t// 2 alu\n"
+"\t\tfragColor = c[0]+(c[2]-c[0])*r.t;\t// 2 alu\n"
+"\t\n"
+"\t\t/*\n"
+"\t\t * T1RGB5:\n"
+"\t\t *\n"
+"\t\t * The transparency bit determines whether to discard pixels (if set).\n"
+"\t\t * What is unknown is how this bit behaves when interpolated. OpenGL\n"
+"\t\t * processes it as an alpha value, so it might concievably be blended\n"
+"\t\t * with neighbors. Here, an arbitrary threshold is chosen.\n"
+"\t\t *\n"
+"\t\t * To-do: blending could probably enabled and this would work even\n"
+"\t\t * better with a hard threshold.\n"
+"\t\t *\n"
+"\t\t * Countour processing also seems to be enabled for RGBA4 textures.\n"
+"\t\t * When the alpha value is 0.0 (or close), pixels are discarded \n"
+"\t\t * entirely.\n"
+"\t\t */\n"
+"\t\tif (fsTexParams.y > 0.5)\t// contour processing enabled\n"
+"\t\t{\n"
+"\t\t\tif (fragColor.a < 0.01)\t// discard anything with alpha == 0\n"
+"\t\t\t\tdiscard;\n"
+"\t\t}\n"
+"\t\t\n"
+"\t\t// If contour texture and not discarded, force alpha to 1.0 because will later be modified by polygon translucency\n"
+"\t\tif (fsTexFormat < 0.5)\t\t// contour (T1RGB5) texture map\n"
+"\t\t\tfragColor.a = 1.0;\n"
+"\t}\n"
+"\n"
+"\t// Compute spotlight and apply lighting\n"
+"\tellipse = (gl_FragCoord.xy-spotEllipse.xy)/spotEllipse.zw;\n"
+"\tinsideSpot = dot(ellipse,ellipse);\n"
+"\tif ((insideSpot <= 1.0) &&  (fsViewZ>=spotRange.x) && (fsViewZ<spotRange.y))\n"
+"\t\tlightIntensity = fsLightIntensity+(1.0-insideSpot)*spotColor;\n"
+"\telse\n"
+"\t\tlightIntensity = fsLightIntensity;\n"
+"\tfragColor.rgb *= lightIntensity;\n"
+"\tfragColor.rgb += vec3(fsSpecularTerm,fsSpecularTerm,fsSpecularTerm);\n"
+"\t\n"
+"\t// Translucency (modulates existing alpha channel for RGBA4 texels)\n"
+"\tfragColor.a *= fsTransLevel;\n"
+"\n"
+"\t// Apply fog under the control of fog factor setting from polygon header\n"
+"\tfragColor.rgb = mix(gl_Fog.color.rgb, fragColor.rgb, fsFogFactor);\n"
+"\n"
+"\t// Store final color\n"
+"\tgl_FragColor = fragColor;\n"
+"}\n"
+};
+
+} // Legacy3D
+
+#endif	// INCLUDED_SHADERS3D_H

Src/Graphics/Legacy3D/TextureRefs.cpp

@@ -0,0 +1,333 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * CTextureRefs.cpp
+ *
+ * Class that tracks unique texture references, eg in a cached model.
+ *
+ * Texture references are stored internally as a 27-bit field (3 bits for format, 6 bits each for x, y, width & height) to save space.
+ * 
+ * A pre-allocated array is used for storing up to TEXREFS_ARRAY_SIZE texture references.  When that limit is exceeded, it switches
+ * to using a hashset to store the texture references, but this requires extra memory allocation.
+ */
+
+#include "Supermodel.h"
+
+namespace Legacy3D {
+
+CTextureRefs::CTextureRefs() : m_size(0), m_hashCapacity(0), m_hashEntries(NULL)
+{
+	//
+}
+
+CTextureRefs::~CTextureRefs()
+{
+	DeleteAllHashEntries();
+}
+
+unsigned CTextureRefs::GetSize() const
+{
+	return m_size;
+}
+
+void CTextureRefs::Clear()
+{
+	// Delete all hash entries
+	DeleteAllHashEntries();
+	m_size = 0;
+	m_hashCapacity = 0;
+	m_hashEntries = NULL;
+}
+
+bool CTextureRefs::ContainsRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height)
+{
+	// Pack texture reference into bitfield
+	unsigned texRef = (fmt&7)<<24|(x&0x7E0)<<13|(y&0x7E0)<<7|(width&0x7E0)<<1|(height&0x7E0)>>5;
+	
+	// Check if using array or hashset
+	if (m_size <= TEXREFS_ARRAY_SIZE)
+	{
+		// See if texture reference held in array
+		for (unsigned i = 0; i < m_size; i++)
+		{
+			if (texRef == m_array[i])
+				return true;
+		}
+		return false;
+	}
+	else
+		// See if texture reference held in hashset
+		return HashContains(texRef);
+}
+
+bool CTextureRefs::AddRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height)
+{
+	// Pack texture reference into bitfield
+	unsigned texRef = (fmt&7)<<24|(x&0x7E0)<<13|(y&0x7E0)<<7|(width&0x7E0)<<1|(height&0x7E0)>>5;
+
+	// Check if using array or hashset
+	if (m_size <= TEXREFS_ARRAY_SIZE)
+	{
+		// See if already held in array, if so nothing to do
+		for (unsigned i = 0; i < m_size; i++)
+		{
+			if (texRef == m_array[i])
+				return true;
+		}
+		// If not, check if array is full
+		if (m_size == TEXREFS_ARRAY_SIZE)
+		{
+			// If so, set initial hashset capacity to 47 to initialize it
+			UpdateHashCapacity(47);
+			// Copy array into hashset
+			for (unsigned i = 0; i < TEXREFS_ARRAY_SIZE; i++)
+				AddToHash(m_array[i]);
+			// Add texture reference to hashset
+			AddToHash(texRef);
+		}
+		else
+		{
+			// Add texture reference to array
+			m_array[m_size] = texRef;
+			m_size++;
+		}
+		return true;
+	}
+	else
+		// Add texture reference to hashset
+		return AddToHash(texRef);
+}
+
+bool CTextureRefs::RemoveRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height)
+{
+	// Pack texture reference into bitfield
+	unsigned texRef = (fmt&7)<<24|(x&0x7E0)<<13|(y&0x7E0)<<7|(width&0x7E0)<<1|(height&0x7E0)>>5;
+
+	// Check if using array or hashset
+	if (m_size <= TEXREFS_ARRAY_SIZE)
+	{
+		for (unsigned i = 0; i < m_size; i++)
+		{
+			if (texRef == m_array[i])
+			{
+				for (unsigned j = i + 1; j < m_size; j++)
+					m_array[j - 1] = m_array[j];
+				m_size--;
+				return true;
+			}
+		}
+		return false;
+	}
+	else 
+	{
+		// Remove texture reference from hashset
+		bool removed = RemoveFromHash(texRef);
+
+		// See if should switch back to array
+		if (m_size == TEXREFS_ARRAY_SIZE)
+		{
+			// Loop through all hash entries and copy texture references into array
+			unsigned j = 0;
+			for (unsigned i = 0; i < m_hashCapacity; i++)
+			{
+				for (HashEntry *entry = m_hashEntries[i]; entry; entry = entry->nextEntry)
+					m_array[j++] = entry->texRef;
+			}
+			// Delete all hash entries
+			DeleteAllHashEntries();
+		}
+		return removed;
+	}
+}
+
+void CTextureRefs::DecodeAllTextures(CLegacy3D *Render3D)
+{
+	// Check if using array or hashset
+	if (m_size <= TEXREFS_ARRAY_SIZE)
+	{
+		// Loop through elements in array and call CLegacy3D::DecodeTexture
+		for (unsigned i = 0; i < m_size; i++)
+		{
+			// Unpack texture reference from bitfield 
+			unsigned texRef = m_array[i];
+			unsigned fmt = texRef>>24;
+			unsigned x = (texRef>>13)&0x7E0;
+			unsigned y = (texRef>>7)&0x7E0;
+			unsigned width = (texRef>>1)&0x7E0;
+			unsigned height = (texRef<<5)&0x7E0;
+			Render3D->DecodeTexture(fmt, x, y, width, height);
+		}
+	}
+	else
+	{
+		// Loop through all hash entriesa and call CLegacy3D::DecodeTexture
+		for (unsigned i = 0; i < m_hashCapacity; i++)
+		{
+			for (HashEntry *entry = m_hashEntries[i]; entry; entry = entry->nextEntry)
+			{
+				// Unpack texture reference from bitfield 
+				unsigned texRef = entry->texRef;
+				unsigned fmt = texRef>>24;
+				unsigned x = (texRef>>13)&0x7E0;
+				unsigned y = (texRef>>7)&0x7E0;
+				unsigned width = (texRef>>1)&0x7E0;
+				unsigned height = (texRef<<5)&0x7E0;
+				Render3D->DecodeTexture(fmt, x, y, width, height);
+			}
+		}
+	}
+}
+
+bool CTextureRefs::UpdateHashCapacity(unsigned capacity)
+{
+	unsigned oldCapacity = m_hashCapacity;
+	HashEntry **oldEntries = m_hashEntries;
+	// Update capacity and create new empty entries array
+	m_hashCapacity = capacity;
+	m_hashEntries = new(std::nothrow) HashEntry*[capacity];
+	if (!m_hashEntries)
+		return false;
+	memset(m_hashEntries, NULL, capacity * sizeof(HashEntry*));
+	if (oldEntries)
+	{
+		// Redistribute entries into new entries array
+		for (unsigned i = 0; i < oldCapacity; i++)
+		{
+			HashEntry *entry = oldEntries[i];
+			while (entry)
+			{
+				HashEntry *nextEntry = entry->nextEntry;
+				unsigned hash = entry->texRef % capacity;
+				entry->nextEntry = m_hashEntries[hash];
+				m_hashEntries[hash] = entry;
+				entry = nextEntry;
+			}
+		}
+		// Delete old entries array
+		delete[] oldEntries;
+	}
+	return true;
+}
+
+HashEntry *CTextureRefs::CreateHashEntry(unsigned texRef, bool &hashCapacityUpdated)
+{
+	// Update size and increase hash capacity if required
+	m_size++;
+	hashCapacityUpdated = m_size >= m_hashCapacity;
+	if (hashCapacityUpdated)
+	{
+		if (m_hashCapacity < 89)
+			UpdateHashCapacity(89); // Capacity of 89 gives good sequence of mostly prime capacities (89, 179, 359, 719, 1439, 2879 etc)
+		else
+			UpdateHashCapacity(2 * m_hashCapacity + 1);
+	}
+	return new(std::nothrow) HashEntry(texRef);
+}
+
+void CTextureRefs::DeleteHashEntry(HashEntry *entry)
+{
+	// Update size and delete hash entry
+	m_size--;
+	delete entry;
+}
+
+void CTextureRefs::DeleteAllHashEntries()
+{
+	if (!m_hashEntries)
+		return;
+	// Delete all hash entries and their storage
+	for (unsigned i = 0; i < m_hashCapacity; i++)
+	{
+		HashEntry *entry = m_hashEntries[i];
+		if (entry)
+			delete entry;
+	}
+	delete[] m_hashEntries;
+}
+
+bool CTextureRefs::AddToHash(unsigned texRef)
+{
+	// Convert texture reference to hash value
+	unsigned hash = texRef % m_hashCapacity;
+	// Loop through linked list for hash value and see if have texture reference already
+	HashEntry *headEntry = m_hashEntries[hash];
+	HashEntry *entry = headEntry;
+	while (entry && texRef != entry->texRef)
+		entry = entry->nextEntry;
+	// If found, nothing to do
+	if (entry)
+		return true;
+	// Otherwise, create new hash entry for texture reference
+	bool hashCapacityUpdated;
+	entry = CreateHashEntry(texRef, hashCapacityUpdated);
+	// If couldn't create entry (ie out of memory), let caller know
+	if (!entry)
+		return false;
+	if (hashCapacityUpdated)
+	{
+		// If hash capacity was increased recalculate hash value
+		hash = texRef % m_hashCapacity;
+		headEntry = m_hashEntries[hash];
+	}
+	// Store hash entry in linked list for hash value
+	entry->nextEntry = headEntry;
+	m_hashEntries[hash] = entry;
+	return true;
+}
+
+bool CTextureRefs::RemoveFromHash(unsigned texRef)
+{
+	// Convert texture reference to hash value
+	unsigned hash = texRef % m_hashCapacity;
+	// Loop through linked list for hash value and see if have texture reference
+	HashEntry *entry = m_hashEntries[hash];
+	HashEntry *prevEntry = NULL;
+	while (entry && texRef != entry->texRef)
+	{
+		prevEntry = entry;
+		entry = entry->nextEntry;
+    }
+	// If not found, nothing to do
+	if (!entry)
+		return false;
+	// Otherwise, remove entry from linked list for hash value
+	if (prevEntry)	
+		prevEntry->nextEntry = entry->nextEntry;
+	else
+		m_hashEntries[hash] = entry->nextEntry;
+	// Delete hash entry storage
+	DeleteHashEntry(entry);
+	return true;
+}
+
+bool CTextureRefs::HashContains(unsigned texRef) const
+{
+	// Convert texture reference to hash value
+	unsigned hash = texRef % m_hashCapacity;
+	// Loop through linked list for hash value and see if have texture reference
+	HashEntry *entry = m_hashEntries[hash];
+	while (entry && texRef != entry->texRef)
+		entry = entry->nextEntry;
+	return !!entry;
+}
+
+} // Legacy3D

Src/Graphics/Legacy3D/TextureRefs.h

@@ -0,0 +1,168 @@
+/**
+ ** Supermodel
+ ** A Sega Model 3 Arcade Emulator.
+ ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ **
+ ** This file is part of Supermodel.
+ **
+ ** Supermodel 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 3 of the License, or (at your option)
+ ** any later version.
+ **
+ ** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+ ** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+ ** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+ ** more details.
+ **
+ ** You should have received a copy of the GNU General Public License along
+ ** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+ **/
+ 
+/*
+ * CTextureRefs.h
+ * 
+ * Class that tracks unique texture references, eg in a cached model.
+ */
+
+#ifndef INCLUDED_TEXTUREREFS_H
+#define INCLUDED_TEXTUREREFS_H
+
+namespace Legacy3D {
+
+#define TEXREFS_ARRAY_SIZE 12
+
+// Hash entry that holds a texture reference in the hashset
+struct HashEntry
+{
+	const unsigned texRef;	// Texture reference as a bitfield
+	HashEntry *nextEntry;   // Next entry with the same hash
+  
+	HashEntry(unsigned theTexRef) : texRef(theTexRef), nextEntry(NULL) { }
+};
+
+class CLegacy3D;
+
+class CTextureRefs
+{
+public:
+	/*
+	 * CTextureRefs():
+	 *
+	 * Constructor.
+	 */
+    CTextureRefs();
+    
+	/*
+	 * ~CTextureRefs():
+	 *
+	 * Destructor.
+	 */
+    ~CTextureRefs();
+
+	/*
+	 * GetSize():
+	 *
+	 * Returns number of unique texture references held.
+	 */
+    unsigned GetSize() const;
+
+	/*
+	 * Clear():
+	 *
+	 * Removes all texture references.
+	 */
+	void Clear();
+
+	/*
+	 * ContainsRef(fmt, x, y, width, height):
+	 *
+	 * Returns true if holds the given texture reference.
+	 */
+	bool ContainsRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height);
+
+	/*
+	 * AddRef(fmt, x, y, width, height):
+	 *
+	 * Adds the given texture reference.  Returns false if it was not possible to add the reference (ie out of memory).
+	 */
+	bool AddRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height);
+
+	/*
+	 * RemoveRef(fmt, x, y, width, height):
+	 *
+	 * Removes the given texture reference.  Return true if the reference was found.
+	 */
+	bool RemoveRef(unsigned fmt, unsigned x, unsigned y, unsigned width, unsigned height);
+
+	/*
+	 * RemoveRef(fmt, x, y, width, height):
+	 *
+	 * Decodes all texture references held, calling CLegacy3D::DecodeTexture for each one.
+	 */
+	void DecodeAllTextures(CLegacy3D *Render3D);
+
+private:
+	// Number of texture references held.
+	unsigned m_size;
+
+	// Pre-allocated array used to hold first TEXREFS_ARRAY_SIZE texture references.
+	unsigned m_array[TEXREFS_ARRAY_SIZE];
+
+	// Dynamically allocated hashset used to hold texture references when there are more than TEXREFS_ARRAY_SIZE.
+	unsigned m_hashCapacity;
+	HashEntry **m_hashEntries;
+    
+	/*
+	 * UpdateHashCapacity(hashCapacity)
+	 *
+	 * Increases capacity of the hashset to given size.
+	 */
+	bool UpdateHashCapacity(unsigned hashCapacity);
+
+	/*
+	 * CreateHashEntry(texRef, hashCapacityUpdated)
+	 *
+	 * Creates and returns a new hash entry, updating the capacity if required (hashCapacityUpdated is set to true).
+	 */
+	HashEntry *CreateHashEntry(unsigned texRef, bool &hashCapacityUpdated);
+
+	/*
+	 * DeleteHashEntry(entry)
+	 *
+	 * Deletes the given hash entry and its storage.
+	 */
+	void DeleteHashEntry(HashEntry *entry);
+	
+	/*
+	 * DeleteAllHashEntries()
+	 *
+	 * Deletes all hash entries and their storage.
+	 */
+    void DeleteAllHashEntries();
+	
+	/*
+	 * AddToHash(texRef)
+	 *
+	 * Adds the given texture reference (as a bitfield) to the hashset.
+	 */
+	bool AddToHash(unsigned texRef);
+
+	/*
+	 * RemoveFromHash(texRef)
+	 *
+	 * Removes the given texture reference (as a bitfield) from the hashset.
+	 */
+	bool RemoveFromHash(unsigned texRef);
+
+	/*
+	 * HashContains(texRef)
+	 *
+	 * Returns true if given texture reference (as a bitfield) is held in the hashset.
+	 */
+    bool HashContains(unsigned texRef) const;
+};
+
+} // Legacy3D
+
+#endif	// INCLUDED_TEXTUREREFS_H

Src/Graphics/New3D/Mat4.cpp

@@ -0,0 +1,353 @@
+#include "Mat4.h"
+#define _USE_MATH_DEFINES
+#include <math.h>
+#include <utility>
+
+#ifndef M_PI
+#define M_PI 3.14159265359
+#endif
+
+namespace New3D {
+
+Mat4::Mat4()
+{
+	LoadIdentity();
+}
+
+void Mat4::LoadIdentity()
+{
+	float *m = currentMatrix;
+
+	m[0] = 1; m[4] = 0; m[8 ] = 0; m[12] = 0;
+	m[1] = 0; m[5] = 1; m[9 ] = 0; m[13] = 0;
+	m[2] = 0; m[6] = 0; m[10] = 1; m[14] = 0;
+	m[3] = 0; m[7] = 0; m[11] = 0; m[15] = 1;
+}
+
+void Mat4::MultiMatrices(const float a[16], const float b[16], float r[16]) {
+
+#define A(row,col)  a[(col<<2)+row]
+#define B(row,col)  b[(col<<2)+row]
+#define P(row,col)  r[(col<<2)+row]
+
+	int i;
+	for (i = 0; i < 4; i++) {
+		const float ai0 = A(i, 0), ai1 = A(i, 1), ai2 = A(i, 2), ai3 = A(i, 3);
+		P(i, 0) = ai0 * B(0, 0) + ai1 * B(1, 0) + ai2 * B(2, 0) + ai3 * B(3, 0);
+		P(i, 1) = ai0 * B(0, 1) + ai1 * B(1, 1) + ai2 * B(2, 1) + ai3 * B(3, 1);
+		P(i, 2) = ai0 * B(0, 2) + ai1 * B(1, 2) + ai2 * B(2, 2) + ai3 * B(3, 2);
+		P(i, 3) = ai0 * B(0, 3) + ai1 * B(1, 3) + ai2 * B(2, 3) + ai3 * B(3, 3);
+	}
+
+#undef A
+#undef B
+#undef p
+}
+
+void Mat4::Copy(const float in[16], float out[16]) {
+
+	for (int i = 0; i<16; i++) {
+		out[i] = in[i];
+	}
+}
+
+void Mat4::Translate(float x, float y, float z) {
+
+	//=============
+	float m[16];
+	float cur[16];
+	//=============
+
+	Mat4::Copy(currentMatrix, cur);
+
+	m[0] = 1;
+	m[1] = 0;
+	m[2] = 0;
+	m[3] = 0;
+
+	m[4] = 0;
+	m[5] = 1;
+	m[6] = 0;
+	m[7] = 0;
+
+	m[8] = 0;
+	m[9] = 0;
+	m[10] = 1;
+	m[11] = 0;
+
+	m[12] = x;
+	m[13] = y;
+	m[14] = z;
+	m[15] = 1;
+
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::Scale(float x, float y, float z) {
+
+	//=============
+	float m[16];
+	float cur[16];
+	//=============
+
+	Mat4::Copy(currentMatrix, cur);
+
+	m[0] = x;
+	m[1] = 0;
+	m[2] = 0;
+	m[3] = 0;
+
+	m[4] = 0;
+	m[5] = y;
+	m[6] = 0;
+	m[7] = 0;
+
+	m[8] = 0;
+	m[9] = 0;
+	m[10] = z;
+	m[11] = 0;
+
+	m[12] = 0;
+	m[13] = 0;
+	m[14] = 0;
+	m[15] = 1;
+
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::Rotate(float angle, float x, float y, float z) {
+
+	//============
+	float c;
+	float s;
+	float m[16];
+	float cur[16];
+	//===========
+
+	Mat4::Copy(currentMatrix, cur);
+
+	// normalise vector first
+	{
+		//========
+		float len;
+		//========
+
+		len = sqrt(x * x + y * y + z * z);
+		
+		x /= len;
+		y /= len;
+		z /= len;
+	}
+
+	c = cos(angle*3.14159265f / 180.0f);
+	s = sin(angle*3.14159265f / 180.0f);
+
+	m[0] = (x*x) * (1 - c) + c;
+	m[1] = (y*x) * (1 - c) + (z*s);
+	m[2] = (x*z) * (1 - c) - (y*s);
+	m[3] = 0;
+
+	m[4] = (x*y) * (1 - c) - (z*s);
+	m[5] = (y*y) * (1 - c) + c;
+	m[6] = (y*z) * (1 - c) + (x*s);
+	m[7] = 0;
+
+	m[8] = (x*z) * (1 - c) + (y*s);
+	m[9] = (y*z) * (1 - c) - (x*s);
+	m[10] = (z*z) * (1 - c) + c;
+	m[11] = 0;
+
+	m[12] = 0;
+	m[13] = 0;
+	m[14] = 0;
+	m[15] = 1;
+
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::Frustum(float left, float right, float bottom, float top, float nearVal, float farVal) {
+
+	//=====================
+	float m[16];
+	float cur[16];
+	float x, y, a, b, c, d;
+	//=====================
+
+	Mat4::Copy(currentMatrix, cur);
+
+	x = (2.0F*nearVal) / (right - left);
+	y = (2.0F*nearVal) / (top - bottom);
+	a = (right + left) / (right - left);
+	b = (top + bottom) / (top - bottom);
+	c = -(farVal + nearVal) / (farVal - nearVal);
+	d = -(2.0F*farVal*nearVal) / (farVal - nearVal);
+
+	m[0] = x;
+	m[1] = 0;
+	m[2] = 0;
+	m[3] = 0;
+
+	m[4] = 0;
+	m[5] = y;
+	m[6] = 0;
+	m[7] = 0;
+
+	m[8] = a;
+	m[9] = b;
+	m[10] = c;
+	m[11] = -1;
+
+	m[12] = 0;
+	m[13] = 0;
+	m[14] = d;
+	m[15] = 0;
+
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::Perspective(float fovy, float aspect, float zNear, float zFar)
+{
+	//=========
+	float ymax;
+	float xmax;
+	//=========
+
+	ymax = zNear * tanf(fovy * (float)M_PI / 360.0f);
+	xmax = ymax * aspect;
+
+	Frustum(-xmax, xmax, -ymax, ymax, zNear, zFar);
+}
+
+void Mat4::Ortho(float left, float right, float bottom, float top, float nearVal, float farVal)
+{
+	//================
+	float m[16];
+	float cur[16];
+	float tx, ty, tz;
+	//================
+
+	Mat4::Copy(currentMatrix, cur);
+
+	tx = -(right + left) / (right - left);
+	ty = -(top + bottom) / (top - bottom);
+	tz = -(farVal + nearVal) / (farVal - nearVal);
+
+	m[0] = 2/(right-left);
+	m[1] = 0;
+	m[2] = 0;
+	m[3] = 0;
+
+	m[4] = 0;
+	m[5] = 2/(top-bottom);
+	m[6] = 0;
+	m[7] = 0;
+
+	m[8] = 0;
+	m[9] = 0;
+	m[10] = -2/(farVal-nearVal);
+	m[11] = 0;
+
+	m[12] = tx;
+	m[13] = ty;
+	m[14] = tz;
+	m[15] = 1;
+
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::Transpose(float m[16]) {
+
+	std::swap(m[1], m[4]);
+	std::swap(m[2], m[8]);
+	std::swap(m[3], m[12]);
+	std::swap(m[6], m[9]);
+	std::swap(m[7], m[13]);
+	std::swap(m[11], m[14]);
+}
+
+void Mat4::MultMatrix(const float *m)
+{
+	if (!m) {
+		return;
+	}
+
+	//============
+	float cur[16];
+	//============
+
+	Mat4::Copy(currentMatrix, cur);
+	Mat4::MultiMatrices(cur, m, currentMatrix);
+}
+
+void Mat4::LoadMatrix(const float *m)
+{
+	if (!m) {
+		return; 
+	}
+
+	Mat4::Copy(m, currentMatrix);
+}
+
+void Mat4::LoadTransposeMatrix(const float *m)
+{
+	if (!m) {
+		return;
+	}
+
+	//=============
+	float copy[16];
+	//=============
+
+	Mat4::Copy(m, copy);
+	Mat4::Transpose(copy);
+	Mat4::LoadMatrix(copy);
+}
+
+void Mat4::MultTransposeMatrix(const float *m)
+{
+	if (!m) {
+		return;
+	}
+
+	//=============
+	float copy[16];
+	//=============
+
+	Mat4::Copy(m, copy);
+	Mat4::Transpose(copy);
+	Mat4::MultMatrix(copy);
+}
+
+void Mat4::PushMatrix()
+{
+	//==============
+	Mat4Container m;
+	//==============
+
+	if (m_vMat4.size() > 128) {
+		return;	// check for overflow
+	}
+
+	Mat4::Copy(currentMatrix, m.mat);
+
+	m_vMat4.push_back(m);
+}
+
+void Mat4::PopMatrix()
+{
+	if (m_vMat4.empty()) {
+		return;	// check for underflow
+	}
+
+	Mat4::Copy(m_vMat4.back().mat, currentMatrix);
+
+	m_vMat4.pop_back();
+}
+
+// flush the matrix stack
+void Mat4::Release()
+{
+	m_vMat4.clear();
+}
+
+} // New3D

Src/Graphics/New3D/Mat4.h

@@ -0,0 +1,48 @@
+#ifndef _MAT4_H_
+#define _MAT4_H_
+
+#include <vector>
+
+namespace New3D {
+
+class Mat4
+{
+public:
+
+	Mat4();
+
+	void LoadIdentity			();
+	void Translate				(float x, float y, float z);
+	void Rotate					(float angle, float x, float y, float z);
+	void Scale					(float x, float y, float z);
+	void Frustum				(float left, float right, float bottom, float top, float nearVal, float farVal);
+	void Ortho					(float left, float right, float bottom, float top, float nearVal, float farVal);
+	void Perspective			(float fovy, float aspect, float zNear, float zFar);
+	void MultMatrix				(const float *m);
+	void LoadMatrix				(const float *m);
+	void LoadTransposeMatrix	(const float *m);
+	void MultTransposeMatrix	(const float *m);
+	void PushMatrix				();
+	void PopMatrix				();
+	void Release				();
+
+	operator const float*		() { return currentMatrix; }
+
+	float currentMatrix[16];
+private:
+
+	void MultiMatrices			(const float a[16], const float b[16], float r[16]);
+	void Copy					(const float in[16], float out[16]);
+	void Transpose				(float m[16]);
+
+	struct Mat4Container {
+		float mat[16];	// we must wrap the matrix inside a struct otherwise vector doesn't work
+	};
+
+	std::vector<Mat4Container> m_vMat4;
+};
+
+} // New3D
+
+#endif
+

Src/Graphics/New3D/Model.cpp

@@ -0,0 +1,60 @@
+#include "Model.h"
+
+namespace New3D {
+
+NodeAttributes::NodeAttributes()
+{
+	currentTexOffsetX = 0;
+	currentTexOffsetY = 0;
+	currentTexOffset = 0;
+}
+
+bool NodeAttributes::Push()
+{
+	//=============
+	NodeAttribs na;
+	//=============
+
+	// check for overflow
+	if (m_vecAttribs.size() >= 128) {
+		return false;
+	}
+
+	na.texOffset = currentTexOffset;
+	na.texOffsetX = currentTexOffsetX;
+	na.texOffsetY = currentTexOffsetY;
+
+	m_vecAttribs.push_back(na);
+
+	return true;
+}
+
+bool NodeAttributes::Pop()
+{
+	if (m_vecAttribs.empty()) {
+		return false;	// check for underflow
+	}
+
+	currentTexOffset = m_vecAttribs.back().texOffset;
+	currentTexOffsetX = m_vecAttribs.back().texOffsetX;
+	currentTexOffsetY = m_vecAttribs.back().texOffsetY;
+
+	m_vecAttribs.pop_back();
+
+	return true;
+}
+
+bool NodeAttributes::StackLimit()
+{
+	return m_vecAttribs.size() >= 1024;
+}
+
+void NodeAttributes::Reset()
+{
+	currentTexOffset = 0;
+	currentTexOffsetX = 0;
+	currentTexOffsetY = 0;
+	m_vecAttribs.clear();
+}
+
+} // New3D

Src/Graphics/New3D/Model.h

@@ -0,0 +1,127 @@
+#ifndef _MODEL_H_
+#define _MODEL_H_
+
+#include "types.h"
+#include <vector>
+#include <unordered_map>
+#include <map>
+#include <memory>
+#include "Texture.h"
+#include "Mat4.h"
+
+namespace New3D {
+
+struct Vertex
+{
+	float pos[3];
+	float normal[3];
+	float texcoords[2];
+	UINT8 color[4];			//rgba
+};
+
+struct Poly		// our polys are always 3 triangles, unlike the real h/w
+{
+	Vertex p1;
+	Vertex p2;
+	Vertex p3;
+};
+
+struct R3DPoly
+{
+	Vertex v[4];			// just easier to have them as an array
+	float faceNormal[3];	// we need this to help work out poly winding, i assume the h/w uses this instead of calculating normals itself
+	int number = 4;
+};
+
+struct Mesh
+{
+	std::shared_ptr<Texture> texture;
+
+	// attributes
+	bool doubleSided	= false;
+	bool textured		= false;
+	bool polyAlpha		= false;		// specified in the rgba colour
+	bool textureAlpha	= false;		// use alpha in texture
+	bool alphaTest		= false;		// discard fragment based on alpha (ogl does this with fixed function)
+	bool lighting		= false;
+	bool testBit		= false;
+	bool clockWise		= true;			// we need to check if the matrix will change the winding
+
+	float fogIntensity = 1.0f;
+
+	// texture
+	bool mirrorU		= false;
+	bool mirrorV		= false;
+
+	// opengl resources
+	int vboOffset		= 0;			// this will be calculated later
+	int triangleCount	= 0;
+};
+
+struct SortingMesh : public Mesh		// This struct temporarily holds the model data, before it gets copied to the main buffer
+{
+	std::vector<Poly> polys;
+};
+
+struct Model
+{
+	std::vector<Mesh> meshes;
+
+	//matrices
+	float modelMat[16];
+
+
+
+	// misc
+	int lutIdx = 0;
+};
+
+struct Viewport
+{
+	Mat4	projectionMatrix;		// projection matrix
+	float	lightingParams[6];		// lighting parameters (see RenderViewport() and vertex shader)
+	float	spotEllipse[4];			// spotlight ellipse (see RenderViewport())
+	float	spotRange[2];			// Z range
+	float	spotColor[3];			// color
+	float	fogParams[5];			// fog parameters (...)
+	int		x, y;					// viewport coordinates (scaled and in OpenGL format)
+	int		width, height;			// viewport dimensions (scaled for display surface size)
+	int		priority;
+};
+
+class NodeAttributes
+{
+public:
+
+	NodeAttributes();
+
+	bool Push();
+	bool Pop();
+	bool StackLimit();
+	void Reset();
+
+	int currentTexOffsetX;
+	int currentTexOffsetY;
+	int currentTexOffset;		// raw value
+
+private:
+
+	struct NodeAttribs
+	{
+		int texOffsetX;
+		int texOffsetY;
+		int texOffset;
+	};
+	std::vector<NodeAttribs> m_vecAttribs;
+};
+
+struct Node
+{
+	Viewport viewport;
+	std::vector<Model> models;
+};
+
+} // New3D
+
+
+#endif

Src/Graphics/New3D/New3D.h

@@ -0,0 +1,216 @@
+/**
+** Supermodel
+** A Sega Model 3 Arcade Emulator.
+** Copyright 2011 Bart Trzynadlowski, Nik Henson
+**
+** This file is part of Supermodel.
+**
+** Supermodel 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 3 of the License, or (at your option)
+** any later version.
+**
+** Supermodel is distributed in the hope that it will be useful, but WITHOUT
+** ANY WARRANTY; without even the implied warranty of MERCHANTABILITY or
+** FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for
+** more details.
+**
+** You should have received a copy of the GNU General Public License along
+** with Supermodel.  If not, see <http://www.gnu.org/licenses/>.
+**/
+
+/*
+* New3D.h
+*
+* Header file defining the CNew3D class: OpenGL Real3D graphics engine.
+*/
+
+#ifndef INCLUDED_NEW3D_H
+#define INCLUDED_NEW3D_H
+
+#include "Pkgs/glew.h"
+#include "Types.h"
+#include "TextureSheet.h"
+#include "Graphics/IRender3D.h"
+#include "Model.h"
+#include "Mat4.h"
+#include "R3DShader.h"
+#include "VBO.h"
+
+namespace New3D {
+
+class CNew3D : public IRender3D
+{
+public:
+	/*
+	* RenderFrame(void):
+	*
+	* Renders the complete scene database. Must be called between BeginFrame() and
+	* EndFrame(). This function traverses the scene database and builds up display
+	* lists.
+	*/
+	void RenderFrame(void);
+
+	/*
+	* BeginFrame(void):
+	*
+	* Prepare to render a new frame. Must be called once per frame prior to
+	* drawing anything.
+	*/
+	void BeginFrame(void);
+
+	/*
+	* EndFrame(void):
+	*
+	* Signals the end of rendering for this frame. Must be called last during
+	* the frame.
+	*/
+	void EndFrame(void);
+
+	/*
+	* UploadTextures(x, y, width, height):
+	*
+	* Signals that a portion of texture RAM has been updated.
+	*
+	* Parameters:
+	*		x		X position within texture RAM.
+	*		y		Y position within texture RAM.
+	*		width	Width of texture data in texels.
+	*		height	Height.
+	*/
+	void UploadTextures(unsigned x, unsigned y, unsigned width, unsigned height);
+
+	/*
+	* AttachMemory(cullingRAMLoPtr, cullingRAMHiPtr, polyRAMPtr, vromPtr,
+	* 				textureRAMPtr):
+	*
+	* Attaches RAM and ROM areas. This must be done prior to any rendering
+	* otherwise the program may crash with an access violation.
+	*
+	* Parameters:
+	*		cullingRAMLoPtr		Pointer to low culling RAM (4 MB).
+	*		cullingRAMHiPtr		Pointer to high culling RAM (1 MB).
+	*		polyRAMPtr			Pointer to polygon RAM (4 MB).
+	*		vromPtr				Pointer to video ROM (64 MB).
+	*		textureRAMPtr		Pointer to texture RAM (8 MB).
+	*/
+	void AttachMemory(const UINT32 *cullingRAMLoPtr,
+		const UINT32 *cullingRAMHiPtr, const UINT32 *polyRAMPtr,
+		const UINT32 *vromPtr, const UINT16 *textureRAMPtr);
+
+	/*
+	* SetStep(stepID):
+	*
+	* Sets the Model 3 hardware stepping, which also determines the Real3D
+	* functionality. The default is Step 1.0. This should be called prior to
+	* any other emulation functions and after Init().
+	*
+	* Parameters:
+	*		stepID	0x10 for Step 1.0, 0x15 for Step 1.5, 0x20 for Step 2.0,
+	*				or 0x21 for Step 2.1. Anything else defaults to 1.0.
+	*/
+	void SetStep(int stepID);
+
+	/*
+	* Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes):
+	*
+	* One-time initialization of the context. Must be called before any other
+	* members (meaning it should be called even before being attached to any
+	* other objects that want to use it).
+	*
+	* External shader files are loaded according to configuration settings.
+	*
+	* Parameters:
+	*		xOffset		X offset of the viewable area within OpenGL display
+	*                  surface, in pixels.
+	*		yOffset		Y offset.
+	*		xRes		Horizontal resolution of the viewable area.
+	*		yRes		Vertical resolution.
+	*		totalXRes	Horizontal resolution of the complete display area.
+	*		totalYRes	Vertical resolution.
+	*
+	* Returns:
+	*		OKAY is successful, otherwise FAILED if a non-recoverable error
+	*		occurred. Any allocated memory will not be freed until the
+	*		destructor is called. Prints own error messages.
+	*/
+	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes);
+
+	/*
+	* CRender3D(void):
+	* ~CRender3D(void):
+	*
+	* Constructor and destructor.
+	*/
+	CNew3D(void);
+	~CNew3D(void);
+
+private:
+	/*
+	* Private Members
+	*/
+
+	// Real3D address translation
+	const UINT32 *TranslateCullingAddress(UINT32 addr);
+	const UINT32 *TranslateModelAddress(UINT32 addr);
+
+	// Matrix stack
+	void MultMatrix(UINT32 matrixOffset, Mat4& mat);
+	void InitMatrixStack(UINT32 matrixBaseAddr, Mat4& mat);
+
+	// Scene database traversal
+	bool DrawModel(UINT32 modelAddr);
+	void DescendCullingNode(UINT32 addr);
+	void DescendPointerList(UINT32 addr);
+	void DescendNodePtr(UINT32 nodeAddr);
+	void RenderViewport(UINT32 addr, int pri);
+
+	// building the scene
+	void CacheModel(Model *m, const UINT32 *data);
+	void CopyVertexData(R3DPoly& r3dPoly, std::vector<Poly>& polyArray);
+
+	enum class AlphaType{ none, poly, texture };
+	void RenderScene(int priority, bool alpha);
+	bool ClockWiseWinding();	// calculate winding with current matrix
+
+	/*
+	* Data
+	*/
+
+	// Stepping
+	int		m_step;
+	int		m_offset;			// offset to subtract for words 3 and higher of culling nodes
+	float	m_vertexFactor;		// fixed-point conversion factor for vertices
+
+	// Memory (passed from outside)
+	const UINT32	*m_cullingRAMLo;	// 4 MB
+	const UINT32	*m_cullingRAMHi;	// 1 MB
+	const UINT32	*m_polyRAM;			// 4 MB
+	const UINT32	*m_vrom;			// 64 MB
+	const UINT16	*m_textureRAM;		// 8 MB
+
+	// Resolution and scaling factors (to support resolutions higher than 496x384) and offsets
+	float		m_xRatio, m_yRatio;
+	unsigned	m_xOffs, m_yOffs;
+	unsigned 	m_totalXRes, m_totalYRes;
+
+	// Real3D Base Matrix Pointer
+	const float	*m_matrixBasePtr;
+
+	TextureSheet	m_texSheet;
+	NodeAttributes	m_nodeAttribs;
+	Mat4			m_modelMat;		// current modelview matrix
+	int				m_listDepth;
+
+	std::vector<Node> m_nodes;		// build the scene
+	std::vector<Poly> m_polyBuffer;	// we actually hold the vertex data here, one buffer to send to opengl, instead of 2000+ small ones.
+
+	VBO m_vboDynamic;				// dynamic data from poly ram, rom polys can go in a different buffer
+
+	R3DShader m_r3dShader;
+	int m_currentVPPriority;
+};
+
+} // New3D
+
+#endif  // INCLUDED_NEW3D_H

Src/Graphics/New3D/OglStateCache.cpp

@@ -0,0 +1,2 @@
+namespace New3D {
+} // New3D

Src/Graphics/New3D/OglStateCache.h

@@ -0,0 +1,2 @@
+namespace New3D {
+} // New3D

Src/Graphics/New3D/PolyHeader.cpp

@@ -0,0 +1,320 @@
+#include "Supermodel.h"
+#include "PolyHeader.h"
+
+namespace New3D {
+
+PolyHeader::PolyHeader()
+{
+	header = NULL;
+}
+
+PolyHeader::PolyHeader(UINT32* h) 
+{
+	header = h;
+}
+
+void PolyHeader::operator = (const UINT32* h) 
+{
+	header = (UINT32*)h;
+}
+
+UINT32* PolyHeader::StartOfData()
+{
+	return header + 7;	// 7 is size of header in bytes, data immediately follows
+}
+
+bool PolyHeader::NextPoly()
+{
+	if (LastPoly()) {
+		return false;
+	}
+
+	header += 7 + (NumVerts() - NumSharedVerts()) * 4;
+
+	return true;
+}
+
+int	PolyHeader::NumPolysTotal()
+{
+	UINT32* start = header;	// save start address
+	int count = 1;
+
+	while (NextPoly()) {
+		count++;
+	}
+
+	header = start;			// restore start address
+
+	return count;
+}
+
+int	PolyHeader::NumTrianglesTotal()
+{
+	if (header[6] == 0) {
+		return 0;			// no poly data
+	}
+
+	UINT32* start = header;	// save start address
+
+	int count = (NumVerts() == 4) ? 2 : 1;
+
+	while (NextPoly()) {
+		count += (NumVerts() == 4) ? 2 : 1;
+	}
+
+	header = start;			// restore start address
+
+	return count;
+}
+
+//
+//  header 0
+//
+
+bool PolyHeader::Specular()
+{
+	return (header[0] & 0x800000000) > 0;
+}
+
+int PolyHeader::PolyNumber() 
+{
+	return (header[0] & 0x000FFFC00) >> 10;				// not all programs pass this, instead they are set to 0
+}
+
+bool PolyHeader::Disabled()
+{
+	if ((header[0] & 0x100) && (header[0] & 0x200)) {	// assuming these two bits mean z and colour writes are disabled
+		return true;
+	}
+
+	return false;
+}
+
+int	PolyHeader::NumVerts() 
+{
+	return (header[0] & 0x40) ? 4 : 3;
+}
+
+int PolyHeader::NumSharedVerts()
+{
+	int num = 0;
+
+	for (int i = 0; i < 4; i++) {
+		if (SharedVertex(i)) {
+			num++;
+		}
+	}
+
+	return num;
+}
+
+bool PolyHeader::SharedVertex(int vertex)
+{
+	UINT32 mask = 1 << vertex;
+
+	return (header[0] & mask) > 0;
+}
+
+//
+//  header 1
+//
+
+void PolyHeader::FaceNormal(float n[3]) 
+{
+	n[0] = (float)(((INT32)header[1]) >> 8) * (1.0f / 4194304.0f);
+	n[1] = (float)(((INT32)header[2]) >> 8) * (1.0f / 4194304.0f);
+	n[2] = (float)(((INT32)header[3]) >> 8) * (1.0f / 4194304.0f);
+}
+
+float PolyHeader::UVScale()
+{
+	return (header[1] & 0x40) ? 1.0f : (1.0f / 8.0f);
+}
+
+bool PolyHeader::DoubleSided()
+{
+	return (header[1] & 0x10) ? true : false;
+}
+
+bool PolyHeader::LastPoly() 
+{
+	if ((header[1] & 4) > 0 || header[6] == 0) {
+		return true;
+	}
+
+	return false;
+}
+
+bool PolyHeader::PolyColor()
+{
+	return (header[1] & 2) > 0;
+}
+
+//
+// header 2
+//
+
+bool PolyHeader::TexUMirror()
+{
+	return (header[2] & 2) > 0;
+}
+
+bool PolyHeader::TexVMirror()
+{
+	return (header[2] & 1) > 0;
+}
+
+//
+// header 3
+
+int	PolyHeader::TexWidth()
+{
+	return 32 << ((header[3] >> 3) & 7);
+}
+int	PolyHeader::TexHeight()
+{
+	return 32 << ((header[3] >> 0) & 7);
+}
+
+//
+// header 4
+//
+
+void PolyHeader::Color(UINT8& r, UINT8& g, UINT8& b)
+{
+	r = (header[4] >> 24);
+	g = (header[4] >> 16) & 0xFF;
+	b = (header[4] >> 8) & 0xFF;
+}
+
+int	PolyHeader::Page()
+{
+	return (header[4] & 0x40) >> 6;
+}
+
+//
+// header 5
+//
+
+int PolyHeader::X(int textureXOffset)
+{
+	//====
+	int x;
+	//====
+
+	x = (32 * (((header[4] & 0x1F) << 1) | ((header[5] >> 7) & 1))) + textureXOffset;
+	x &= 2047;
+	return x;
+}
+
+int PolyHeader::Y(int textureYOffset)
+{
+	//=======
+	int y;
+	int page;
+	//=======
+
+	if (Page()) {
+		page = 1024;
+	}
+	else {
+		page = 0;
+	}
+
+	y = (32 * (header[5] & 0x1F) + page) + textureYOffset;	// if we hit 2nd page add 1024 to y coordinate
+	y &= 2047;
+
+	return y;
+}
+
+//
+// header 6
+//
+
+int	PolyHeader::TexFormat()
+{
+	return (header[6] >> 7) & 7;
+}
+
+bool PolyHeader::TexEnabled()
+{
+	return (header[6] & 0x04000000) > 0;
+}
+
+bool PolyHeader::LightEnabled()
+{
+	return !(header[6] & 0x00010000);
+}
+
+bool PolyHeader::AlphaTest()
+{
+	return (header[6] & 0x80000000) > 0;
+}
+
+UINT8 PolyHeader::Transparency()
+{
+	return (UINT8)(((header[6] >> 18) & 0x1F)  * 255.f / 0x1F);	
+}
+
+bool PolyHeader::FixedShading()
+{
+	return (header[6] & 0x2000000) > 0;
+}
+
+UINT8 PolyHeader::ShadeValue()
+{
+	return (UINT8)(((header[6] >> 26) & 0x1F) * (255.f / 0x1F));
+}
+
+bool PolyHeader::PolyAlpha()
+{
+	return (header[6] & 0x00800000) == 0;
+}
+
+bool PolyHeader::TextureAlpha()
+{
+	return (header[6] & 0x1);
+}
+
+bool PolyHeader::StencilPoly()
+{
+	return (header[6] & 1000000) > 0;
+}
+
+bool PolyHeader::Luminous()
+{
+	return (header[6] & 0x00010000) > 0;
+}
+
+float PolyHeader::LightModifier()
+{
+	return (float)((header[6] >> 11) & 0x1F) * (1.0f / 31.0f);
+}
+
+//
+// misc
+//
+
+UINT64 PolyHeader::Hash(int textureXOffset, int textureYOffset)
+{
+	UINT64 hash = 0;
+
+	hash |= (header[2] & 3);					// bits 0-1 uv mirror bits
+	hash |= ((header[3] >> 0) & 7) << 2;		// bits 2-4 tex height
+	hash |= ((header[3] >> 3) & 7) << 5;		// bits 5-7 tex width
+	hash |= X(textureXOffset) << 8;				// bits 8-17 x offset
+	hash |= Y(textureYOffset) << 18;			// bits 18-27 y offset
+	hash |= TexFormat() << 28;					// bits 28-30 tex format
+	hash |= (UINT64)TexEnabled() << 31;			// bits 31 textures enabled
+	hash |= (UINT64)LightEnabled() << 32;		// bits 32 light enabled
+	hash |= (UINT64)DoubleSided() << 33;		// bits 33 double sided
+	hash |= (UINT64)AlphaTest() << 34;			// bits 34 contour processing
+	hash |= (UINT64)PolyAlpha() << 35;			// bits 35 poly alpha processing
+	hash |= (UINT64)TextureAlpha() << 36;		// bits 35 poly alpha processing
+
+	//to do add the rest of the states
+
+	return hash;
+}
+
+} // New3D

Src/Graphics/New3D/PolyHeader.h

@@ -0,0 +1,139 @@
+#ifndef _POLY_HEADER_H_
+#define _POLY_HEADER_H_
+
+namespace New3D {
+
+/*
+Polygon Data
+
+0x00:
+x------- -------- -------- --------		Specular enable
+-xxxxx-- -------- -------- --------		?
+------xx xxxxxxxx xxxxxx-- --------		Polygon number (not always present)
+-------- -------- ------xx --------		Possibly disable z and colour writing
+-------- -------- -------- -x------		0 = Triangle, 1 = Quad
+-------- -------- -------- ----x---		Vertex 3 shared from previous polygon
+-------- -------- -------- -----x--		Vertex 2 shared from previous polygon
+-------- -------- -------- ------x-		Vertex 1 shared from previous polygon
+-------- -------- -------- -------x		Vertex 0 shared from previous polygon
+-------- -------- -------- x-xx----		?
+
+0x01: 
+xxxxxxxx xxxxxxxx xxxxxxxx--------		Polygon normal X coordinate(2.22 fixed point)
+-------- -------- -------- -x------		UV scale (0 = 13.3, 1 = 16.0)
+-------- -------- -------- ---x----		1 = Double-sided polygon
+-------- -------- -------- -----x--		If set, this is the last polygon
+-------- -------- -------- ------x-		Poly color, 1 = RGB, 0 = color table
+-------- -------- -------- x-x-x--x		?
+
+0x02: 
+xxxxxxxx xxxxxxxx xxxxxxxx --------		Polygon normal Y coordinate(2.22 fixed point)
+-------- -------- -------- ------x-		Texture U mirror enable
+-------- -------- -------- -------x		Texture V mirror enable
+-------- -------- -------- xxxxxx--		H/W also supports texture clamp/mirror, so guessing some of these bits must be that
+
+0x03: 
+xxxxxxxx xxxxxxxx xxxxxxxx --------		Polygon normal Z coordinate(2.22 fixed point)
+-------- -------- -------- --xxx---		Texture width(in 8 - pixel tiles)
+-------- -------- -------- ---- - xxx	Texture height(in 8 - pixel tiles)
+
+0x04: 
+xxxxxxxx xxxxxxxx xxxxxxxx --------		Color(RGB888)
+-------- -------- -------- -x------		Texture page
+-------- -------- -------- ---xxxxx		Upper 5 bits of texture U coordinate
+-------- -------- -------- x-------		?
+-------- -------- -------- --x-----		?
+
+0x05 : 
+xxxxxxxx xxxxxxxx xxxxxxxx --------		Specular color ?
+-------- -------- -------- x-------		Low bit of texture U coordinate
+-------- -------- -------- ---xxxxx		Low 5 bits of texture V coordinate
+-------- -------- -------- -xx-----		?
+
+0x06: 
+x------- -------- -------- --------		Alpha testing / contour
+-xxxxx-- -------- -------- --------		Fixed shading ?
+------x- -------- -------- --------		Enable fixed shading ?
+-------x -------- -------- --------		Possible stencil
+-------- x------- -------- --------		1 = disable transparency ?
+-------- -xxxxx-- -------- --------		Polygon translucency(0 = fully transparent)
+-------- -------x -------- --------		1 = disable lighting
+-------- -------- xxxxx--- --------		Polygon light modifier(Amount that a luminous polygon will burn through fog. Valid range is 0.0 to 1.0. 0.0 is completely fogged; 1.0 has no fog.)
+-------- -------- -----x-- --------		Texture enable
+-------- -------- ------xx x-------		Texture format
+-------- -------- -------- -------x		Alpha enable ?
+-------- ------x - ------- --------		Never seen set ?
+-------- -------- -------- -----xx-		Always set ?
+-------- -------- -------- -xxxx---		?
+*/
+
+class PolyHeader
+{
+public:
+	PolyHeader();
+	PolyHeader(UINT32* h);
+
+	void operator =(const UINT32* h);
+
+	UINT32*	StartOfData();
+	bool	NextPoly();
+	int		NumPolysTotal();		// could be quads or triangles
+	int		NumTrianglesTotal();
+
+	//header 0
+	bool	Specular();
+	int		PolyNumber();
+	bool	Disabled();		// z & colour disabled
+	int		NumVerts();
+	int		NumSharedVerts();
+	bool	SharedVertex(int vertex);
+
+	//header 1
+	void	FaceNormal(float n[3]);
+	float	UVScale();
+	bool	DoubleSided();
+	bool	LastPoly();
+	bool	PolyColor();	// if false uses LUT from ram
+
+	//header 2
+	bool	TexUMirror();
+	bool	TexVMirror();
+
+	// header 3
+	int		TexWidth();
+	int		TexHeight();
+
+	//header 4
+	void	Color(UINT8& r, UINT8& g, UINT8& b);
+	int		Page();
+
+	// header 5
+	int		X(int textureXOffset);
+	int		Y(int textureYOffset);
+
+	//header 6
+	int		TexFormat();
+	bool	TexEnabled();
+	bool	LightEnabled();
+	bool	AlphaTest();
+	UINT8	Transparency();			// 0-255
+	bool	FixedShading();
+	UINT8	ShadeValue();
+	bool	PolyAlpha();
+	bool	TextureAlpha();
+	bool	StencilPoly();
+	bool	Luminous();
+	float	LightModifier();
+
+	// misc
+	UINT64	Hash(int textureXOffset, int textureYOffset);		// make a unique hash for sorting by state
+
+
+	//=============
+	UINT32* header;
+	//=============
+};
+
+} // New3D
+
+#endif

Src/Graphics/New3D/R3DShader.cpp

@@ -0,0 +1,263 @@
+#include "R3DShader.h"
+#include "Graphics/Shader.h"
+
+namespace New3D {
+
+static char *vertexShaderBasic =
+
+// uniforms
+"uniform float	fogIntensity;\n"
+"uniform float	fogDensity;\n"
+"uniform float	fogStart;\n"
+
+//outputs to fragment shader
+"varying float	fsFogFactor;\n"
+"varying float	fsSpecularTerm;\n"		// specular light term (additive)
+"varying float	fsViewZ;\n"
+"varying vec3	fsViewNormal;\n"		// per vertex normal vector
+
+"void main(void)\n"
+"{\n"
+	"vec3	viewVertex;\n"
+
+	"viewVertex		= vec3(gl_ModelViewMatrix * gl_Vertex);\n"
+	"fsViewNormal	= normalize(gl_NormalMatrix*gl_Normal);\n"
+	"float z		= length(viewVertex);\n"
+	"fsFogFactor	= clamp(1.0 - fogIntensity*(fogStart + z*fogDensity), 0.0, 1.0);\n"
+	"fsViewZ		= -viewVertex.z;\n"	// convert Z from GL->Real3D convention (want +Z to be further into screen)
+
+	"gl_FrontColor	= gl_Color;\n"
+	"gl_TexCoord[0]	= gl_MultiTexCoord0;\n"
+	"gl_Position	= gl_ModelViewProjectionMatrix * gl_Vertex;\n"
+"}\n";
+
+static char *fragmentShaderBasic =
+
+"uniform sampler2D tex;\n"
+"uniform int	textureEnabled;\n"
+"uniform int	alphaTest;\n"
+"uniform int	textureAlpha;\n"
+"uniform vec3	fogColour;\n"
+"uniform vec4	spotEllipse;\n"		// spotlight ellipse position: .x=X position (screen coordinates), .y=Y position, .z=half-width, .w=half-height)
+"uniform vec2	spotRange;\n"			// spotlight Z range: .x=start (viewspace coordinates), .y=limit
+"uniform vec3	spotColor;\n"			// spotlight RGB color
+"uniform vec3	lighting[2];\n"		// lighting state (lighting[0] = sun direction, lighting[1].x,y = diffuse, ambient intensities from 0-1.0)
+"uniform int	lightEnable;\n"		// lighting enabled (1.0) or luminous (0.0), drawn at full intensity
+"uniform float	shininess;\n"		// specular shininess (if >= 0.0) or disable specular lighting (negative)
+
+//interpolated inputs from vertex shader
+"varying float	fsFogFactor;\n"
+"varying float	fsSpecularTerm;\n"		// specular light term (additive)
+"varying float	fsViewZ;\n"
+"varying vec3	fsViewNormal;\n"		// per vertex normal vector
+
+"void main()\n"
+"{\n"
+	"vec4 texData;\n"
+	"vec4 colData;\n"
+	"vec4 finalData;\n"
+
+	"texData = vec4(1.0, 1.0, 1.0, 1.0);\n"
+
+	"if(textureEnabled==1) {\n"
+		"texData = texture2D( tex, gl_TexCoord[0].st);\n"
+
+		"if (alphaTest==1) {\n"			// does it make any sense to do this later?
+			"if (texData.a < (8.0/16.0)) {\n"
+				"discard;\n"
+			"}\n"
+		"}\n"
+
+		"if (textureAlpha == 0) {\n"
+			"texData.a = 1.0;\n"
+		"}\n"
+	"}\n"
+
+	"colData =  gl_Color;\n"
+
+	"finalData = texData * colData;\n"
+	"if (finalData.a < (1.0/16.0)) {\n"		// basically chuck out any totally transparent pixels value = 1/16 the smallest transparency level h/w supports
+		"discard;\n"
+	"}\n"
+
+	"vec3 lightIntensity;\n"
+
+	"if (lightEnable==1)\n"
+	"{\n"
+		"vec3	sunVector;\n"		// sun lighting vector (as reflecting away from vertex)
+		"float	sunFactor;\n"		// sun light projection along vertex normal (0.0 to 1.0)
+
+		// Real3D -> OpenGL view space convention (TO-DO: do this outside of shader)
+		"sunVector = lighting[0] * vec3(1.0, -1.0, -1.0);\n"
+
+		// Compute diffuse factor for sunlight
+		"sunFactor = max(dot(sunVector, fsViewNormal), 0.0);\n"
+
+		// Total light intensity: sum of all components
+		"lightIntensity = vec3(sunFactor*lighting[1].x + lighting[1].y);\n"
+		"lightIntensity = clamp(lightIntensity,0.0,1.0);\n"
+	"}\n"
+	"else {\n"
+			"lightIntensity = vec3(1.0,1.0,1.0);\n"
+	"}\n"
+
+	"finalData.rgb *= lightIntensity;\n"
+
+	/*
+	"vec2	ellipse;\n"
+	"vec3	lightIntensity;\n"
+	"float	insideSpot;\n"
+
+	// Compute spotlight and apply lighting
+	"ellipse	= (gl_FragCoord.xy - spotEllipse.xy) / spotEllipse.zw;\n"
+	"insideSpot = dot(ellipse, ellipse);\n"
+
+	"if ((insideSpot <= 1.0) && (fsViewZ >= spotRange.x) && (fsViewZ<spotRange.y)) {\n"
+		"lightIntensity = fsLightIntensity + (1.0 - insideSpot)*spotColor;\n"
+	"}\n"
+	"else {\n"
+		"lightIntensity = fsLightIntensity;\n"
+	"}\n"
+
+	"finalData.rgb *= lightIntensity;\n"
+	"finalData.rgb += vec3(fsSpecularTerm, fsSpecularTerm, fsSpecularTerm);\n"
+	*/
+
+	"finalData.rgb = mix(fogColour, finalData.rgb, fsFogFactor);\n"
+
+	"gl_FragColor = finalData;\n"
+"}\n";
+
+R3DShader::R3DShader()
+{
+	m_shaderProgram		= 0;
+	m_vertexShader		= 0;
+	m_fragmentShader	= 0;
+
+	Start();	// reset attributes
+}
+
+void R3DShader::Start()
+{
+	m_textured		= false;
+	m_textureAlpha	= false;		// use alpha in texture
+	m_alphaTest		= false;		// discard fragment based on alpha (ogl does this with fixed function)
+	m_doubleSided	= false;
+
+	m_dirty			= true;			// dirty means all the above are dirty, ie first run
+}
+
+bool R3DShader::LoadShader(const char* vertexShader, const char* fragmentShader)
+{
+	const char* vShader;
+	const char* fShader;
+	bool success;
+
+	if (vertexShader) {
+		vShader = vertexShader;
+	}
+	else {
+		vShader = vertexShaderBasic;
+	}
+
+	if (fragmentShader) {
+		fShader = fragmentShader;
+	}
+	else {
+		fShader = fragmentShaderBasic;
+	}
+
+	success = LoadShaderProgram(&m_shaderProgram, &m_vertexShader, &m_fragmentShader, nullptr, nullptr, vShader, fShader);
+
+	m_locTexture		= glGetUniformLocation(m_shaderProgram, "tex");
+	m_locTextureEnabled	= glGetUniformLocation(m_shaderProgram, "textureEnabled");
+	m_locTextureAlpha	= glGetUniformLocation(m_shaderProgram, "textureAlpha");
+	m_locAlphaTest		= glGetUniformLocation(m_shaderProgram, "alphaTest");
+
+	m_locFogIntensity	= glGetUniformLocation(m_shaderProgram, "fogIntensity");
+	m_locFogDensity		= glGetUniformLocation(m_shaderProgram, "fogDensity");
+	m_locFogStart		= glGetUniformLocation(m_shaderProgram, "fogStart");
+	m_locFogColour		= glGetUniformLocation(m_shaderProgram, "fogColour");
+
+	m_locLighting		= glGetUniformLocation(m_shaderProgram, "lighting");
+	m_locLightEnable	= glGetUniformLocation(m_shaderProgram, "lightEnable");
+	m_locShininess		= glGetUniformLocation(m_shaderProgram, "shininess");
+	m_locSpotEllipse	= glGetUniformLocation(m_shaderProgram, "spotEllipse");
+	m_locSpotRange		= glGetUniformLocation(m_shaderProgram, "spotRange");
+	m_locSpotColor		= glGetUniformLocation(m_shaderProgram, "spotColor");
+
+	return success;
+}
+
+void R3DShader::SetShader(bool enable)
+{
+	if (enable) {
+		glUseProgram(m_shaderProgram);
+		Start();
+	}
+	else {
+		glUseProgram(0);
+	}
+}
+
+void R3DShader::SetMeshUniforms(const Mesh* m)
+{
+	if (m == nullptr) {
+		return;			// sanity check
+	}
+
+	if (m_dirty) {
+		glUniform1i(m_locTexture, 0);
+	}
+
+	if (m_dirty || m->textured != m_textured) {
+		glUniform1i(m_locTextureEnabled, m->textured);
+		m_textured = m->textured;
+	}
+
+	if (m_dirty || m->alphaTest != m_alphaTest) {
+		glUniform1i(m_locAlphaTest, m->alphaTest);
+		m_alphaTest = m->alphaTest;
+	}
+
+	if (m_dirty || m->textureAlpha != m_textureAlpha) {
+		glUniform1i(m_locTextureAlpha, m->textureAlpha);
+		m_textureAlpha = m->textureAlpha;
+	}
+
+	if (m_dirty || m->fogIntensity != m_fogIntensity) {
+		glUniform1f(m_locFogIntensity, m->fogIntensity);
+		m_fogIntensity = m->fogIntensity;
+	}
+
+	glUniform1i(m_locLightEnable, m->lighting);
+	glUniform1f(m_locShininess, 1);
+
+	// technically not uniforms
+	if (m_dirty || m->doubleSided != m_doubleSided) {
+		m_doubleSided = m->doubleSided;
+		if (m_doubleSided) {
+			glDisable(GL_CULL_FACE);
+		}
+		else {
+			glEnable(GL_CULL_FACE);
+		}
+	}
+
+	m_dirty = false;
+}
+
+void R3DShader::SetViewportUniforms(const Viewport *vp)
+{	
+	//didn't bother caching these, they don't get frequently called anyway
+	glUniform1f(m_locFogDensity, vp->fogParams[3]);
+	glUniform1f(m_locFogStart, vp->fogParams[4]);
+	glUniform3fv(m_locFogColour, 1, vp->fogParams);
+
+	glUniform3fv(m_locLighting, 2, vp->lightingParams);
+	glUniform4fv(m_locSpotEllipse, 1, vp->spotEllipse);
+	glUniform2fv(m_locSpotRange, 1, vp->spotRange);
+	glUniform3fv(m_locSpotColor, 1, vp->spotColor);
+}
+
+} // New3D

Src/Graphics/New3D/R3DShader.h

@@ -0,0 +1,57 @@
+#ifndef _R3DSHADER_H_
+#define _R3DSHADER_H_
+
+#include "Pkgs/glew.h"
+#include "Model.h"
+
+namespace New3D {
+
+class R3DShader
+{
+public:
+	R3DShader();
+
+	bool LoadShader(const char* vertexShader = nullptr, const char* fragmentShader = nullptr);
+	void SetMeshUniforms(const Mesh* m);
+	void SetViewportUniforms(const Viewport *vp);
+	void Start();
+	void SetShader(bool enable = true);
+
+private:
+
+	GLuint m_shaderProgram;
+	GLuint m_vertexShader;
+	GLuint m_fragmentShader;
+
+	// mesh uniform data
+	GLint m_locTexture;
+	GLint m_locTextureEnabled;
+	GLint m_locTextureAlpha;
+	GLint m_locAlphaTest;
+
+	bool m_textured;
+	bool m_textureAlpha;	// use alpha in texture
+	bool m_alphaTest;		// discard fragment based on alpha (ogl does this with fixed function)
+	float m_fogIntensity;
+	bool m_doubleSided;
+
+	bool m_dirty;
+
+	// viewport uniform data
+	GLint m_locFogIntensity;
+	GLint m_locFogDensity;
+	GLint m_locFogStart;
+	GLint m_locFogColour;
+
+	// lighting
+	GLint m_locLighting;
+	GLint m_locLightEnable;
+	GLint m_locShininess;
+	GLint m_locSpotEllipse;
+	GLint m_locSpotRange;
+	GLint m_locSpotColor;
+};
+
+} // New3D
+
+#endif

Src/Graphics/New3D/Texture.cpp

@@ -0,0 +1,264 @@
+#include "Texture.h"
+#include <stdio.h>
+#include <math.h>
+
+namespace New3D {
+
+Texture::Texture()
+{
+	Reset();
+}
+
+Texture::~Texture()
+{
+	DeleteTexture();		// make sure to have valid context before destroying
+}
+
+void Texture::DeleteTexture()
+{
+	if (m_textureID) {
+		glDeleteTextures(1, &m_textureID);
+		printf("-----> deleting %i %i %i %i %i\n", m_format, m_x, m_y, m_width, m_height);
+		Reset();
+	}
+}
+
+void Texture::Reset()
+{
+	m_x = 0;
+	m_y = 0;
+	m_width = 0;
+	m_height = 0;
+	m_format = 0;
+	m_textureID = 0;
+	m_mirrorU = false;
+	m_mirrorV = false;
+}
+
+void Texture::BindTexture()
+{
+	glBindTexture(GL_TEXTURE_2D, m_textureID);
+}
+
+void Texture::GetCoordinates(UINT16 uIn, UINT16 vIn, float uvScale, float& uOut, float& vOut)
+{
+	uOut = ((uIn*uvScale)+0.0f) / m_width;
+	vOut = ((vIn*uvScale)+0.0f) / m_height;
+}
+
+void Texture::SetWrapMode(bool mirrorU, bool mirrorV)
+{
+	if (mirrorU != m_mirrorU) {
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, mirrorU ? GL_MIRRORED_REPEAT : GL_REPEAT);
+		m_mirrorU = mirrorU;
+	}
+
+	if (mirrorV != m_mirrorV) {
+		glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, mirrorV ? GL_MIRRORED_REPEAT : GL_REPEAT);
+		m_mirrorV = mirrorV;
+	}
+}
+
+UINT32 Texture::UploadTexture(const UINT16* src, UINT8* scratch, int format, bool mirrorU, bool mirrorV, int x, int y, int width, int height)
+{
+	int		xi, yi, i;
+	GLubyte	texel;
+	GLubyte	c, a;
+
+	if (!src || !scratch) {
+		return 0;		// sanity checking
+	}
+
+	DeleteTexture();	// free any existing texture
+
+	i = 0;
+
+	switch (format)
+	{
+	default:	// Debug texture, use TEXTURE_DEBUG mask
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				scratch[i++] = 255;	// R
+				scratch[i++] = 0;	// G
+				scratch[i++] = 0;	// B
+				scratch[i++] = 255;	// A
+			}
+		}
+		break;
+
+	case 0:	// T1RGB5 <- correct
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				scratch[i++] = (GLubyte)(((src[yi * 2048 + xi] >> 10) & 0x1F) * 255.f / 0x1F);	// R
+				scratch[i++] = (GLubyte)(((src[yi * 2048 + xi] >> 5) & 0x1F) * 255.f / 0x1F);	// G
+				scratch[i++] = (GLubyte)(((src[yi * 2048 + xi] >> 0) & 0x1F) * 255.f / 0x1F);	// B
+				scratch[i++] = ((src[yi * 2048 + xi] & 0x8000) ? 0 : 255);						// T
+			}
+		}
+		break;
+
+	case 1:	// Interleaved A4L4 (low byte)
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				// Interpret as A4L4
+				texel = src[yi * 2048 + xi] & 0xFF;
+				c = (texel & 0xF) * 17;
+				a = (texel >> 4) * 17;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = a;
+			}
+		}
+		break;
+
+	case 2:	// luminance alpha texture <- this one is correct
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				texel = src[yi * 2048 + xi] & 0xFF;
+				c = ((texel >> 4) & 0xF) * 17;
+				a = (texel & 0xF) * 17;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = a;
+			}
+		}
+		break;
+
+	case 3:	// Interleaved A4L4 (high byte)
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				texel = src[yi * 2048 + xi] >> 8;
+				c = (texel & 0xF) * 17;
+				a = (texel >> 4) * 17;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = a;
+			}
+		}
+		break;
+
+	case 4: // 8-bit, L4A4
+
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				texel = src[yi * 2048 + xi] & 0xFF;
+				c = ((texel >> 4) & 0xF) * 17;	// seems to work better in Lost World (raptor shadows)
+				a = (texel & 0xF) * 17;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = c;
+				scratch[i++] = a;
+			}
+		}
+		break;
+
+	case 5:	// 8-bit grayscale
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				// Interpret as 8-bit grayscale
+				texel = src[yi * 2048 + xi] & 0xFF;
+
+				scratch[i++] = texel;
+				scratch[i++] = texel;
+				scratch[i++] = texel;
+				scratch[i++] = 255;
+			}
+		}
+		break;
+
+	case 6:	// 8-bit grayscale <-- this one is correct
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				texel = src[yi * 2048 + xi] >> 8;
+
+				scratch[i++] = texel;
+				scratch[i++] = texel;
+				scratch[i++] = texel;
+				scratch[i++] = 255;
+			}
+		}
+		break;
+
+	case 7:	// RGBA4
+		for (yi = y; yi < (y + height); yi++)
+		{
+			for (xi = x; xi < (x + width); xi++)
+			{
+				scratch[i++] = ((src[yi * 2048 + xi] >> 12) & 0xF) * 17;// R
+				scratch[i++] = ((src[yi * 2048 + xi] >> 8) & 0xF) * 17;	// G
+				scratch[i++] = ((src[yi * 2048 + xi] >> 4) & 0xF) * 17;	// B
+				scratch[i++] = ((src[yi * 2048 + xi] >> 0) & 0xF) * 17;	// A
+			}
+		}
+		break;
+	}
+
+	//remove debug mask
+	format &= 7;
+
+	GLfloat maxAnistrophy;
+
+	glGetFloatv(GL_MAX_TEXTURE_MAX_ANISOTROPY_EXT, &maxAnistrophy);
+
+	if (maxAnistrophy > 8) {
+		maxAnistrophy = 8.0f;	//anymore than 8 can get expensive for little gain
+	}
+
+	glPixelStorei(GL_UNPACK_ALIGNMENT, 4);	// rgba is always 4 byte aligned
+	glActiveTexture(GL_TEXTURE0);           // activate correct texture unit
+
+	glGenTextures(1, &m_textureID);
+	glBindTexture(GL_TEXTURE_2D, m_textureID);
+
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, mirrorU ? GL_MIRRORED_REPEAT : GL_REPEAT);	//todo this in shaders?
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, mirrorV ? GL_MIRRORED_REPEAT : GL_REPEAT);
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+	glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_MIN_FILTER, GL_LINEAR_MIPMAP_LINEAR);
+	glTexParameterf(GL_TEXTURE_2D, GL_TEXTURE_MAX_ANISOTROPY_EXT, maxAnistrophy);
+	glTexParameteri(GL_TEXTURE_2D, GL_GENERATE_MIPMAP, GL_TRUE);
+	glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, width, height, 0, GL_RGBA, GL_UNSIGNED_BYTE, scratch);
+
+	// assuming successful we can copy details
+
+	m_x = x;
+	m_y = y;
+	m_width = width;
+	m_height = height;
+	m_format = format;
+	m_mirrorU = mirrorU;
+	m_mirrorV = mirrorV;
+
+	printf("create format %i x: %i y: %i width: %i height: %i\n", format, x, y, width, height);
+
+	return m_textureID;
+}
+
+void Texture::GetDetails(int& x, int&y, int& width, int& height, int& format)
+{
+	x = m_x;
+	y = m_y;
+	width = m_width;
+	height = m_height;
+	format = m_format;
+}
+
+} // New3D

Src/Graphics/New3D/Texture.h

@@ -0,0 +1,42 @@
+#ifndef _TEXTURE_H_
+#define _TEXTURE_H_
+
+#include "Types.h"
+#include "Pkgs/glew.h"	//arg
+
+namespace New3D {
+  
+#define TEXTURE_DEBUG		0x8
+#define TEXTURE_DEBUG_MASK	0x7
+
+class Texture
+{
+public:
+
+	Texture();
+	~Texture();
+
+	UINT32	UploadTexture	(const UINT16* src, UINT8* scratch, int format, bool mirrorU, bool mirrorV, int x, int y, int width, int height);
+	void	DeleteTexture	();
+	void	BindTexture		();
+	void	GetCoordinates	(UINT16 uIn, UINT16 vIn, float uvScale, float& uOut, float& vOut);
+	void	GetDetails		(int& x, int&y, int& width, int& height, int& format);
+	void	SetWrapMode		(bool mirrorU, bool mirrorV);
+
+private:
+
+	void Reset();
+
+	int m_x;
+	int m_y;
+	int m_width;
+	int m_height;
+	int m_format;
+	bool m_mirrorU;
+	bool m_mirrorV;
+	GLuint m_textureID;
+};
+
+} // New3D
+
+#endif

Src/Graphics/New3D/TextureSheet.cpp

@@ -0,0 +1,104 @@
+#include "TextureSheet.h"
+
+namespace New3D {
+
+TextureSheet::TextureSheet()
+{
+	m_temp.resize(512 * 512 * 4);	// temporay buffer for textures
+}
+
+int TextureSheet::ToIndex(int x, int y)
+{
+	return (y * 2048) + x;
+}
+
+std::shared_ptr<Texture> TextureSheet::BindTexture(const UINT16* src, int format, bool mirrorU, bool mirrorV, int x, int y, int width, int height)
+{
+	//========
+	int	index;
+	//========
+
+	x &= 2047;
+	y &= 2047;
+
+	if ((x + width) > 2048 || (y + height) > 2048) {
+		return 0;
+	}
+
+	if (width > 512 || height > 512) {
+		return 0;
+	}
+
+	index = ToIndex(x, y);
+
+	if (m_texMap[format&TEXTURE_DEBUG_MASK].count(index) == 0) {
+
+		//no textures at this position or format so add it to the map
+
+		std::shared_ptr<Texture> t(new Texture());
+		m_texMap[format&TEXTURE_DEBUG_MASK].insert(std::pair<int, std::shared_ptr<Texture>>(index, t));
+		t->UploadTexture(src, m_temp.data(), format, mirrorU, mirrorV, x, y, width, height);
+		return t;
+	}
+	else {
+		//scan for duplicates
+		//only texture width/height and wrap modes can change here. Since key is based on x/y pos, and each map is a separate format
+
+		auto range = m_texMap[format&TEXTURE_DEBUG_MASK].equal_range(index);
+
+		for (auto it = range.first; it != range.second; ++it) {
+
+			int x2, y2, width2, height2, format2;
+
+			it->second->GetDetails(x2, y2, width2, height2, format2);
+
+			if (width == width2 && height == height2) {
+				return it->second;
+			}
+		}
+
+		std::shared_ptr<Texture> t(new Texture());
+		m_texMap[format&TEXTURE_DEBUG_MASK].insert(std::pair<int, std::shared_ptr<Texture>>(index, t));
+		t->UploadTexture(src, m_temp.data(), format, mirrorU, mirrorV, x, y, width, height);
+
+		return t;
+	}
+}
+
+void TextureSheet::Release()
+{
+	for (int i = 0; i < 8; i++) {
+		m_texMap[i].clear();
+	}
+}
+
+void TextureSheet::Invalidate(int x, int y, int width, int height)
+{
+	//==========
+	int count;
+	int sWidth;		// sample width
+	int sHeight;	// sample height
+	//==========
+
+	// since the smallest sized texture is 32x32 pixels?
+	// we can invalidate 32x32 tiles over the width/height of the area
+
+	sWidth	= width / 32;
+	sHeight	= height / 32;
+	count	= sWidth * sHeight;
+
+	for (int i = 0; i < count; i++) {
+
+		int index = ToIndex(x + ((i%sWidth) * 32), y + ((i / sWidth) * 32));
+
+		for (int j = 0; j<8; j++) {
+
+			if (m_texMap[j].count(index) > 0) {
+
+				m_texMap[j].erase(index);
+			}
+		}
+	}
+}
+
+} // New3D

Src/Graphics/New3D/TextureSheet.h

@@ -0,0 +1,38 @@
+#ifndef _TEXTURE_SHEET_H_
+#define _TEXTURE_SHEET_H_
+
+#include "Types.h"
+#include <unordered_map>
+#include <vector>
+#include <memory>
+#include "Texture.h"
+
+namespace New3D {
+
+#define REAL_SHEET_WIDTH	2048
+#define REAL_SHEET_HEIGHT	2048
+
+class TextureSheet
+{
+public:
+	TextureSheet();
+
+	std::shared_ptr<Texture>	BindTexture		(const UINT16* src, int format, bool mirrorU, bool mirrorV, int x, int y, int width, int height);
+	void						Invalidate		(int x, int y, int width, int height); // release parts of the memory
+	void						Release			();		// release all texture objects and memory
+
+private:
+
+	int ToIndex(int x, int y);
+
+	std::unordered_multimap<int, std::shared_ptr<Texture>> m_texMap[8];
+
+	// the key for the above maps is the x/y position in the 2048x2048 texture
+	// array of 8 planes for each texture type
+
+	std::vector<UINT8> m_temp;
+};
+
+} // New3D
+
+#endif

Src/Graphics/New3D/VBO.cpp

@@ -0,0 +1,46 @@
+#include "VBO.h"
+
+namespace New3D {
+
+VBO::VBO()
+{
+	m_id = 0;
+	m_target = 0;
+}
+
+void VBO::Create(GLenum target, GLenum usage, GLsizeiptr size, const void* data)
+{
+	glGenBuffers(1, &m_id);							// create a vbo
+	glBindBuffer(target, m_id);						// activate vbo id to use
+	glBufferData(target, size, data, usage);		// upload data to video card
+
+	m_target = target;
+
+	Bind(false);		// unbind
+}
+
+void VBO::BufferSubData(GLintptr offset, GLsizeiptr size, const GLvoid* data)
+{
+	glBufferSubData(m_target, offset, size, data);
+}
+
+void VBO::Destroy()
+{
+	if (m_id) {
+		glDeleteBuffers(1, &m_id);
+		m_id = 0;
+		m_target = 0;
+	}
+}
+
+void VBO::Bind(bool enable)
+{
+	if (enable) {
+		glBindBuffer(m_target, m_id);
+	}
+	else {
+		glBindBuffer(m_target, 0);
+	}
+}
+
+} // New3D

Src/Graphics/New3D/VBO.h

@@ -0,0 +1,25 @@
+#ifndef _VBO_H_
+#define _VBO_H_
+
+#include "Pkgs/glew.h"
+
+namespace New3D {
+
+class VBO
+{
+public:
+	VBO();
+
+	void Create			(GLenum target, GLenum usage, GLsizeiptr size, const void* data=nullptr);
+	void BufferSubData	(GLintptr offset, GLsizeiptr size, const GLvoid* data);
+	void Destroy		();
+	void Bind			(bool enable);
+
+private:
+	GLuint m_id;
+	GLenum m_target;
+};
+
+} // New3D
+
+#endif

Src/Graphics/New3D/Vec.cpp

@@ -0,0 +1,237 @@
+#include <windows.h>
+#include "Vec.h"
+#include <Math.h>
+
+namespace New3D {
+
+static float fastSqrt(float number) {
+    long i;
+    float x, y;
+    const float f = 1.5F;
+
+    x = number * 0.5F;
+    y  = number;
+    i  = * ( long * ) &y;
+    i  = 0x5f375a86 - ( i >> 1 );
+    y  = * ( float * ) &i;
+    y  = y * ( f - ( x * y * y ) );
+    y  = y * ( f - ( x * y * y ) );
+    return number * y;
+}
+
+static float fastInvSqrt(float x)
+{
+	float xhalf = 0.5f*x;
+	int i = *(int*)&x;			// get bits for floating value
+	i = 0x5f375a86- (i>>1);		// gives initial guess y0
+	x = *(float*)&i;			// convert bits back to float
+	x = x*(1.5f-xhalf*x*x);		// Newton step, repeating increases accuracy
+	return x;
+}
+
+void V3::subtract(const Vec3 a, const Vec3 b, Vec3 out) {
+
+	out[0] = a[0] - b[0];
+	out[1] = a[1] - b[1];
+	out[2] = a[2] - b[2];
+}
+
+void V3::subtract(Vec3 a, const Vec3 b) {
+
+	a[0] -= b[0];
+	a[1] -= b[1];
+	a[2] -= b[2];
+}
+
+void V3::add(const Vec3 a, const Vec3 b, Vec3 out) {
+
+	out[0] = a[0] + b[0];
+	out[1] = a[1] + b[1];
+	out[2] = a[2] + b[2];
+}
+
+void V3::add(Vec3 a, const Vec3 b) {
+
+	a[0] += b[0];
+	a[1] += b[1];
+	a[2] += b[2];
+}
+
+void V3::divide(Vec3 a, float number) {
+
+	multiply(a,1/number);
+}
+
+void V3::multiply(Vec3 a, float number) {
+
+	a[0] *= number;
+	a[1] *= number;
+	a[2] *= number;
+}
+
+void V3::multiply(Vec3 a, const Vec3 b) {
+
+	a[0] *= b[0];
+	a[1] *= b[1];
+	a[2] *= b[2];
+}
+
+void V3::multiply(const Vec3 a, const Vec3 b, Vec3 out) {
+
+	out[0] = a[0] * b[0];
+	out[1] = a[1] * b[1];
+	out[2] = a[2] * b[2];
+}
+
+void V3::crossProduct(const Vec3 v1, const Vec3 v2, Vec3 cross) {
+									
+	cross[0] = v1[1]*v2[2] - v1[2]*v2[1];
+	cross[1] = v1[2]*v2[0] - v1[0]*v2[2];
+	cross[2] = v1[0]*v2[1] - v1[1]*v2[0];						
+}
+
+float V3::dotProduct(const Vec3 v1, const Vec3 v2) {
+
+	return v1[0]*v2[0] + v1[1]*v2[1] + v1[2]*v2[2];
+}
+
+void V3::copy(const Vec3 in, Vec3 out) {
+
+	out[0] = in[0];
+	out[1] = in[1];
+	out[2] = in[2];
+}
+
+void V3::inverse(Vec3 v) {
+
+	v[0] = -v[0];
+	v[1] = -v[1];
+	v[2] = -v[2];
+}
+
+float V3::length(const Vec3 v) {
+
+	//===========
+	float length;
+	//===========
+	
+	length = v[0]*v[0] + v[1]*v[1] + v[2]*v[2];
+	length = sqrt(length);
+
+	return length;
+}
+
+void V3::normalise(Vec3 v) {
+
+	//========
+	float len;
+	//========
+
+	len = v[0]*v[0] + v[1]*v[1] + v[2]*v[2];
+	len = fastInvSqrt(len);
+
+	multiply(v,len);
+}
+
+void V3::multiplyAdd(const Vec3 a, float scale, const Vec3 b, Vec3 out) {
+
+	out[0] = a[0] + scale*b[0];
+	out[1] = a[1] + scale*b[1];
+	out[2] = a[2] + scale*b[2];
+}
+
+void V3::reset(Vec3 v) {
+
+	v[0] = 0;
+	v[1] = 0;
+	v[2] = 0;
+}
+
+void V3::set(Vec3 v, float value) {
+
+	v[0] = value;
+	v[1] = value;
+	v[2] = value;
+}
+
+void V3::set(Vec3 v, float x, float y, float z) {
+
+	v[0] = x;
+	v[1] = y;
+	v[2] = z;
+}
+
+void V3::reflect(const Vec3 a, const Vec3 b, Vec3 out) {
+
+	//===========
+	float	temp;
+	Vec3	v;
+	//===========
+
+	//Vect2 = Vect1 - 2 * WallN * (WallN DOT Vect1)
+
+	V3::copy(a,v);
+
+	temp = V3::dotProduct(a,b) * 2;
+
+	V3::multiply(v,temp);
+	V3::subtract(b,v,out);
+}
+
+void V3::createNormal(const Vec3 a, const Vec3 b, const Vec3 c, Vec3 outNormal) {
+
+	//======
+	Vec3 v1;
+	Vec3 v2;
+	//======
+
+	V3::subtract	(a,b,v1);
+	V3::subtract	(c,b,v2);
+	V3::crossProduct(v1,v2,outNormal);
+}
+
+void V3::_max(Vec3 a, const Vec3 compare) {
+
+	if(a[0] < compare[0]) a[0] = compare[0];
+	if(a[1] < compare[1]) a[1] = compare[1];
+	if(a[2] < compare[2]) a[2] = compare[2];
+}
+
+void V3::_min(Vec3 a, const Vec3 compare) {
+
+	if(a[0] > compare[0]) a[0] = compare[0];
+	if(a[1] > compare[1]) a[1] = compare[1];
+	if(a[2] > compare[2]) a[2] = compare[2];
+}
+
+bool V3::cmp(const Vec3 a, float b) {
+
+	if(a[0]!=b) return false;
+	if(a[1]!=b) return false;
+	if(a[2]!=b) return false;
+
+	return true;
+}
+
+bool V3::cmp(const Vec3 a, const Vec3 b) {
+
+	if(a[0]!=b[0]) return false;
+	if(a[1]!=b[1]) return false;
+	if(a[2]!=b[2]) return false;
+
+	return true;
+}
+
+void V3::clamp(Vec3 a, float _min, float _max) {
+	
+	if(a[0] < _min) a[0] = _min;
+	if(a[0] > _max) a[0] = _max;
+
+	if(a[1] < _min) a[1] = _min;
+	if(a[1] > _max) a[1] = _max;
+
+	if(a[2] < _min) a[2] = _min;
+	if(a[2] > _max) a[2] = _max;
+}
+
+} // New3D

Src/Graphics/New3D/Vec.h

@@ -0,0 +1,37 @@
+#ifndef VEC_H
+#define VEC_H
+
+namespace New3D {
+namespace V3 
+{
+	typedef float Vec3[3];
+
+	void	subtract		(const Vec3 a, const Vec3 b, Vec3 out);
+	void	subtract		(Vec3 a, const Vec3 b);
+	void	add				(const Vec3 a, const Vec3 b, Vec3 out);
+	void	add				(Vec3 a, const Vec3 b);
+	void	divide			(Vec3 a, float number);
+	void	multiply		(Vec3 a, float number);
+	void	multiply		(Vec3 a, const Vec3 b);
+	void	multiply		(const Vec3 a, const Vec3 b, Vec3 out);
+	void	crossProduct	(const Vec3 v1, const Vec3 v2, Vec3 cross);
+	float	dotProduct		(const Vec3 v1, const Vec3 v2);
+	void	copy			(const Vec3 in, Vec3 out);
+	void	inverse			(Vec3 v);
+	float	length			(const Vec3 v);
+	void	normalise		(Vec3 v);
+	void	multiplyAdd		(const Vec3 a, float scale, const Vec3 b, Vec3 out);
+	void	reset			(Vec3 v);
+	void	set				(Vec3 v, float value);
+	void	set				(Vec3 v, float x, float y, float z);
+	void	reflect			(const Vec3 a, const Vec3 b, Vec3 out);
+	void	createNormal	(const Vec3 a, const Vec3 b, const Vec3 c, Vec3 outNormal);	// assume a,b,c are wound clockwise
+	void	_max			(Vec3 a, const Vec3 compare);
+	void	_min			(Vec3 a, const Vec3 compare);
+	bool	cmp				(const Vec3 a, float b);
+	bool	cmp				(const Vec3 a, const Vec3 b);
+	void	clamp			(Vec3 a, float _min, float _max);
+}
+} // New3D
+
+#endif

Src/Graphics/Render2D.cpp

@@ -710,7 +710,7 @@ void CRender2D::BeginFrame(void)
 
 // Top layers
 void CRender2D::EndFrame(void)
-{	
+{
 	// Display top surface
 	Setup2D(false, false);
 	glEnable(GL_BLEND);

Src/Model3/Real3D.cpp

@@ -107,6 +107,7 @@ void CReal3D::LoadState(CBlockFile *SaveState)
 	}
 	
 	SaveState->Read(memoryPool, MEM_POOL_SIZE_RW);
+
 	// If multi-threaded, update read-only snapshots too
 	if (g_Config.gpuMultiThreaded)
 		UpdateSnapshots(true);
@@ -706,7 +707,6 @@ void CReal3D::UploadTexture(UINT32 header, UINT16 *texData)
 	unsigned	x, y, page, width, height, bytesPerTexel,
 				mipYPos, mipWidth, mipHeight, mipNum, mipX, mipY;
 	
-	
 	// Position: texture RAM is arranged as 2 2048x1024 texel sheets
 	x = 32*(header&0x3F);
 	y = 32*((header>>7)&0x1F);
@@ -855,7 +855,7 @@ void CReal3D::WriteTexturePort(unsigned reg, UINT32 data)
 	case 0x8:	// VROM texture length (also used to trigger uploads)
 	case 0x14:
 		UploadTexture(vromTextureHeader,(UINT16 *)&vrom[vromTextureAddr&0xFFFFFF]);
-		//printf("texture upload: addr=%08X\n", vromTextureAddr);
+    //printf("texture upload: addr=%08X\n", vromTextureAddr);
 		break;
 	default:
 		DebugLog("Real3D texture port write: %X=%08X\n", reg, data);
@@ -969,7 +969,7 @@ void CReal3D::Reset(void)
  Configuration, Initialization, and Shutdown
 ******************************************************************************/
 
-void CReal3D::AttachRenderer(CRender3D *Render3DPtr)
+void CReal3D::AttachRenderer(IRender3D *Render3DPtr)
 {
 	Render3D = Render3DPtr;
 

Src/Model3/Real3D.h

@@ -319,7 +319,7 @@ public:
 	void Reset(void);
 	
 	/*
-	 * AttachRenderer(render2DPtr):
+	 * AttachRenderer(render3DPtr):
 	 *
 	 * Attaches a 3D renderer for the Real3D to use. This function will
 	 * immediately pass along the information that a CRender3D object needs to
@@ -328,7 +328,7 @@ public:
 	 * Parameters:
 	 *		Render3DPtr		Pointer to a 3D renderer object.
 	 */
-	void AttachRenderer(CRender3D *Render3DPtr);
+	void AttachRenderer(IRender3D *Render3DPtr);
 	
 	/*
 	 * SetStep(stepID):
@@ -388,7 +388,7 @@ private:
 	UINT32		UpdateSnapshot(bool copyWhole, UINT8 *src, UINT8 *dst, unsigned size, UINT8 *dirty);
 
 	// Renderer attached to the Real3D
-	CRender3D	*Render3D;
+	IRender3D	*Render3D;
 	
 	// Data passed from Model 3 object
 	const UINT32	*vrom;	// Video ROM

Src/OSD/SDL/Main.cpp

@@ -1,7 +1,7 @@
 /**
  ** Supermodel
  ** A Sega Model 3 Arcade Emulator.
- ** Copyright 2011-2012 Bart Trzynadlowski, Nik Henson
+ ** Copyright 2011-2016 Bart Trzynadlowski, Nik Henson
  **
  ** This file is part of Supermodel.
  **
@@ -28,6 +28,7 @@
  * -------------------------
  * - Add UI keys for balance setting? 
  * - 5.1 audio support?
+ * - Stretch video option
  *
  * Compile-Time Options
  * --------------------
@@ -216,10 +217,10 @@ static bool CreateGLScreen(const char *caption, unsigned *xOffsetPtr, unsigned *
     	return ErrorLog("Unable to initialize SDL video subsystem: %s\n", SDL_GetError());
     
     // Important GL attributes
-    SDL_GL_SetAttribute(SDL_GL_RED_SIZE,5);	// need at least RGB555 for Model 3 textures
-    SDL_GL_SetAttribute(SDL_GL_GREEN_SIZE,5);
-    SDL_GL_SetAttribute(SDL_GL_BLUE_SIZE,5);
-    SDL_GL_SetAttribute(SDL_GL_DEPTH_SIZE,16);
+    SDL_GL_SetAttribute(SDL_GL_RED_SIZE,8);
+    SDL_GL_SetAttribute(SDL_GL_GREEN_SIZE,8);
+    SDL_GL_SetAttribute(SDL_GL_BLUE_SIZE,8);
+    SDL_GL_SetAttribute(SDL_GL_DEPTH_SIZE,24);
     SDL_GL_SetAttribute(SDL_GL_DOUBLEBUFFER,1);
 
   	// Set vsync
@@ -350,6 +351,87 @@ static void PrintGLInfo(bool createScreen, bool infoLog, bool printExtensions)
 	else			printf("\n");
 }
 
+static void PrintBAT(unsigned regu, unsigned regl)
+{
+#ifdef DEBUG
+  UINT32 batu = ppc_read_spr(regu);
+  UINT32 batl = ppc_read_spr(regl);
+  UINT32 bepi = batu >> (31 - 14);
+  UINT32 bl = (batu >> (31 - 29)) & 0x7ff;
+  bool vs = batu & 2;
+  bool vp = batu & 1;
+  UINT32 brpn = batl >> (31 - 14);
+  UINT32 wimg = (batl >> (31 - 28)) & 0xf;
+  UINT32 pp = batl & 3;
+  UINT32 size = (bl + 1) * 128 * 1024;
+  UINT32 ea_base = bepi << (31 - 14);
+  UINT32 ea_limit = ea_base + size - 1;
+  UINT32 pa_base = brpn << (31 - 14);
+  UINT32 pa_limit = pa_base + size - 1;  
+  printf("%08X-%08X -> %08X-%08X ", ea_base, ea_limit, pa_base, pa_limit);
+  printf("%c%c%c%c ", (wimg&8)?'W':'-', (wimg&4)?'I':'-', (wimg&2)?'M':'-', (wimg&1)?'G':'-');
+  printf("PP=");
+  if (pp == 0)
+    printf("NA");
+  else if (pp == 2)
+    printf("RW");
+  else
+    printf("RO");
+  printf(" Vs=%d Vp=%d", vs, vp);
+#endif
+}
+
+static void DumpPPCRegisters(CBus *bus)
+{
+#ifdef DEBUG
+	for (int i = 0; i < 32; i += 4)
+        printf("R%d=%08X\tR%d=%08X\tR%d=%08X\tR%d=%08X\n",
+        i + 0, ppc_get_gpr(i + 0),
+        i + 1, ppc_get_gpr(i + 1),
+        i + 2, ppc_get_gpr(i + 2),
+        i + 3, ppc_get_gpr(i + 3));
+	printf("PC =%08X\n", ppc_get_pc());
+	printf("LR =%08X\n", ppc_get_lr());
+	printf("DBAT0U=%08X\tIBAT0U=%08X\n", ppc_read_spr(SPR603E_DBAT0U), ppc_read_spr(SPR603E_IBAT0U));
+	printf("DBAT0L=%08X\tIBAT0L=%08X\n", ppc_read_spr(SPR603E_DBAT0L), ppc_read_spr(SPR603E_IBAT0L));
+	printf("DBAT1U=%08X\tIBAT1U=%08X\n", ppc_read_spr(SPR603E_DBAT1U), ppc_read_spr(SPR603E_IBAT1U));
+	printf("DBAT1L=%08X\tIBAT1L=%08X\n", ppc_read_spr(SPR603E_DBAT1L), ppc_read_spr(SPR603E_IBAT1L));
+	printf("DBAT2U=%08X\tIBAT2U=%08X\n", ppc_read_spr(SPR603E_DBAT2U), ppc_read_spr(SPR603E_IBAT2U));
+	printf("DBAT2L=%08X\tIBAT2L=%08X\n", ppc_read_spr(SPR603E_DBAT2L), ppc_read_spr(SPR603E_IBAT2L));
+	printf("DBAT3U=%08X\tIBAT3U=%08X\n", ppc_read_spr(SPR603E_DBAT3U), ppc_read_spr(SPR603E_IBAT3U));
+	printf("DBAT3L=%08X\tIBAT3L=%08X\n", ppc_read_spr(SPR603E_DBAT3L), ppc_read_spr(SPR603E_IBAT3L));
+	for (int i = 0; i < 10; i++)
+		printf("SR%d =%08X VSID=%06X\n", i, ppc_read_sr(i), ppc_read_sr(i) & 0x00ffffff);
+	for (int i = 10; i < 16; i++)
+		printf("SR%d=%08X VSID=%06X\n", i, ppc_read_sr(i), ppc_read_sr(i) & 0x00ffffff);
+	printf("SDR1=%08X\n", ppc_read_spr(SPR603E_SDR1));
+	printf("\n");
+	printf("DBAT0: "); PrintBAT(SPR603E_DBAT0U, SPR603E_DBAT0L); printf("\n");
+	printf("DBAT1: "); PrintBAT(SPR603E_DBAT1U, SPR603E_DBAT1L); printf("\n");
+	printf("DBAT2: "); PrintBAT(SPR603E_DBAT2U, SPR603E_DBAT2L); printf("\n");
+	printf("DBAT3: "); PrintBAT(SPR603E_DBAT3U, SPR603E_DBAT3L); printf("\n");
+	printf("IBAT0: "); PrintBAT(SPR603E_IBAT0U, SPR603E_IBAT0L); printf("\n");
+	printf("IBAT1: "); PrintBAT(SPR603E_IBAT1U, SPR603E_IBAT1L); printf("\n");
+	printf("IBAT2: "); PrintBAT(SPR603E_IBAT2U, SPR603E_IBAT2L); printf("\n");
+	printf("IBAT3: "); PrintBAT(SPR603E_IBAT3U, SPR603E_IBAT3L); printf("\n");
+	printf("\n");
+	/*
+	printf("First PTEG:\n");
+	UINT32 ptab = ppc_read_spr(SPR603E_SDR1) & 0xffff0000;
+	for (int i = 0; i < 65536/8; i++)
+	{
+	  UINT64 pte = bus->Read64(ptab + i*8);
+	  UINT32 vsid = (pte >> (32 + (31 - 24))) & 0x00ffffff;
+	  UINT32 rpn = pte & 0xfffff000;
+	  int wimg = (pte >> 3) & 0xf;
+	  bool v = pte & 0x8000000000000000ULL;
+	  printf("  %d: %016llX V=%d VSID=%06X RPN=%08X WIMG=%c%c%c%c\n", i, pte, v, vsid, rpn, (wimg&8)?'W':'-', (wimg&4)?'I':'-', (wimg&2)?'M':'-', (wimg&1)?'G':'-');
+  }
+  */
+	
+#endif
+}
+
 /******************************************************************************
  Configuration
  
@@ -433,6 +515,7 @@ static void ApplySettings(CINIFile *INI, const char *section)
 #endif // SUPERMODEL_WIN32
 	
 	// OSD
+	INI->Get(section, "New3DEngine", g_Config.new3DEngine);
 	INI->Get(section, "XResolution", g_Config.xRes);
 	INI->Get(section, "YResolution", g_Config.yRes);
 	INI->Get(section, "FullScreen", g_Config.fullScreen);
@@ -479,6 +562,7 @@ static void LogConfig(void)
 	InfoLog("Program settings:");
 	
 	// COSDConfig
+	InfoLog("\tNew3DEngine                   = %d", g_Config.new3DEngine);
 	InfoLog("\tXResolution                   = %d", g_Config.xRes);
 	InfoLog("\tYResolution                   = %d", g_Config.yRes);
 	InfoLog("\tFullScreen                    = %d", g_Config.fullScreen);
@@ -528,7 +612,7 @@ static void LogConfig(void)
 	InfoLog("\tForceFeedback                 = %d", g_Config.forceFeedback);
 #endif
 
-	// CRender3DConfig
+	// CLegacy3DConfig
 	InfoLog("\tVertexShader                  = %s", g_Config.vertexShaderFile.c_str());
 	InfoLog("\tFragmentShader                = %s", g_Config.fragmentShaderFile.c_str());
 
@@ -800,8 +884,8 @@ int Supermodel(const char *zipFile, CInputs *Inputs, COutputs *Outputs, CINIFile
 	CModel3			*Model3 = new CModel3();
 #endif // SUPERMODEL_DEBUGGER
 	char			baseTitleStr[128], titleStr[128];
-	CRender2D		*Render2D = new CRender2D();
-	CRender3D		*Render3D = new CRender3D();
+	CRender2D		*Render2D;
+	IRender3D		*Render3D;
 	unsigned		prevFPSTicks, currentFPSTicks, currentTicks, targetTicks, startTicks;
 	unsigned		fpsFramesElapsed, framesElapsed;
 	bool			gameHasLightguns = false;
@@ -852,7 +936,9 @@ int Supermodel(const char *zipFile, CInputs *Inputs, COutputs *Outputs, CINIFile
 	if (Outputs != NULL)
 		Model3->AttachOutputs(Outputs);
 	
-	// Initialize the renderer
+	// Initialize the renderers
+	Render2D = new CRender2D();
+	Render3D = g_Config.new3DEngine ? ((IRender3D *) new New3D::CNew3D()) : ((IRender3D *) new Legacy3D::CLegacy3D());
 	if (OKAY != Render2D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
 		goto QuitError;
 	if (OKAY != Render3D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
@@ -985,7 +1071,7 @@ int Supermodel(const char *zipFile, CInputs *Inputs, COutputs *Outputs, CINIFile
 
 			// Recreate renderers and attach to the emulator
 			Render2D = new CRender2D();
-			Render3D = new CRender3D();
+			Render3D = g_Config.new3DEngine ? ((IRender3D *) new New3D::CNew3D()) : ((IRender3D *) new Legacy3D::CLegacy3D());
 			if (OKAY != Render2D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
 				goto QuitError;
 			if (OKAY != Render3D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
@@ -1204,33 +1290,6 @@ int Supermodel(const char *zipFile, CInputs *Inputs, COutputs *Outputs, CINIFile
 	delete Render2D;
 	delete Render3D;
 	
-	// Dump PowerPC registers
-#ifdef DEBUG
-	for (int i = 0; i < 32; i += 4)
-        printf("R%d=%08X\tR%d=%08X\tR%d=%08X\tR%d=%08X\n",
-        i + 0, ppc_get_gpr(i + 0),
-        i + 1, ppc_get_gpr(i + 1),
-        i + 2, ppc_get_gpr(i + 2),
-        i + 3, ppc_get_gpr(i + 3));
-	printf("PC =%08X\n", ppc_get_pc());
-  	printf("LR =%08X\n", ppc_get_lr());
-	/*
-	printf("DBAT0U=%08X\tIBAT0U=%08X\n", ppc_read_spr(SPR603E_DBAT0U), ppc_read_spr(SPR603E_IBAT0U));
-	printf("DBAT0L=%08X\tIBAT0L=%08X\n", ppc_read_spr(SPR603E_DBAT0L), ppc_read_spr(SPR603E_IBAT0L));
-	printf("DBAT1U=%08X\tIBAT1U=%08X\n", ppc_read_spr(SPR603E_DBAT1U), ppc_read_spr(SPR603E_IBAT1U));
-	printf("DBAT1L=%08X\tIBAT1L=%08X\n", ppc_read_spr(SPR603E_DBAT1L), ppc_read_spr(SPR603E_IBAT1L));
-	printf("DBAT2U=%08X\tIBAT2U=%08X\n", ppc_read_spr(SPR603E_DBAT2U), ppc_read_spr(SPR603E_IBAT2U));
-	printf("DBAT2L=%08X\tIBAT2L=%08X\n", ppc_read_spr(SPR603E_DBAT2L), ppc_read_spr(SPR603E_IBAT2L));
-	printf("DBAT3U=%08X\tIBAT3U=%08X\n", ppc_read_spr(SPR603E_DBAT3U), ppc_read_spr(SPR603E_IBAT3U));
-	printf("DBAT3L=%08X\tIBAT3L=%08X\n", ppc_read_spr(SPR603E_DBAT3L), ppc_read_spr(SPR603E_IBAT3L));
-	for (int i = 0; i < 10; i++)
-		printf("SR%d =%08X\n", i, ppc_read_sr(i));
-	for (int i = 10; i < 16; i++)
-		printf("SR%d=%08X\n", i, ppc_read_sr(i));
-	printf("SDR1=%08X\n", ppc_read_spr(SPR603E_SDR1));
-	*/
-#endif
-	
 	return 0;
 
 	// Quit with an error
@@ -1316,7 +1375,7 @@ static int DisassembleCROM(const char *zipFile, UINT32 addr, unsigned n)
 static void Title(void)
 {
 	puts("Supermodel: A Sega Model 3 Arcade Emulator (Version "SUPERMODEL_VERSION")");
-	puts("Copyright (C) 2011-2012 by Bart Trzynadlowski and Nik Henson\n");
+	puts("Copyright 2011-2016 by Bart Trzynadlowski and Nik Henson\n");
 }
 
 // Print usage information
@@ -1326,7 +1385,7 @@ static void Help(void)
 	puts("ROM set must be a valid ZIP file containing a single game.");
 	puts("");
 	puts("General Options:");
-	puts("    -?, -h                 Print this help text");
+	puts("    -?, -h, -help, --help  Print this help text");
 	puts("    -print-games           List supported games and quit");
 	puts("");
 	puts("Core Options:");
@@ -1346,8 +1405,10 @@ static void Help(void)
 	puts("    -show-fps              Display frame rate in window title bar");
 	puts("    -crosshairs=<n>        Crosshairs configuration for gun games:");
 	puts("                            0=none [Default], 1=P1 only, 2=P2 only, 3=P1 & P2");
-	puts("    -vert-shader=<file>    Load 3D vertex shader from external file");
-	puts("    -frag-shader=<file>    Load 3D fragment shader from external file");
+	puts("    -legacy3d              Legacy 3D engine [Default]");
+	puts("    -new3d                 New 3D engine by Ian Curtis");
+	puts("    -vert-shader=<file>    Load vertex shader from file (legacy 3D engine)");
+	puts("    -frag-shader=<file>    Load fragment shader from file (legacy 3D engine)");
 	puts("    -print-gl-info         Print OpenGL driver information and quit");
 	puts("");
 	puts("Audio Options:");
@@ -1371,7 +1432,7 @@ static void Help(void)
 	puts("");
 #ifdef SUPERMODEL_DEBUGGER
 	puts("Debug Options:");
-	puts("    -disable-debugger	     Completely disable debugger functionality");
+	puts("    -disable-debugger      Completely disable debugger functionality");
 	puts("    -enter-debugger        Enter debugger at start of emulation");
 	puts("    -dis=<addr>[,n]        Disassemble PowerPC code from CROM");
 	puts("");
@@ -1449,7 +1510,17 @@ int main(int argc, char **argv)
 	CmdLine.SetDefaultSectionName("Global");	// command line settings are global-level
 	for (i = 1; i < argc; i++)
 	{
-		if (!strcmp(argv[i],"-h") || !strcmp(argv[i],"-?"))
+		if (!strcmp(argv[i],"-new3d"))
+		{
+			n = 1;
+			CmdLine.Set("Global", "New3DEngine", n);
+		}
+		else if (!strcmp(argv[i],"-legacy3d"))
+	  {
+	    n = 0;
+	    CmdLine.Set("Global", "New3DEngine", n);
+	  }
+		else if (!strcmp(argv[i],"-h") || !strcmp(argv[i],"-?") || !strcmp(argv[i],"-help") || !strcmp(argv[i],"--help"))
 		{
 			Help();
 			return 0;

Src/OSD/SDL/OSDConfig.h

@@ -1,7 +1,7 @@
 /**
  ** Supermodel
  ** A Sega Model 3 Arcade Emulator.
- ** Copyright 2011 Bart Trzynadlowski, Nik Henson 
+ ** Copyright 2011-2016 Bart Trzynadlowski, Nik Henson 
  **
  ** This file is part of Supermodel.
  **
@@ -43,6 +43,7 @@ using namespace std;
 class COSDConfig
 {
 public:
+	bool		new3DEngine;	// New 3D engine
 	unsigned	xRes, yRes;		// X and Y resolution, in pixels
 	bool 		fullScreen;		// Full screen mode (if true)
 	bool		wideScreen;		// Wide screen hack
@@ -135,6 +136,7 @@ public:
 	// Defaults
 	COSDConfig(void)
 	{
+		new3DEngine = false;
 		xRes = 496;
 		yRes = 384;
 		fullScreen = false;