Add supersampling anti-aliasing

Late christmas present. Due to the way alpha works on the model3 adding regular anti-aliasing doesn't really work. Supersampling is very much a brute force solution, render the scene at a higher resolution and mipmap it.

It's enabled via command line with the -ss option, for example -ss=4 for 4x supersampling or by adding Supersampling = 4 in the config file.

Note non power of two values work as well, so 3 gives a very good balance between speed and quality. 8 will make your GPU bleed, since it is essentially rendering 64 pixels for every visible pixel on the screen.

Makefiles/Rules.inc

@@ -116,6 +116,7 @@ SRC_FILES = \
 	Src/Graphics/New3D/R3DScrollFog.cpp \
 	Src/Graphics/FBO.cpp \
 	Src/Graphics/Render2D.cpp \
+	Src/Graphics/SuperAA.cpp \
 	Src/Model3/TileGen.cpp \
 	Src/Model3/Model3.cpp \
 	Src/CPU/PowerPC/ppc.cpp \

Src/Graphics/IRender3D.h

@@ -17,7 +17,7 @@ public:
   virtual void UploadTextures(unsigned level, 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 SetStepping(int stepping) = 0;
-  virtual bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes) = 0;
+  virtual bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes, unsigned aaTarget) = 0;
   virtual void SetSunClamp(bool enable) = 0;
   virtual void SetSignedShade(bool enable) = 0;
   virtual float GetLosValue(int layer) = 0;

Src/Graphics/Legacy3D/Legacy3D.cpp

@@ -964,6 +964,10 @@ void CLegacy3D::RenderFrame(void)
   // Begin frame
   ClearErrors();  // must be cleared each frame
 
+  if (m_aaTarget) {
+      glBindFramebuffer(GL_FRAMEBUFFER, m_aaTarget);			// if we have an AA target draw to it instead of the default back buffer
+  }
+  
   // Z buffering (Z buffer is cleared by display list viewport nodes)
   glDepthFunc(GL_LESS);
   glEnable(GL_DEPTH_TEST);
@@ -1039,6 +1043,10 @@ void CLegacy3D::RenderFrame(void)
   glDisableClientState(GL_TEXTURE_COORD_ARRAY);
   glDisableClientState(GL_NORMAL_ARRAY);
   glDisableClientState(GL_VERTEX_ARRAY);
+
+  if (m_aaTarget) {
+      glBindFramebuffer(GL_FRAMEBUFFER, 0);			// restore target if needed
+  }
 }
 
 void CLegacy3D::EndFrame(void)
@@ -1089,7 +1097,7 @@ void CLegacy3D::SetStepping(int stepping)
   DebugLog("Legacy3D set to Step %d.%d\n", (step>>4)&0xF, step&0xF);
 }
   
-bool CLegacy3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXResParam, unsigned totalYResParam)
+bool CLegacy3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXResParam, unsigned totalYResParam, unsigned aaTarget)
 {
   // Allocate memory for texture buffer
   textureBuffer = new(std::nothrow) GLfloat[1024*1024*4];
@@ -1098,6 +1106,8 @@ bool CLegacy3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned
     
   glGetError(); // clear error flag
 
+  m_aaTarget = aaTarget;
+  
   // Create model caches and VBOs
   if (CreateModelCache(&VROMCache, NUM_STATIC_VERTS, NUM_LOCAL_VERTS, NUM_STATIC_MODELS, 0x4000000/4, NUM_DISPLAY_LIST_ITEMS, false))
     return FAIL;
@@ -1310,7 +1320,8 @@ float CLegacy3D::GetLosValue(int layer)
 }
 
 CLegacy3D::CLegacy3D(const Util::Config::Node &config)
-  : m_config(config)
+  : m_config(config),
+    m_aaTarget(0)
 { 
   cullingRAMLo = NULL;
   cullingRAMHi = NULL;

Src/Graphics/Legacy3D/Legacy3D.h

@@ -328,7 +328,7 @@ public:
 	 *		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);
+	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes, unsigned aaTarget);
 
 	/*
 	* SetSunClamp(bool enable);
@@ -457,6 +457,7 @@ private:
 	GLfloat	spotColor[3];
 	GLint	viewportX, viewportY;
 	GLint	viewportWidth, viewportHeight;
+	GLuint  m_aaTarget;
 	
 	// Scene graph processing
 	int		listDepth;	        // how many lists have we recursed into

Src/Graphics/New3D/New3D.cpp

@@ -22,7 +22,8 @@ CNew3D::CNew3D(const Util::Config::Node &config, const std::string& gameName) :
 	m_r3dScrollFog(config),
 	m_gameName(gameName),
 	m_textureBuffer(0),
-	m_vao(0)
+	m_vao(0),
+	m_aaTarget(0)
 {
 	m_cullingRAMLo	= nullptr;
 	m_cullingRAMHi	= nullptr;
@@ -121,7 +122,7 @@ void CNew3D::SetStepping(int stepping)
 	}
 }
 
-bool CNew3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXResParam, unsigned totalYResParam)
+bool CNew3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXResParam, unsigned totalYResParam, unsigned aaTarget)
 {
 	// Resolution and offset within physical display area
 	m_xRatio	= xRes * (float)(1.0 / 496.0);
@@ -132,6 +133,7 @@ bool CNew3D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yR
 	m_yRes		= yRes;
 	m_totalXRes	= totalXResParam;
 	m_totalYRes = totalYResParam;
+	m_aaTarget	= aaTarget;
 
 	m_r3dFrameBuffers.DestroyFBO();		// remove any old ones if created
 	m_r3dFrameBuffers.CreateFBO(totalXResParam, totalYResParam);
@@ -440,7 +442,16 @@ void CNew3D::RenderFrame(void)
 	}
 
 	m_r3dFrameBuffers.SetFBO(Layer::none);
+
+	if (m_aaTarget) {
+		glBindFramebuffer(GL_FRAMEBUFFER, m_aaTarget);			// if we have an AA target draw to it instead of the default back buffer
+	}
+
 	m_r3dFrameBuffers.Draw();
+
+	if (m_aaTarget) {
+		glBindFramebuffer(GL_FRAMEBUFFER, 0);
+	}
 }
 
 void CNew3D::BeginFrame(void)
@@ -1009,6 +1020,7 @@ void CNew3D::RenderViewport(UINT32 addr)
 		m_planes.bnrv = Util::Uint32AsFloat(vpnode[0x11]);
 		m_planes.bnbu = Util::Uint32AsFloat(vpnode[0x12]);
 		m_planes.bnbw = Util::Uint32AsFloat(vpnode[0x13]);
+		m_planes.correction = 1.0f;		// might get changed by the calc viewport method
 
 		vp->angle_left		= (0.0f - jo) / cv;
 		vp->angle_right		= (1.0f - jo) / cv;

Src/Graphics/New3D/New3D.h

@@ -141,7 +141,7 @@ public:
 	*		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);
+	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes, unsigned aaTarget);
 
 	/*
 	* SetSunClamp(bool enable);
@@ -223,6 +223,7 @@ private:
 	void DisableRenderStates();
 	void TranslateLosPosition(int inX, int inY, int& outX, int& outY);
 	bool ProcessLos(int priority);
+	void CalcViewport(Viewport* vp);
 
 	/*
 	* Data
@@ -286,6 +287,7 @@ private:
 	R3DShader m_r3dShader;
 	R3DScrollFog m_r3dScrollFog;
 	R3DFrameBuffers m_r3dFrameBuffers;
+	GLuint m_aaTarget;						// optional, maybe zero
 
 	int m_currentPriority;
 
@@ -301,8 +303,6 @@ private:
 		float bnbw;
 		float correction;
 	} m_planes;	
-
-	void CalcViewport		(Viewport* vp);
 };
 
 } // New3D

Src/Graphics/New3D/R3DFrameBuffers.cpp

@@ -298,7 +298,6 @@ void R3DFrameBuffers::AllocShaderTrans()
 
 void R3DFrameBuffers::Draw()
 {
-	SetFBO		(Layer::none);						// make sure to draw on the back buffer
 	glViewport	(0, 0, m_width, m_height);			// cover the entire screen
 	glDisable	(GL_DEPTH_TEST);					// disable depth testing / writing
 	glDisable	(GL_CULL_FACE);

Src/Graphics/Render2D.cpp

@@ -310,11 +310,19 @@ void CRender2D::Setup2D(bool isBottom)
 	// Clear everything if requested or just overscan areas for wide screen mode
 	if (isBottom)
 	{
+		if (m_aaTarget) {
+			glBindFramebuffer(GL_FRAMEBUFFER, m_aaTarget);	// set target if needed
+		}
+
 		glClearColor(0.0, 0.0, 0.0, 0.0);
 		glViewport	(0, 0, m_totalXPixels, m_totalYPixels);
 		glDisable	(GL_SCISSOR_TEST);							// scissor is enabled to fix the 2d/3d miss match problem
 		glClear		(GL_COLOR_BUFFER_BIT);						// we want to clear outside the scissored areas so must disable it
 		glEnable	(GL_SCISSOR_TEST);
+
+		if (m_aaTarget) {
+			glBindFramebuffer(GL_FRAMEBUFFER, 0);			// restore target if needed
+		}
 	}
 
 	// Set up the viewport and orthogonal projection
@@ -327,6 +335,10 @@ void CRender2D::Setup2D(bool isBottom)
 
 void CRender2D::DrawSurface(GLuint textureID)
 {
+	if (m_aaTarget) {
+		glBindFramebuffer(GL_FRAMEBUFFER, m_aaTarget);	// set target if needed
+	}
+
 	m_shader.EnableShader();
 
 	glEnable			(GL_BLEND);
@@ -338,6 +350,10 @@ void CRender2D::DrawSurface(GLuint textureID)
 	glDisable			(GL_BLEND);
 
 	m_shader.DisableShader();	
+
+	if (m_aaTarget) {
+		glBindFramebuffer(GL_FRAMEBUFFER, 0);			// restore target if needed
+	}
 }
 
 float CRender2D::LineToPercentStart(int lineNumber)
@@ -470,7 +486,7 @@ void CRender2D::AttachVRAM(const uint8_t* vramPtr)
 	DebugLog("Render2D attached VRAM\n");
 }
 
-bool CRender2D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes)
+bool CRender2D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes, unsigned aaTarget)
 {
 	// Resolution
 	m_xPixels		= xRes;
@@ -480,6 +496,7 @@ bool CRender2D::Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned
 	m_totalXPixels	= totalXRes;
 	m_totalYPixels	= totalYRes;
 	m_correction	= (UINT32)(((yRes / 384.f) * 2) + 0.5f);		// for some reason the 2d layer is 2 pixels off the 3D
+	m_aaTarget		= aaTarget;
 
 	return OKAY;
 }

Src/Graphics/Render2D.h

@@ -158,7 +158,7 @@ public:
 	 *    OKAY is successful, otherwise FAILED if a non-recoverable error
 	 *    occurred. Prints own error messages.
 	 */
-	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes);
+	bool Init(unsigned xOffset, unsigned yOffset, unsigned xRes, unsigned yRes, unsigned totalXRes, unsigned totalYRes, unsigned aaTarget);
 
 	/*
 	 * CRender2D(config):
@@ -198,6 +198,7 @@ private:
 	unsigned  m_totalXPixels = 0;   // total display surface resolution
 	unsigned  m_totalYPixels = 0;
 	unsigned  m_correction = 0;
+	GLuint m_aaTarget = 0;
 
 	GLuint m_vao;
 	GLSLShader m_shader;

Src/Graphics/SuperAA.cpp

@@ -0,0 +1,137 @@
+#include "SuperAA.h"
+#include <string>
+
+SuperAA::SuperAA(int aaValue) :
+	m_aa(aaValue),
+	m_vao(0),
+	m_width(0),
+	m_height(0)
+{
+	if (aaValue > 1) {
+
+		const char* vertexShader = R"glsl(
+
+			#version 410 core
+
+			// outputs
+			out vec2 fsTexCoord;
+
+			void main(void)
+			{
+				const vec4 vertices[] = vec4[](vec4(-1.0, -1.0, 0.0, 1.0),
+												vec4(-1.0,  1.0, 0.0, 1.0),
+												vec4( 1.0, -1.0, 0.0, 1.0),
+												vec4( 1.0,  1.0, 0.0, 1.0));
+
+				fsTexCoord = (vertices[gl_VertexID % 4].xy + 1.0) / 2.0;
+				gl_Position = vertices[gl_VertexID % 4];	
+			}
+
+			)glsl";
+
+
+		std::string fragmentShaderVersion = R"glsl(
+			#version 410 core
+
+		)glsl";
+
+		std::string aaString = "const int aa = ";
+		aaString += std::to_string(aaValue);
+		aaString += ";\n";
+
+		std::string fragmentShader = R"glsl(
+
+		// inputs
+		uniform sampler2D tex1;			// base tex
+		in vec2 fsTexCoord;
+
+		// outputs
+		out vec4 fragColor;
+
+		vec4 GetTextureValue(sampler2D s, vec2 texCoord)
+		{
+			vec2 size		= vec2(textureSize(s,0));        // want the values as floats
+			ivec2 texPos	= ivec2(size * texCoord);
+			vec4 texColour	= vec4(0.0);
+
+			for(int i=0; i < aa; i++) {
+				for(int j=0; j < aa; j++) {
+					texColour += texelFetch(s,ivec2(texPos.x+i,texPos.y+j),0);
+				}
+			}
+
+			texColour /= float(aa * aa);
+
+			return texColour;
+		}
+
+		void main()
+		{
+			fragColor = GetTextureValue(tex1, fsTexCoord);
+		}
+
+		)glsl";
+
+		std::string fragmentShaderString = fragmentShaderVersion + aaString + fragmentShader;
+		
+		// load shaders
+		m_shader.LoadShaders(vertexShader, fragmentShaderString.c_str());
+		m_shader.GetUniformLocationMap("tex1");
+
+		// setup uniform memory
+		m_shader.EnableShader();
+		glUniform1i(m_shader.attribLocMap["tex1"], 0);		// texture will be bound to unit zero
+		m_shader.DisableShader();
+
+		glGenVertexArrays(1, &m_vao);
+		glBindVertexArray(m_vao);
+		// no states needed since we do it in the shader
+		glBindVertexArray(0);
+	}
+}
+
+// need an active context bound to the current thread to destroy our objects
+SuperAA::~SuperAA()
+{
+	m_shader.UnloadShaders();
+	m_fbo.Destroy();
+
+	if (m_vao) {
+		glDeleteVertexArrays(1, &m_vao);
+		m_vao = 0;
+	}
+}
+
+void SuperAA::Init(int width, int height)
+{
+	if (m_aa > 1) {
+		m_fbo.Destroy();
+		m_fbo.Create(width * m_aa, height * m_aa);
+
+		m_width = width;
+		m_height = height;
+	}
+}
+
+void SuperAA::Draw()
+{
+	if (m_aa > 1) {
+		glDisable(GL_DEPTH_TEST);
+		glDisable(GL_STENCIL_TEST);
+		glDisable(GL_SCISSOR_TEST);
+		glDisable(GL_BLEND);
+
+		glBindTexture(GL_TEXTURE_2D, m_fbo.GetTextureID());
+		glBindVertexArray(m_vao);
+		glViewport(0, 0, m_width, m_height);
+		m_shader.EnableShader();
+		glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
+		m_shader.DisableShader();
+		glBindVertexArray(0);
+	}
+}
+
+GLuint SuperAA::GetTargetID()
+{
+	return m_fbo.GetFBOID();	// will return 0 if no render target which will be our default frame buffer (back buffer)
+}

Src/Graphics/SuperAA.h

@@ -0,0 +1,31 @@
+#pragma once
+
+#include "FBO.h"
+#include "New3D/GLSLShader.h"
+
+// This class just implements super sampling. Super sampling looks fantastic but is quite expensive.
+// 8x and beyond values can start to eat ridiculous amounts of memory / gpu time, for less and less noticable returns
+// 4x works and looks great
+// values such as 3 are also possible, that works out 9 samples per pixel
+// The algorithm is super simple, just add up all samples and divide by the number
+
+class SuperAA
+{
+public:
+	SuperAA(int aaValue);		
+	~SuperAA();
+
+	void Init(int width, int height);		// width & height are real window dimensions
+	void Draw();							// this is a no-op if AA is 1, since we'll be drawing straight on the back buffer anyway
+
+	GLuint GetTargetID();
+
+private:
+	FBO m_fbo;
+	GLSLShader m_shader;
+	const int m_aa;
+	GLuint m_vao;
+	int m_width;
+	int m_height;
+};
+

Src/Model3/IEmulator.h

@@ -35,6 +35,7 @@ class CRender2D;
 class IRender3D;
 class CInputs;
 class COutputs;
+class SuperAA;
 
 /*
  * IEmulator:
@@ -150,7 +151,7 @@ public:
    *    Render2DPtr   Pointer to a tile renderer object.
    *    Render3DPtr   Same as above but for a 3D renderer.
    */
-  virtual void AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr) = 0;
+  virtual void AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr, SuperAA *superAA) = 0;
   
   /*
    * AttachInputs(InputsPtr):

Src/Model3/Model3.cpp

@@ -2163,6 +2163,7 @@ void CModel3::RenderFrame(void)
     TileGen.RenderFrameTop();
     GPU.EndFrame();
     TileGen.EndFrame();
+    m_superAA->Draw();
   }
 
   EndFrameVideo();
@@ -3062,10 +3063,11 @@ bool CModel3::LoadGame(const Game &game, const ROMSet &rom_set)
   return OKAY;
 }
 
-void CModel3::AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr)
+void CModel3::AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr, SuperAA *superAA)
 {
   TileGen.AttachRenderer(Render2DPtr);
   GPU.AttachRenderer(Render3DPtr);
+  m_superAA = superAA;
 }
 
 void CModel3::AttachInputs(CInputs *InputsPtr)
@@ -3209,7 +3211,8 @@ CModel3::CModel3(Util::Config::Node &config)
     TileGen(config),
     GPU(config),
     SoundBoard(config),
-    m_jtag(GPU)
+    m_jtag(GPU),
+    m_superAA(nullptr)
 {
   // Initialize pointers so dtor can know whether to free them
   memoryPool = NULL;

Src/Model3/Model3.h

@@ -45,6 +45,8 @@
 #include "Network/INetBoard.h"
 #endif // NET_BOARD
 #include "Util/NewConfig.h"
+#include "Graphics/SuperAA.h"
+
 
 /*
  * FrameTimings
@@ -92,7 +94,7 @@ public:
   void RenderFrame(void);
   void Reset(void);
   const Game &GetGame(void) const;
-  void AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr);
+  void AttachRenderers(CRender2D *Render2DPtr, IRender3D *Render3DPtr, SuperAA *superAA);
   void AttachInputs(CInputs *InputsPtr);
   void AttachOutputs(COutputs *OutputsPtr);
   bool Init(void);
@@ -321,6 +323,7 @@ private:
   CDriveBoard *DriveBoard;    // Drive board
   CCrypto     m_cryptoDevice; // Encryption device
   CJTAG       m_jtag;         // JTAG interface
+  SuperAA     *m_superAA;
 #ifdef NET_BOARD
   INetBoard   *NetBoard;      // Net board
   bool		m_runNetBoard;

Src/OSD/SDL/Main.cpp

@@ -94,6 +94,7 @@
 #include "Model3/Model3.h"
 #include "OSD/Audio.h"
 #include "Graphics/New3D/VBO.h"
+#include "Graphics/SuperAA.h"
 
 #include <iostream>
 #include "Util/BMPFile.h"
@@ -122,6 +123,7 @@ SDL_Window *s_window = nullptr;
 static unsigned  xOffset, yOffset;      // offset of renderer output within OpenGL viewport
 static unsigned  xRes, yRes;            // renderer output resolution (can be smaller than GL viewport)
 static unsigned  totalXRes, totalYRes;  // total resolution (the whole GL viewport)
+static int aaValue = 1;                 // default is 1 which is no aa
 
 /*
  * Crosshair stuff
@@ -185,11 +187,11 @@ static bool SetGLGeometry(unsigned *xOffsetPtr, unsigned *yOffsetPtr, unsigned *
   // Scissor box (to clip visible area)
   if (s_runtime_config["WideScreen"].ValueAsDefault<bool>(false))
   {
-    glScissor(0, correction, *totalXResPtr, *totalYResPtr - (correction * 2));
+    glScissor(0* aaValue, correction* aaValue, *totalXResPtr * aaValue, (*totalYResPtr - (correction * 2)) * aaValue);
   }
   else
   {
-    glScissor(*xOffsetPtr + correction, *yOffsetPtr + correction, *xResPtr - (correction * 2), *yResPtr - (correction * 2));
+      glScissor((*xOffsetPtr + correction) * aaValue, (*yOffsetPtr + correction) * aaValue, (*xResPtr - (correction * 2)) * aaValue, (*yResPtr - (correction * 2)) * aaValue);
   }
   return OKAY;
 }
@@ -971,13 +973,17 @@ int Supermodel(const Game &game, ROMSet *rom_set, IEmulator *Model3, CInputs *In
   uint64_t nextTime = 0;
 
   // Initialize the renderers
+  SuperAA* superAA = new SuperAA(aaValue);
+  superAA->Init(totalXRes, totalYRes);  // pass actual frame sizes here
   CRender2D *Render2D = new CRender2D(s_runtime_config);
   IRender3D *Render3D = s_runtime_config["New3DEngine"].ValueAs<bool>() ? ((IRender3D *) new New3D::CNew3D(s_runtime_config, Model3->GetGame().name)) : ((IRender3D *) new Legacy3D::CLegacy3D(s_runtime_config));
-  if (OKAY != Render2D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
+
+  if (OKAY != Render2D->Init(xOffset*aaValue, yOffset*aaValue, xRes*aaValue, yRes*aaValue, totalXRes*aaValue, totalYRes*aaValue, superAA->GetTargetID()))
     goto QuitError;
-  if (OKAY != Render3D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
+  if (OKAY != Render3D->Init(xOffset*aaValue, yOffset*aaValue, xRes*aaValue, yRes*aaValue, totalXRes*aaValue, totalYRes*aaValue, superAA->GetTargetID()))
     goto QuitError;
-  Model3->AttachRenderers(Render2D,Render3D);
+ 
+  Model3->AttachRenderers(Render2D,Render3D, superAA);
 
   // Reset emulator
   Model3->Reset();
@@ -1105,8 +1111,8 @@ int Supermodel(const Game &game, ROMSet *rom_set, IEmulator *Model3, CInputs *In
       // Delete renderers and recreate them afterwards since GL context will most likely be lost when switching from/to fullscreen
       delete Render2D;
       delete Render3D;
-      Render2D = NULL;
+      Render2D = nullptr;
-      Render3D = NULL;
+      Render3D = nullptr;
 
       // Resize screen
       totalXRes = xRes = s_runtime_config["XResolution"].ValueAs<unsigned>();
@@ -1117,13 +1123,15 @@ int Supermodel(const Game &game, ROMSet *rom_set, IEmulator *Model3, CInputs *In
         goto QuitError;
 
       // Recreate renderers and attach to the emulator
+      superAA->Init(totalXRes, totalYRes);
       Render2D = new CRender2D(s_runtime_config);
       Render3D = s_runtime_config["New3DEngine"].ValueAs<bool>() ? ((IRender3D *) new New3D::CNew3D(s_runtime_config, Model3->GetGame().name)) : ((IRender3D *) new Legacy3D::CLegacy3D(s_runtime_config));
-      if (OKAY != Render2D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
+      if (OKAY != Render2D->Init(xOffset * aaValue, yOffset * aaValue, xRes * aaValue, yRes * aaValue, totalXRes * aaValue, totalYRes * aaValue, superAA->GetTargetID()))
         goto QuitError;
-      if (OKAY != Render3D->Init(xOffset, yOffset, xRes, yRes, totalXRes, totalYRes))
+      if (OKAY != Render3D->Init(xOffset * aaValue, yOffset * aaValue, xRes * aaValue, yRes * aaValue, totalXRes * aaValue, totalYRes * aaValue, superAA->GetTargetID()))
         goto QuitError;
-      Model3->AttachRenderers(Render2D,Render3D);
+
+      Model3->AttachRenderers(Render2D, Render3D, superAA);
 
       Render3D->UploadTextures(0, 0, 0, 2048, 2048);    // sync texture memory
 
@@ -1331,6 +1339,7 @@ int Supermodel(const Game &game, ROMSet *rom_set, IEmulator *Model3, CInputs *In
   // Shut down renderers
   delete Render2D;
   delete Render3D;
+  delete superAA;
 
   return 0;
 
@@ -1338,6 +1347,8 @@ int Supermodel(const Game &game, ROMSet *rom_set, IEmulator *Model3, CInputs *In
 QuitError:
   delete Render2D;
   delete Render3D;
+  delete superAA;
+
   return 1;
 }
 
@@ -1478,7 +1489,7 @@ static Util::Config::Node DefaultConfig()
   config.SetEmpty("WindowYPosition");
   config.Set("FullScreen", false);
   config.Set("BorderlessWindow", false);
-
+  config.Set("Supersampling", 1);
   config.Set("WideScreen", false);
   config.Set("Stretch", false);
   config.Set("WideBackground", false);
@@ -1558,6 +1569,7 @@ static void Help(void)
   puts("");
   puts("Video Options:");
   puts("  -res=<x>,<y>            Resolution [Default: 496,384]");
+  puts("  -ss=<n>                 Supersampling (range 1-8)");
   puts("  -window-pos=<x>,<y>     Window position [Default: centered]");
   puts("  -window                 Windowed mode [Default]");
   puts("  -borderless             Windowed mode with no border");
@@ -1819,6 +1831,29 @@ static ParsedCommandLine ParseCommandLine(int argc, char **argv)
               }
           }
       }
+      else if (arg == "-ss" || arg.find("-ss=") == 0) {
+
+          std::vector<std::string> parts = Util::Format(arg).Split('=');
+
+          if (parts.size() != 2)
+          {
+              ErrorLog("'-ss' requires an integer argument (e.g., '-ss=2').");
+              cmd_line.error = true;
+          }
+          else {
+
+              try {
+                  int val = std::stoi(parts[1]);
+                  val = std::clamp(val, 1, 8);
+
+                  cmd_line.config.Set("Supersampling", val);
+              }
+              catch (...) {
+                  ErrorLog("'-ss' requires an integer argument (e.g., '-ss=2').");
+                  cmd_line.error = true;
+              }
+          }
+      }
       else if (arg == "-true-hz")
         cmd_line.config.Set("RefreshRate", 57.524f);
       else if (arg == "-print-gl-info")
@@ -1967,6 +2002,8 @@ int main(int argc, char **argv)
 #endif // SUPERMODEL_DEBUGGER
   std::string selectedInputSystem = s_runtime_config["InputSystem"].ValueAs<std::string>();
 
+  aaValue = s_runtime_config["Supersampling"].ValueAs<int>();
+
   // Create a window
   xRes = 496;
   yRes = 384;

VS2008/Supermodel.vcxproj

@@ -328,6 +328,7 @@ xcopy /D /Y "$(ProjectDir)..\Assets\*" "$(TargetDir)Assets"</Command>
     <ClCompile Include="..\Src\Graphics\New3D\Vec.cpp" />
     <ClCompile Include="..\Src\Graphics\Render2D.cpp" />
     <ClCompile Include="..\Src\Graphics\Shader.cpp" />
+    <ClCompile Include="..\Src\Graphics\SuperAA.cpp" />
     <ClCompile Include="..\Src\Inputs\Input.cpp" />
     <ClCompile Include="..\Src\Inputs\Inputs.cpp" />
     <ClCompile Include="..\Src\Inputs\InputSource.cpp" />
@@ -507,6 +508,7 @@ xcopy /D /Y "$(ProjectDir)..\Assets\*" "$(TargetDir)Assets"</Command>
     <ClInclude Include="..\Src\Graphics\Render2D.h" />
     <ClInclude Include="..\Src\Graphics\Shader.h" />
     <ClInclude Include="..\Src\Graphics\Shaders2D.h" />
+    <ClInclude Include="..\Src\Graphics\SuperAA.h" />
     <ClInclude Include="..\Src\Inputs\Input.h" />
     <ClInclude Include="..\Src\Inputs\Inputs.h" />
     <ClInclude Include="..\Src\Inputs\InputSource.h" />

VS2008/Supermodel.vcxproj.filters

@@ -470,6 +470,9 @@
     <ClCompile Include="..\Src\Graphics\FBO.cpp">
       <Filter>Source Files\Graphics</Filter>
     </ClCompile>
+    <ClCompile Include="..\Src\Graphics\SuperAA.cpp">
+      <Filter>Source Files\Graphics</Filter>
+    </ClCompile>
   </ItemGroup>
   <ItemGroup>
     <MASM Include="..\Src\CPU\68K\Turbo68K\Turbo68K.asm">
@@ -856,6 +859,9 @@
     <ClInclude Include="..\Src\Graphics\New3D\R3DShaderCommon.h">
       <Filter>Header Files\Graphics\New</Filter>
     </ClInclude>
+    <ClInclude Include="..\Src\Graphics\SuperAA.h">
+      <Filter>Header Files\Graphics</Filter>
+    </ClInclude>
   </ItemGroup>
   <ItemGroup>
     <CustomBuild Include="..\Src\Debugger\ReadMe.txt">