3D: Add FXAA

--HG--
branch : multipass-stereo

code/nel/include/nel/3d/fxaa.h

@@ -46,6 +46,7 @@ namespace NL3D {
 class ITexture;
 class CTextureUser;
 class CPixelProgram;
+class CVertexProgram;
 
 /**
  * \brief CFXAA
@@ -66,8 +67,9 @@ private:
 	UDriver *m_Driver;
 
 	NL3D::UMaterial m_Mat;
-	NL3D::CVertexBuffer m_VB;
+	// NL3D::CVertexBuffer m_VB;
 	NLMISC::CQuadUV m_QuadUV;
+	CVertexProgram *m_VP;
 	CPixelProgram *m_PP;
 
 	uint m_Width;

code/nel/src/3d/bloom_effect.cpp

@@ -264,6 +264,7 @@ void CBloomEffect::applyBloom()
 	CRect rect1(0, 0, width, height);
 	CRect rect2(0, 0, _BlurWidth, _BlurHeight);
 	dru->stretchRect(_Scene, txt1, rect1, txt2, rect2);
+	_Driver->setMatrixMode2D11();
 
 	// horizontal blur pass
 	doBlur(true);
@@ -319,10 +320,7 @@ void CBloomEffect::applyBlur()
 	matObjectFinal->texConstantColor(0, constCoeff);
 
 	// display quad
-	UCamera	pCam = _Scene->getCam();
-	_Driver->setMatrixMode2D11();
 	_Driver->drawQuad(_BlurQuad, displayBlurMat);
-	_Driver->setMatrixMode3D(pCam);
 
 	// disable vertex program
 	drvInternal->activeVertexProgram(NULL);
@@ -404,15 +402,12 @@ void CBloomEffect::doBlur(bool horizontalBlur)
 	matObject->setTexture(3, startTexture);
 
 	// display
-	UCamera	pCam = _Scene->getCam();
-	_Driver->setMatrixMode2D11();
 	_Driver->drawQuad(_BlurQuad, _BlurMat);
 
 	// disable render target and vertex program
 	drvInternal->activeVertexProgram(NULL);
 	CTextureUser cu;
 	((CDriverUser *)_Driver)->setRenderTarget(cu, 0, 0, 0, 0);
-	_Driver->setMatrixMode3D(pCam);
 }
 
 }; // NL3D

code/nel/src/3d/fxaa.cpp

@@ -54,7 +54,7 @@ namespace {
 
 } /* anonymous namespace */
 
-CFXAA::CFXAA(NL3D::UDriver *driver) : m_Driver(driver), m_PP(NULL), m_Width(~0), m_Height(~0)
+CFXAA::CFXAA(NL3D::UDriver *driver) : m_Driver(driver), m_PP(NULL), m_VP(NULL), m_Width(~0), m_Height(~0)
 {
 	nldebug("3D: Create FXAA");
 
@@ -82,14 +82,54 @@ CFXAA::CFXAA(NL3D::UDriver *driver) : m_Driver(driver), m_PP(NULL), m_Width(~0),
 		}
 		if (!drv->compilePixelProgram(m_PP))
 		{
-			nlwarning("No supported pixel program for FXAA effect");
+			nlwarning("3D: No supported pixel program for FXAA effect");
 
 			delete m_PP;
 			m_PP = NULL;
 		}
+		else
+		{
+			nldebug("3D: FXAA pixel program available");
+		}
 	}
 
-	if (m_PP)
+	if (!m_PP)
+	{
+		return;
+	}
+
+	// create vp
+	{
+		m_VP = new CVertexProgram();
+		// nelvp
+		{
+			IProgram::CSource *source = new IProgram::CSource();
+			source->Features.MaterialFlags = CProgramFeatures::TextureStages;
+			source->Profile = IProgram::nelvp;
+			source->setSourcePtr(a_nelvp);
+			m_VP->addSource(source);
+		}
+		if (!drv->compileVertexProgram(m_VP))
+		{
+			nlwarning("3D: No supported vertex program for FXAA effect");
+
+			delete m_VP;
+			m_VP = NULL;
+			delete m_PP;
+			m_PP = NULL;
+		}
+		else
+		{
+			nldebug("3D: FXAA vertex program available");
+		}
+	}
+
+	if (!m_VP)
+	{
+		return;
+	}
+	
+	// create material and vb
 	{
 		m_Mat = m_Driver->createMaterial();
 		m_Mat.initUnlit();
@@ -113,28 +153,14 @@ CFXAA::CFXAA(NL3D::UDriver *driver) : m_Driver(driver), m_PP(NULL), m_Width(~0),
 		m_QuadUV.Uv2 = CUV(1.f, 1.f);
 		m_QuadUV.Uv3 = CUV(0.f,  1.f);
 
-		CVertexBuffer &vb = m_VB;
+		/*CVertexBuffer &vb = m_VB;
 		vb.clearValueEx();
 		vb.addValueEx(CVertexBuffer::Position, CVertexBuffer::Float3);
 		vb.addValueEx(CVertexBuffer::TexCoord0, CVertexBuffer::Float2);
 		vb.addValueEx(CVertexBuffer::TexCoord1, CVertexBuffer::Float4);
 		vb.initEx();
-		vb.setPreferredMemory(CVertexBuffer::AGPPreferred, false);
+		vb.setPreferredMemory(CVertexBuffer::RAMVolatile, false);
-		vb.setNumVertices(4);
+		vb.setNumVertices(4);*/
-		/*CVertexBufferReadWrite vba;
-		vb.lock(vba);
-		vba.setVertexCoord(0, 0.f, 0.f, 0.5f);
-		vba.setVertexCoord(1, 1.f, 0.f, 0.5f);
-		vba.setVertexCoord(2, 1.f, 1.f, 0.5f);
-		vba.setVertexCoord(3, 0.f, 1.f, 0.5f);
-		vba.setTexCoord(0, 0, 0.f, 0.f);
-		vba.setTexCoord(1, 0, 1.f, 0.f);
-		vba.setTexCoord(2, 0, 1.f, 1.f);
-		vba.setTexCoord(3, 0, 0.f, 1.f);*/
-		/*vba.setTexCoord(0, 1, 0.f, 0.f);
-		vba.setTexCoord(1, 1, 1.f, 0.f);
-		vba.setTexCoord(2, 1, 1.f, 1.f);
-		vba.setTexCoord(3, 1, 0.f, 1.f);*/
 	}
 }
 
@@ -147,6 +173,8 @@ CFXAA::~CFXAA()
 		m_Driver->deleteMaterial(m_Mat);
 	}
 
+	delete m_VP;
+	m_VP = NULL;
 	delete m_PP;
 	m_PP = NULL;
 
@@ -172,6 +200,33 @@ void CFXAA::applyEffect()
 
 	float fwidth = (float)width;
 	float fheight = (float)height;
+	nldebug("%f, %f", fwidth, fheight);
+	float pwidth = 1.0f / fwidth;
+	float pheight = 1.0f / fheight;
+	float hpwidth = pwidth * 0.5f;
+	float hpheight = pheight * 0.5f;
+	float n = 0.5f;
+
+	//if (width != m_Width || height != m_Height)
+	/*{
+		// Build VB
+		m_Width = width;
+		m_Height = height;
+		CVertexBufferReadWrite vba;
+		m_VB.lock(vba);
+		vba.setValueFloat3Ex(CVertexBuffer::Position, 0, 0.f, 0.f, 0.5f); // BL
+		vba.setValueFloat3Ex(CVertexBuffer::Position, 1, 1.f, 0.f, 0.5f); // BR
+		vba.setValueFloat3Ex(CVertexBuffer::Position, 2, 1.f, 1.f, 0.5f); // TR
+		vba.setValueFloat3Ex(CVertexBuffer::Position, 3, 0.f, 1.f, 0.5f); // TL
+		vba.setValueFloat2Ex(CVertexBuffer::TexCoord0, 0, 0.f, 0.f);
+		vba.setValueFloat2Ex(CVertexBuffer::TexCoord0, 1, 1.f, 0.f);
+		vba.setValueFloat2Ex(CVertexBuffer::TexCoord0, 2, 1.f, 1.f);
+		vba.setValueFloat2Ex(CVertexBuffer::TexCoord0, 3, 0.f, 1.f);
+		vba.setValueFloat4Ex(CVertexBuffer::TexCoord1, 0, 0.f - hpwidth, 0.f - hpheight, 0.f + hpwidth, 0.f + hpheight);
+		vba.setValueFloat4Ex(CVertexBuffer::TexCoord1, 1, 1.f - hpwidth, 0.f - hpheight, 1.f + hpwidth, 0.f + hpheight);
+		vba.setValueFloat4Ex(CVertexBuffer::TexCoord1, 2, 1.f - hpwidth, 1.f - hpheight, 1.f + hpwidth, 1.f + hpheight);
+		vba.setValueFloat4Ex(CVertexBuffer::TexCoord1, 3, 0.f - hpwidth, 1.f - hpheight, 0.f + hpwidth, 1.f + hpheight);
+	}*/
 
 	// create render target
 	CTextureUser *otherRenderTarget = m_Driver->getRenderTargetManager().getRenderTarget(width, height, mode2D);
@@ -182,11 +237,30 @@ void CFXAA::applyEffect()
 	drv->swapTextureHandle(*renderTarget, *otherRenderTarget->getITexture());
 	drv->setRenderTarget(renderTarget);
 
+	// debug
+	m_Driver->clearBuffers(CRGBA(128, 128, 128, 128));
+
+	// activate program
+	bool vpok = drv->activeVertexProgram(m_VP);
+	nlassert(vpok);
+	bool ppok = drv->activePixelProgram(m_PP);
+	nlassert(ppok);
+	drv->setUniform4f(IDriver::PixelProgram, 0, -n / fwidth, -n / fheight, n / fwidth, n / fheight); // fxaaConsoleRcpFrameOpt
+	drv->setUniform4f(IDriver::PixelProgram, 1, -2.0f / fwidth, -2.0f / fheight, 2.0f / fwidth, 2.0f / fheight); // fxaaConsoleRcpFrameOpt2
+	drv->setUniformMatrix(IDriver::VertexProgram, 0, IDriver::ModelViewProjection, IDriver::Identity);
+	drv->setUniform4f(IDriver::VertexProgram, 9, -hpwidth, -hpheight, hpwidth, hpheight);
+
 	// render effect
 	m_Mat.getObjectPtr()->setTexture(0, otherRenderTarget->getITexture());
+	/*drv->activeVertexBuffer(m_VB);
+	drv->renderRawQuads(*m_Mat.getObjectPtr(), 0, 1);*/
 	m_Driver->drawQuad(m_QuadUV, m_Mat);
 	m_Mat.getObjectPtr()->setTexture(0, NULL);
 
+	// deactivate program
+	drv->activeVertexProgram(NULL);
+	drv->activePixelProgram(NULL);
+
 	// recycle render target
 	m_Driver->getRenderTargetManager().recycleRenderTarget(otherRenderTarget);
 }

code/nel/src/3d/fxaa_program.h

@@ -1,6 +1,101 @@
 
+const char *a_nelvp =
+	"!!VP1.0\n"
+	"DP4 o[HPOS].x, c[0], v[OPOS];\n"
+	"DP4 o[HPOS].y, c[1], v[OPOS];\n"
+	"DP4 o[HPOS].z, c[2], v[OPOS];\n"
+	"DP4 o[HPOS].w, c[3], v[OPOS];\n"
+	"MOV o[TEX0].xy, v[TEX0];\n"
+	"ADD o[TEX1], v[TEX0].xyxy, c[9];\n"
+	"END\n";
+
+const char *a_arbfp1_test =
+	"!!ARBfp1.0\n"
+	"OPTION ARB_precision_hint_fastest;\n"
+	"TEX result.color, fragment.texcoord[1].zwzw, texture[0], 2D;\n"
+	"END\n";
+
 const char *a_arbfp1 = 
 	"!!ARBfp1.0\n"
+	"OPTION ARB_precision_hint_fastest;\n"
+	//# cgc version 3.1.0013, build date Apr 18 2012
+	//# command line args: -profile arbfp1 -O3 -fastmath -fastprecision
+	//# source file: fxaa_fp.cg
+	//#vendor NVIDIA Corporation
+	//#version 3.1.0.13
+	//#profile arbfp1
+	//#program fxaa_fp
+	//#semantic fxaa_fp.fxaaConsoleRcpFrameOpt
+	//#semantic fxaa_fp.fxaaConsoleRcpFrameOpt2
+	//#semantic fxaa_fp.nlTex0 : TEX0
+	//#var float2 pos : $vin.TEXCOORD0 : TEX0 : 0 : 1
+	//#var float4 fxaaConsolePosPos : $vin.TEXCOORD1 : TEX1 : 1 : 1
+	//#var float4 fxaaConsoleRcpFrameOpt :  : c[0] : 2 : 1
+	//#var float4 fxaaConsoleRcpFrameOpt2 :  : c[1] : 3 : 1
+	//#var sampler2D nlTex0 : TEX0 : texunit 0 : 4 : 1
+	//#var float4 oCol : $vout.COLOR : COL : 5 : 1
+	//#const c[2] = 0.125 0 -2 2
+	//#const c[3] = 0.001953125 0.5
+	"PARAM c[4] = { program.env[0..1],\n"
+	"		{ 0.125, 0, -2, 2 },\n"
+	"		{ 0.001953125, 0.5 } };\n"
+	"TEMP R0;\n"
+	"TEMP R1;\n"
+	"TEMP R2;\n"
+	"TEMP R3;\n"
+	"TEMP R4;\n"
+	"TEMP R5;\n"
+	"TEX R1.w, fragment.texcoord[1].zyzw, texture[0], 2D;\n"
+	"ADD R0.x, R1.w, c[3];\n"
+	"TEX R0.w, fragment.texcoord[1].xwzw, texture[0], 2D;\n"
+	"TEX R1.w, fragment.texcoord[1], texture[0], 2D;\n"
+	"ADD R0.y, -R0.x, R0.w;\n"
+	"ADD R0.z, R1.w, R0.y;\n"
+	"TEX R2.w, fragment.texcoord[1].zwzw, texture[0], 2D;\n"
+	"ADD R0.y, -R1.w, R0;\n"
+	"ADD R1.x, R2.w, R0.y;\n"
+	"ADD R1.y, R0.z, -R2.w;\n"
+	"MUL R2.xy, R1, R1;\n"
+	"ADD R0.y, R2.x, R2;\n"
+	"RSQ R0.y, R0.y;\n"
+	"MUL R2.xy, R0.y, R1;\n"
+	"MAD R3.xy, R2, c[0].zwzw, fragment.texcoord[0];\n"
+	"ABS R0.z, R2.y;\n"
+	"ABS R0.y, R2.x;\n"
+	"MIN R0.y, R0, R0.z;\n"
+	"RCP R0.y, R0.y;\n"
+	"MUL R1.xy, R0.y, R2;\n"
+	"MUL R1.xy, R1, c[2].x;\n"
+	"MIN R1.xy, R1, c[2].w;\n"
+	"TEX R4, R3, texture[0], 2D;\n"
+	"MAD R2.xy, -R2, c[0].zwzw, fragment.texcoord[0];\n"
+	"TEX R3, R2, texture[0], 2D;\n"
+	"ADD R3, R3, R4;\n"
+	"MAX R1.xy, R1, c[2].z;\n"
+	"MAD R2.xy, R1, c[1].zwzw, fragment.texcoord[0];\n"
+	"MUL R5, R3, c[3].y;\n"
+	"MAD R1.xy, -R1, c[1].zwzw, fragment.texcoord[0];\n"
+	"MIN R0.z, R0.x, R2.w;\n"
+	"MIN R0.y, R0.w, R1.w;\n"
+	"MIN R0.y, R0, R0.z;\n"
+	"MAX R0.z, R0.x, R2.w;\n"
+	"MAX R0.x, R0.w, R1.w;\n"
+	"MAX R0.x, R0, R0.z;\n"
+	"TEX R4, R2, texture[0], 2D;\n"
+	"TEX R3, R1, texture[0], 2D;\n"
+	"ADD R3, R3, R4;\n"
+	"MAD R3, R3, c[3].y, R5;\n"
+	"MUL R3, R3, c[3].y;\n"
+	"SLT R0.z, R0.x, R3.w;\n"
+	"SLT R0.x, R3.w, R0.y;\n"
+	"ADD_SAT R0.x, R0, R0.z;\n"
+	"CMP result.color, -R0.x, R5, R3;\n"
+	"END\n";
+	//# 45 instructions, 6 R-regs
+
+const char *a_arbfp1_earlyexit = 
+	"!!ARBfp1.0\n"
+	"OPTION ARB_precision_hint_fastest;\n"
 	//"# cgc version 3.1.0013, build date Apr 18 2012\n"
 	//"# command line args: -profile arbfp1\n"
 	//"# source file: fxaa_fp.cg\n"
@@ -82,7 +177,127 @@ const char *a_arbfp1 =
 	"END\n";
 	//"# 51 instructions, 6 R-regs\n"
 
+const char *a_ps_2_0_test_t0 =
+	"ps_2_x\n"
+	"dcl_2d s0\n"
+	"dcl t0.xyz\n"
+	"mov r0.xy, t0.xy\n"
+	"texld r0, r0, s0\n"
+	"mov oC0, r0\n";
+
+const char *a_ps_2_0_test_avg = 
+	"ps_2_x\n"
+	"dcl_2d s0\n"
+	"def c0, 0.25000000, 0, 0, 0\n"
+	"dcl t1\n"
+	"mov r0.xy, t1.xwzw\n"
+	"mov r1.xy, t1.zyzw\n"
+	"texld r0, r0, s0\n"
+	"texld r1, r1, s0\n"
+	"add r2, r1, r0\n"
+	"mov r0.xy, t1.zwzw\n"
+	"texld r1, t1, s0\n"
+	"texld r0, r0, s0\n"
+	"add r1, r2, r1\n"
+	"add r0, r1, r0\n"
+	"mul r0, r0, c0.x\n"
+	"mov oC0, r0\n";
+
 const char *a_ps_2_0 = 
+	"ps_2_0\n"
+	// cgc version 3.1.0013, build date Apr 18 2012
+	// command line args: -profile ps_2_0 -O3 -fastmath -fastprecision
+	// source file: fxaa_pp.cg
+	//vendor NVIDIA Corporation
+	//version 3.1.0.13
+	//profile ps_2_0
+	//program fxaa_pp
+	//semantic fxaa_pp.fxaaConsoleRcpFrameOpt
+	//semantic fxaa_pp.fxaaConsoleRcpFrameOpt2
+	//semantic fxaa_pp.nlTex0 : TEX0
+	//var float2 pos : $vin.TEXCOORD0 : TEX0 : 0 : 1
+	//var float4 fxaaConsolePosPos : $vin.TEXCOORD1 : TEX1 : 1 : 1
+	//var float4 fxaaConsoleRcpFrameOpt :  : c[0] : 2 : 1
+	//var float4 fxaaConsoleRcpFrameOpt2 :  : c[1] : 3 : 1
+	//var sampler2D nlTex0 : TEX0 : texunit 0 : 4 : 1
+	//var float4 oCol : $vout.COLOR : COL : 5 : 1
+	//const c[2] = 0.001953125 0.125 2 -2
+	//const c[3] = 0.5 0 1
+	"dcl_2d s0\n"
+	"def c2, 0.00195313, 0.12500000, 2.00000000, -2.00000000\n"
+	"def c3, 0.50000000, 0.00000000, 1.00000000, 0\n"
+	"dcl t1\n"
+	"dcl t0.xy\n"
+	"texld r5, t1, s0\n"
+	"mov r1.y, t1.w\n"
+	"mov r1.x, t1.z\n"
+	"mov r2.xy, r1\n"
+	"mov r0.y, t1.w\n"
+	"mov r0.x, t1\n"
+	"mov r1.y, t1\n"
+	"mov r1.x, t1.z\n"
+	"texld r1, r1, s0\n"
+	"texld r0, r0, s0\n"
+	"texld r6, r2, s0\n"
+	"add r0.x, r1.w, c2\n"
+	"add r2.x, -r0, r0.w\n"
+	"add r1.x, r5.w, r2\n"
+	"add r2.z, r1.x, -r6.w\n"
+	"add r2.x, -r5.w, r2\n"
+	"add r2.x, r6.w, r2\n"
+	"mov r3.x, r2\n"
+	"mov r3.y, r2.z\n"
+	"mov r2.y, r2.z\n"
+	"mov r1.y, r2.z\n"
+	"mov r1.x, r2\n"
+	"mul r1.xy, r3, r1\n"
+	"add r1.x, r1, r1.y\n"
+	"rsq r1.x, r1.x\n"
+	"mul r4.xy, r1.x, r2\n"
+	"abs r2.x, r4.y\n"
+	"abs r1.x, r4\n"
+	"min r1.x, r1, r2\n"
+	"rcp r1.x, r1.x\n"
+	"mul r1.xy, r1.x, r4\n"
+	"mul r1.xy, r1, c2.y\n"
+	"min r1.xy, r1, c2.z\n"
+	"max r2.xy, r1, c2.w\n"
+	"mov r1.y, c1.w\n"
+	"mov r1.x, c1.z\n"
+	"mad r3.xy, r2, r1, t0\n"
+	"mov r1.y, c1.w\n"
+	"mov r1.x, c1.z\n"
+	"mad r5.xy, -r2, r1, t0\n"
+	"mov r1.y, c0.w\n"
+	"mov r1.x, c0.z\n"
+	"mad r2.xy, -r4, r1, t0\n"
+	"mov r1.y, c0.w\n"
+	"mov r1.x, c0.z\n"
+	"mad r1.xy, r4, r1, t0\n"
+	"texld r4, r5, s0\n"
+	"texld r3, r3, s0\n"
+	"texld r1, r1, s0\n"
+	"texld r2, r2, s0\n"
+	"add r1, r2, r1\n"
+	"mul r2, r1, c3.x\n"
+	"add r1, r4, r3\n"
+	"max r3.x, r0, r6.w\n"
+	"mad r1, r1, c3.x, r2\n"
+	"mul r4, r1, c3.x\n"
+	"max r1.x, r0.w, r5.w\n"
+	"max r1.x, r1, r3\n"
+	"add r1.x, -r4.w, r1\n"
+	"min r3.x, r0.w, r5.w\n"
+	"min r0.x, r0, r6.w\n"
+	"min r0.x, r3, r0\n"
+	"add r0.x, r4.w, -r0\n"
+	"cmp r1.x, r1, c3.y, c3.z\n"
+	"cmp r0.x, r0, c3.y, c3.z\n"
+	"add_pp_sat r0.x, r0, r1\n"
+	"cmp r0, -r0.x, r4, r2\n"
+	"mov oC0, r0\n";
+
+const char *a_ps_2_0_earlyexit = 
 	"ps_2_x\n"
 	// cgc version 3.1.0013, build date Apr 18 2012
 	// command line args: -profile ps_2_x

code/nel/src/3d/shaders/compile.bat

@@ -0,0 +1,3 @@
+cgc -entry fxaa_pp fxaa_pp.cg -profile arbfp1 -O3 -fastmath -fastprecision -o fxaa_pp_arbfp1.txt
+cgc -entry fxaa_pp fxaa_pp.cg -profile ps_2_0 -O3 -fastmath -fastprecision -o fxaa_pp_ps_2_0.txt
+cgc -entry fxaa_vp fxaa_vp.cg -profile arbvp1 -fastmath -fastprecision -o fxaa_vp_arbvp1.txt

code/nel/src/3d/shaders/fxaa_pp.cg

@@ -0,0 +1,70 @@
+
+#define FXAA_PS3 1
+#define FXAA_HLSL_3 1
+#define FXAA_QUALITY__PRESET 12
+#define FXAA_EARLY_EXIT 0
+
+#define h4tex2Dlod tex2Dlod
+#define half4 float4
+#define half3 float3
+#define half2 float2
+#define half float
+
+#include "fxaa3_11.h"
+
+void fxaa_pp(
+	// Per fragment parameters
+	float2 pos : TEXCOORD0, 
+	float4 fxaaConsolePosPos : TEXCOORD1, 
+	
+	// Fragment program constants
+	uniform float4 fxaaConsoleRcpFrameOpt,
+	uniform float4 fxaaConsoleRcpFrameOpt2,
+	uniform sampler2D nlTex0 : TEX0, 
+	
+	// Output color
+	out float4 oCol : COLOR
+)
+{
+	oCol = FxaaPixelShader(
+		pos, 
+		fxaaConsolePosPos, 
+		nlTex0,
+		fxaaConsoleRcpFrameOpt,
+		fxaaConsoleRcpFrameOpt2
+		);
+}
+
+/*
+Have FXAA vertex shader run as a full screen triangle,
+and output "pos" and "fxaaConsolePosPos"
+such that inputs in the pixel shader provide,
+ 
+// {xy} = center of pixel
+FxaaFloat2 pos,
+ 
+// {xy__} = upper left of pixel
+// {__zw} = lower right of pixel
+FxaaFloat4 fxaaConsolePosPos,
+*/
+
+// fxaaConsoleRcpFrameOpt:
+// Only used on FXAA Console.
+// This must be from a constant/uniform.
+// This effects sub-pixel AA quality and inversely sharpness.
+// Where N ranges between,
+// N = 0.50 (default)
+// N = 0.33 (sharper)
+// {x___} = -N/screenWidthInPixels
+// {_y__} = -N/screenHeightInPixels
+// {__z_} = N/screenWidthInPixels
+// {___w} = N/screenHeightInPixels
+
+// fxaaConsoleRcpFrameOpt2:
+// Only used on FXAA Console.
+// Not used on 360, but used on PS3 and PC.
+// This must be from a constant/uniform.
+// {x___} = -2.0/screenWidthInPixels
+// {_y__} = -2.0/screenHeightInPixels
+// {__z_} = 2.0/screenWidthInPixels
+// {___w} = 2.0/screenHeightInPixels

code/nel/src/3d/shaders/fxaa_pp_arbfp1.txt

@@ -0,0 +1,76 @@
+!!ARBfp1.0
+OPTION ARB_precision_hint_fastest;
+# cgc version 3.1.0013, build date Apr 18 2012
+# command line args: -profile arbfp1 -O3 -fastmath -fastprecision
+# source file: fxaa_pp.cg
+#vendor NVIDIA Corporation
+#version 3.1.0.13
+#profile arbfp1
+#program fxaa_pp
+#semantic fxaa_pp.fxaaConsoleRcpFrameOpt
+#semantic fxaa_pp.fxaaConsoleRcpFrameOpt2
+#semantic fxaa_pp.nlTex0 : TEX0
+#var float2 pos : $vin.TEXCOORD0 : TEX0 : 0 : 1
+#var float4 fxaaConsolePosPos : $vin.TEXCOORD1 : TEX1 : 1 : 1
+#var float4 fxaaConsoleRcpFrameOpt :  : c[0] : 2 : 1
+#var float4 fxaaConsoleRcpFrameOpt2 :  : c[1] : 3 : 1
+#var sampler2D nlTex0 : TEX0 : texunit 0 : 4 : 1
+#var float4 oCol : $vout.COLOR : COL : 5 : 1
+#const c[2] = 0.125 0 -2 2
+#const c[3] = 0.001953125 0.5
+PARAM c[4] = { program.local[0..1],
+		{ 0.125, 0, -2, 2 },
+		{ 0.001953125, 0.5 } };
+TEMP R0;
+TEMP R1;
+TEMP R2;
+TEMP R3;
+TEMP R4;
+TEMP R5;
+TEX R1.w, fragment.texcoord[1].zyzw, texture[0], 2D;
+ADD R0.x, R1.w, c[3];
+TEX R0.w, fragment.texcoord[1].xwzw, texture[0], 2D;
+TEX R1.w, fragment.texcoord[1], texture[0], 2D;
+ADD R0.y, -R0.x, R0.w;
+ADD R0.z, R1.w, R0.y;
+TEX R2.w, fragment.texcoord[1].zwzw, texture[0], 2D;
+ADD R0.y, -R1.w, R0;
+ADD R1.x, R2.w, R0.y;
+ADD R1.y, R0.z, -R2.w;
+MUL R2.xy, R1, R1;
+ADD R0.y, R2.x, R2;
+RSQ R0.y, R0.y;
+MUL R2.xy, R0.y, R1;
+MAD R3.xy, R2, c[0].zwzw, fragment.texcoord[0];
+ABS R0.z, R2.y;
+ABS R0.y, R2.x;
+MIN R0.y, R0, R0.z;
+RCP R0.y, R0.y;
+MUL R1.xy, R0.y, R2;
+MUL R1.xy, R1, c[2].x;
+MIN R1.xy, R1, c[2].w;
+TEX R4, R3, texture[0], 2D;
+MAD R2.xy, -R2, c[0].zwzw, fragment.texcoord[0];
+TEX R3, R2, texture[0], 2D;
+ADD R3, R3, R4;
+MAX R1.xy, R1, c[2].z;
+MAD R2.xy, R1, c[1].zwzw, fragment.texcoord[0];
+MUL R5, R3, c[3].y;
+MAD R1.xy, -R1, c[1].zwzw, fragment.texcoord[0];
+MIN R0.z, R0.x, R2.w;
+MIN R0.y, R0.w, R1.w;
+MIN R0.y, R0, R0.z;
+MAX R0.z, R0.x, R2.w;
+MAX R0.x, R0.w, R1.w;
+MAX R0.x, R0, R0.z;
+TEX R4, R2, texture[0], 2D;
+TEX R3, R1, texture[0], 2D;
+ADD R3, R3, R4;
+MAD R3, R3, c[3].y, R5;
+MUL R3, R3, c[3].y;
+SLT R0.z, R0.x, R3.w;
+SLT R0.x, R3.w, R0.y;
+ADD_SAT R0.x, R0, R0.z;
+CMP result.color, -R0.x, R5, R3;
+END
+# 45 instructions, 6 R-regs

code/nel/src/3d/shaders/fxaa_pp_ps_2_0.txt

@@ -0,0 +1,92 @@
+ps_2_0
+// cgc version 3.1.0013, build date Apr 18 2012
+// command line args: -profile ps_2_0 -O3 -fastmath -fastprecision
+// source file: fxaa_pp.cg
+//vendor NVIDIA Corporation
+//version 3.1.0.13
+//profile ps_2_0
+//program fxaa_pp
+//semantic fxaa_pp.fxaaConsoleRcpFrameOpt
+//semantic fxaa_pp.fxaaConsoleRcpFrameOpt2
+//semantic fxaa_pp.nlTex0 : TEX0
+//var float2 pos : $vin.TEXCOORD0 : TEX0 : 0 : 1
+//var float4 fxaaConsolePosPos : $vin.TEXCOORD1 : TEX1 : 1 : 1
+//var float4 fxaaConsoleRcpFrameOpt :  : c[0] : 2 : 1
+//var float4 fxaaConsoleRcpFrameOpt2 :  : c[1] : 3 : 1
+//var sampler2D nlTex0 : TEX0 : texunit 0 : 4 : 1
+//var float4 oCol : $vout.COLOR : COL : 5 : 1
+//const c[2] = 0.001953125 0.125 2 -2
+//const c[3] = 0.5 0 1
+dcl_2d s0
+def c2, 0.00195313, 0.12500000, 2.00000000, -2.00000000
+def c3, 0.50000000, 0.00000000, 1.00000000, 0
+dcl t1
+dcl t0.xy
+texld r5, t1, s0
+mov r1.y, t1.w
+mov r1.x, t1.z
+mov r2.xy, r1
+mov r0.y, t1.w
+mov r0.x, t1
+mov r1.y, t1
+mov r1.x, t1.z
+texld r1, r1, s0
+texld r0, r0, s0
+texld r6, r2, s0
+add r0.x, r1.w, c2
+add r2.x, -r0, r0.w
+add r1.x, r5.w, r2
+add r2.z, r1.x, -r6.w
+add r2.x, -r5.w, r2
+add r2.x, r6.w, r2
+mov r3.x, r2
+mov r3.y, r2.z
+mov r2.y, r2.z
+mov r1.y, r2.z
+mov r1.x, r2
+mul r1.xy, r3, r1
+add r1.x, r1, r1.y
+rsq r1.x, r1.x
+mul r4.xy, r1.x, r2
+abs r2.x, r4.y
+abs r1.x, r4
+min r1.x, r1, r2
+rcp r1.x, r1.x
+mul r1.xy, r1.x, r4
+mul r1.xy, r1, c2.y
+min r1.xy, r1, c2.z
+max r2.xy, r1, c2.w
+mov r1.y, c1.w
+mov r1.x, c1.z
+mad r3.xy, r2, r1, t0
+mov r1.y, c1.w
+mov r1.x, c1.z
+mad r5.xy, -r2, r1, t0
+mov r1.y, c0.w
+mov r1.x, c0.z
+mad r2.xy, -r4, r1, t0
+mov r1.y, c0.w
+mov r1.x, c0.z
+mad r1.xy, r4, r1, t0
+texld r4, r5, s0
+texld r3, r3, s0
+texld r1, r1, s0
+texld r2, r2, s0
+add r1, r2, r1
+mul r2, r1, c3.x
+add r1, r4, r3
+max r3.x, r0, r6.w
+mad r1, r1, c3.x, r2
+mul r4, r1, c3.x
+max r1.x, r0.w, r5.w
+max r1.x, r1, r3
+add r1.x, -r4.w, r1
+min r3.x, r0.w, r5.w
+min r0.x, r0, r6.w
+min r0.x, r3, r0
+add r0.x, r4.w, -r0
+cmp r1.x, r1, c3.y, c3.z
+cmp r0.x, r0, c3.y, c3.z
+add_pp_sat r0.x, r0, r1
+cmp r0, -r0.x, r4, r2
+mov oC0, r0

code/nel/src/3d/shaders/fxaa_vp.cg

@@ -0,0 +1,20 @@
+
+void fxaa_vp(
+	// Per vertex parameters
+	float3 position : POSITION, 
+	float2 texCoord0 : TEXCOORD0, 
+	
+	// Vertex program constants
+	uniform float4x4 modelViewProjection,
+	uniform float4 fxaaConsolePosPos,
+	
+	// Output position
+	out float4 oPosition : POSITION, 
+	out float2 oTexCoord0 : TEXCOORD0, 
+	out float4 oTexCoord1 : TEXCOORD1
+)
+{
+	oPosition = mul(modelViewProjection, float4(position, 0.0));
+	oTexCoord0 = texCoord0;
+	oTexCoord1 = texCoord0.xyxy + fxaaConsolePosPos;
+}

code/nel/src/3d/shaders/fxaa_vp_arbvp1.txt

@@ -0,0 +1,31 @@
+!!ARBvp1.0
+# cgc version 3.1.0013, build date Apr 18 2012
+# command line args: -profile arbvp1 -fastmath -fastprecision
+# source file: fxaa_vp.cg
+#vendor NVIDIA Corporation
+#version 3.1.0.13
+#profile arbvp1
+#program fxaa_vp
+#semantic fxaa_vp.modelViewProjection
+#semantic fxaa_vp.fxaaConsolePosPos
+#var float3 position : $vin.POSITION : POSITION : 0 : 1
+#var float2 texCoord0 : $vin.TEXCOORD0 : TEXCOORD0 : 1 : 1
+#var float4x4 modelViewProjection :  : c[1], 4 : 2 : 1
+#var float4 fxaaConsolePosPos :  : c[5] : 3 : 1
+#var float4 oPosition : $vout.POSITION : HPOS : 4 : 1
+#var float2 oTexCoord0 : $vout.TEXCOORD0 : TEX0 : 5 : 1
+#var float4 oTexCoord1 : $vout.TEXCOORD1 : TEX1 : 6 : 1
+#const c[0] = 0
+PARAM c[6] = { { 0 },
+		program.local[1..5] };
+TEMP R0;
+MOV R0.w, c[0].x;
+MOV R0.xyz, vertex.position;
+DP4 result.position.w, R0, c[4];
+DP4 result.position.z, R0, c[3];
+DP4 result.position.y, R0, c[2];
+DP4 result.position.x, R0, c[1];
+ADD result.texcoord[1], vertex.texcoord[0].xyxy, c[5];
+MOV result.texcoord[0].xy, vertex.texcoord[0];
+END
+# 8 instructions, 1 R-regs

code/nel/src/3d/shaders/readme.txt

@@ -0,0 +1,4 @@
+Compiled shaders are embedded in the source.
+Must compile and re-embed manually.
+
+FXAA is in public domain.

code/ryzom/client/src/main_loop.cpp

@@ -1730,12 +1730,12 @@ bool mainLoop()
 					{
 						if (effectRender)
 						{
-							if (ClientCfg.Bloom)
+							if (StereoDisplay) Driver->setViewport(NL3D::CViewport());
-							{
+							UCamera	pCam = Scene->getCam();
-								if (StereoDisplay) Driver->setViewport(NL3D::CViewport());
+							Driver->setMatrixMode2D11();
-								CBloomEffect::instance().applyBloom();
+							if (ClientCfg.Bloom) CBloomEffect::instance().applyBloom();
-								if (StereoDisplay) Driver->setViewport(StereoDisplay->getCurrentViewport());
+							Driver->setMatrixMode3D(pCam);
-							}
+							if (StereoDisplay) Driver->setViewport(StereoDisplay->getCurrentViewport());
 							effectRender = false;
 						}
 

code/snowballs2/client/src/snowballs_client.cpp

@@ -801,8 +801,11 @@ void loopIngame()
 					if (effectRender)
 					{
 						if (StereoDisplay) Driver->setViewport(NL3D::CViewport());
-						if (s_EnableBloom) CBloomEffect::instance().applyBloom();
+						UCamera	pCam = Scene->getCam();
+						Driver->setMatrixMode2D11();
 						if (s_FXAA) s_FXAA->applyEffect();
+						if (s_EnableBloom) CBloomEffect::instance().applyBloom();
+						Driver->setMatrixMode3D(pCam);
 						if (StereoDisplay) Driver->setViewport(StereoDisplay->getCurrentViewport());
 						effectRender = false;
 					}