Swapped R and B channels in lut-default.png, and adjusted D3D9 LUT-application behaviour. (#11004) [Ryan Holtz]

* lut-default.png: Swapped red and blue channels (fixes GitHub #11001).
* render/d3d/d3dlsl.cpp: Changed screen LUT application to be applied during the color convolution pass on raster systems.

hlsl/color.fx

@@ -1,14 +1,23 @@
 // license:BSD-3-Clause
-// copyright-holders:Ryan Holtz
+// copyright-holders:Ryan Holtz, W. M. Martinez
 //-----------------------------------------------------------------------------
 // Color-Convolution Effect
 //-----------------------------------------------------------------------------
 
+//-----------------------------------------------------------------------------
+// Macros
+//-----------------------------------------------------------------------------
+
+#define LUT_TEXTURE_WIDTH 4096.0f
+#define LUT_SIZE 64.0f
+#define LUT_SCALE float2(1.0f / LUT_TEXTURE_WIDTH, 1.0f / LUT_SIZE)
+
 //-----------------------------------------------------------------------------
 // Sampler Definitions
 //-----------------------------------------------------------------------------
 
 texture Diffuse;
+texture LutTexture;
 
 sampler DiffuseSampler = sampler_state
 {
@@ -21,6 +30,35 @@ sampler DiffuseSampler = sampler_state
 	AddressW = CLAMP;
 };
 
+sampler2D LutSampler = sampler_state
+{
+	Texture = <LutTexture>;
+	MinFilter = LINEAR;
+	MagFilter = LINEAR;
+	MipFilter = LINEAR;
+	AddressU = CLAMP;
+	AddressV = CLAMP;
+	AddressW = CLAMP;
+};
+
+//-----------------------------------------------------------------------------
+// Utilities
+//-----------------------------------------------------------------------------
+
+float3 apply_lut(float3 color)
+{
+	// NOTE: Do not change the order of parameters here.
+	float3 lutcoord = float3((color.rg * (LUT_SIZE - 1.0f) + 0.5f) *
+		LUT_SCALE, color.b * (LUT_SIZE - 1.0f));
+	float shift = floor(lutcoord.z);
+
+	lutcoord.x += shift * LUT_SCALE.y;
+	color.rgb = lerp(tex2D(LutSampler, lutcoord.xy).rgb, tex2D(LutSampler,
+		float2(lutcoord.x + LUT_SCALE.y, lutcoord.y)).rgb,
+		lutcoord.z - shift);
+	return color;
+}
+
 //-----------------------------------------------------------------------------
 // Vertex Definitions
 //-----------------------------------------------------------------------------
@@ -83,11 +121,15 @@ uniform float3 BluRatios = float3(0.0f, 0.0f, 1.0f);
 uniform float3 Offset = float3(0.0f, 0.0f, 0.0f);
 uniform float3 Scale = float3(1.0f, 1.0f, 1.0f);
 uniform float Saturation = 1.0f;
+uniform bool LutEnable;
 
 float4 ps_main(PS_INPUT Input) : COLOR
 {
 	float4 BaseTexel = tex2D(DiffuseSampler, Input.TexCoord);
 
+	if (LutEnable)
+		BaseTexel.rgb = apply_lut(BaseTexel.rgb);
+
 	float3 OutRGB = BaseTexel.rgb;
 
 	// RGB Tint & Shift

hlsl/primary.fx

@@ -166,11 +166,7 @@ uniform bool UiLutEnable;
 
 float4 ps_screen_main(PS_INPUT Input) : COLOR
 {
-	float4 BaseTexel = tex2D(DiffuseSampler, Input.TexCoord);
-
-	if (LutEnable)
-		BaseTexel.rgb = apply_lut(BaseTexel.rgb);
-	return BaseTexel;
+	return tex2D(DiffuseSampler, Input.TexCoord);
 }
 
 float4 ps_vector_buffer_main(PS_INPUT Input) : COLOR

src/osd/modules/render/d3d/d3dhlsl.cpp

@@ -879,6 +879,7 @@ int shaders::create_resources()
 	color_effect->add_uniform("Scale", uniform::UT_VEC3, uniform::CU_COLOR_SCALE);
 	color_effect->add_uniform("Saturation", uniform::UT_FLOAT, uniform::CU_COLOR_SATURATION);
 	color_effect->add_uniform("SourceDims", uniform::UT_VEC2, uniform::CU_SOURCE_DIMS);
+	color_effect->add_uniform("LutEnable", uniform::UT_BOOL, uniform::CU_LUT_ENABLE);
 
 	deconverge_effect->add_uniform("ConvergeX", uniform::UT_VEC3, uniform::CU_CONVERGE_LINEAR_X);
 	deconverge_effect->add_uniform("ConvergeY", uniform::UT_VEC3, uniform::CU_CONVERGE_LINEAR_Y);
@@ -932,9 +933,6 @@ int shaders::create_resources()
 
 	prescale_point_effect->add_uniform("SourceDims", uniform::UT_VEC2, uniform::CU_SOURCE_DIMS);
 
-	default_effect->add_uniform("LutEnable", uniform::UT_BOOL, uniform::CU_LUT_ENABLE);
-	default_effect->add_uniform("UiLutEnable", uniform::UT_BOOL, uniform::CU_UI_LUT_ENABLE);
-
 	ui_effect->add_uniform("LutEnable", uniform::UT_BOOL, uniform::CU_LUT_ENABLE);
 	ui_effect->add_uniform("UiLutEnable", uniform::UT_BOOL, uniform::CU_UI_LUT_ENABLE);
 
@@ -942,7 +940,6 @@ int shaders::create_resources()
 	ui_wrap_effect->add_uniform("UiLutEnable", uniform::UT_BOOL, uniform::CU_UI_LUT_ENABLE);
 
 	vector_buffer_effect->add_uniform("LutEnable", uniform::UT_BOOL, uniform::CU_LUT_ENABLE);
-	vector_buffer_effect->add_uniform("UiLutEnable", uniform::UT_BOOL, uniform::CU_UI_LUT_ENABLE);
 
 	return 0;
 }
@@ -1159,6 +1156,8 @@ int shaders::color_convolution_pass(d3d_render_target *rt, int source_index, pol
 	uint32_t tint = (uint32_t)poly->tint();
 	float prim_tint[3] = { ((tint >> 16) & 0xff) / 255.0f, ((tint >> 8) & 0xff) / 255.0f, (tint & 0xff) / 255.0f };
 	curr_effect->set_vector("PrimTint", 3, prim_tint);
+	curr_effect->set_texture("LutTexture", !lut_texture ? nullptr : lut_texture->get_finaltex());
+	curr_effect->set_bool("UiLutEnable", false);
 
 	next_index = rt->next_index(next_index);
 	blit(rt->source_surface[next_index].Get(), false, D3DPT_TRIANGLELIST, 0, 2);
@@ -1467,6 +1466,7 @@ int shaders::vector_buffer_pass(d3d_render_target *rt, int source_index, poly_in
 
 	curr_effect->set_texture("Diffuse", rt->target_texture[next_index].Get());
 	curr_effect->set_texture("LutTexture", !lut_texture ? nullptr : lut_texture->get_finaltex());
+	curr_effect->set_bool("UiLutEnable", false);
 
 	// we need to clear the vector render target here
 	next_index = rt->next_index(next_index);
@@ -1486,7 +1486,9 @@ int shaders::screen_pass(d3d_render_target *rt, int source_index, poly_info *pol
 	curr_effect->update_uniforms();
 
 	curr_effect->set_texture("Diffuse", rt->target_texture[next_index].Get());
-	curr_effect->set_texture("LutTexture", !lut_texture ? nullptr : lut_texture->get_finaltex());
+	curr_effect->set_texture("LutTexture", nullptr);
+	curr_effect->set_bool("LutEnable", false);
+	curr_effect->set_bool("UiLutEnable", false);
 
 	blit(backbuffer.Get(), false, poly->type(), vertnum, poly->count());