dolphin/Source/Core/VideoCommon/TextureConverterShaderGen.cpp
Stenzek f74dbc794c EFB2RAM: Apply copy filter as a float coefficient after sampling
Using 8-bit integer math here lead to precision loss for depth copies,
which broke various effects in games, e.g. lens flare in MK:DD.

It's unlikely the console implements this as a floating-point multiply
(fixed-point perhaps), but since we have the float round trip in our
EFB2RAM shaders anyway, it's not going to make things any worse. If we
do rewrite our shaders to use integer math completely, then it might be
worth switching this conversion back to integers.

However, the range of the values (format) should be known, or we should
expand all values out to 24-bits first.
2018-05-22 12:24:08 +10:00

288 lines
9.8 KiB
C++

// Copyright 2017 Dolphin Emulator Project
// Licensed under GPLv2+
// Refer to the license.txt file included.
#include <array>
#include <cstring>
#include "Common/Assert.h"
#include "Common/CommonTypes.h"
#include "VideoCommon/BPMemory.h"
#include "VideoCommon/TextureConverterShaderGen.h"
#include "VideoCommon/VideoCommon.h"
#include "VideoCommon/VideoConfig.h"
namespace TextureConversionShaderGen
{
TCShaderUid GetShaderUid(EFBCopyFormat dst_format, bool is_depth_copy, bool is_intensity,
bool scale_by_half, bool copy_filter)
{
TCShaderUid out;
UidData* uid_data = out.GetUidData<UidData>();
memset(uid_data, 0, sizeof(*uid_data));
uid_data->dst_format = dst_format;
uid_data->efb_has_alpha = bpmem.zcontrol.pixel_format == PEControl::RGBA6_Z24;
uid_data->is_depth_copy = is_depth_copy;
uid_data->is_intensity = is_intensity;
uid_data->scale_by_half = scale_by_half;
uid_data->copy_filter = copy_filter;
return out;
}
ShaderCode GenerateShader(APIType api_type, const UidData* uid_data)
{
const bool mono_depth = uid_data->is_depth_copy && g_ActiveConfig.bStereoEFBMonoDepth;
ShaderCode out;
if (api_type == APIType::OpenGL)
{
out.Write("SAMPLER_BINDING(9) uniform sampler2DArray samp9;\n"
"uniform float3 filter_coefficients;\n"
"uniform float gamma_rcp;\n"
"uniform float2 clamp_tb;\n"
"uniform float pixel_height;\n");
out.Write("float4 SampleEFB(float3 uv, float y_offset) {\n"
" return texture(samp9, float3(uv.x, clamp(uv.y - (y_offset * pixel_height), "
"clamp_tb.x, clamp_tb.y), %s));\n"
"}\n",
mono_depth ? "0.0" : "uv.z");
out.Write("#define uv0 f_uv0\n"
"in vec3 uv0;\n"
"out vec4 ocol0;\n"
"void main(){\n");
}
else if (api_type == APIType::Vulkan)
{
out.Write("UBO_BINDING(std140, 1) uniform PSBlock {\n"
" float3 filter_coefficients;\n"
" float gamma_rcp;\n"
" float2 clamp_tb;\n"
" float pixel_height;\n"
"};\n");
out.Write("SAMPLER_BINDING(0) uniform sampler2DArray samp0;\n");
out.Write("float4 SampleEFB(float3 uv, float y_offset) {\n"
" return texture(samp0, float3(uv.x, clamp(uv.y + (y_offset * pixel_height), "
"clamp_tb.x, clamp_tb.y), %s));\n"
"}\n",
mono_depth ? "0.0" : "uv.z");
out.Write("layout(location = 0) in vec3 uv0;\n"
"layout(location = 1) in vec4 col0;\n"
"layout(location = 0) out vec4 ocol0;"
"void main(){\n");
}
else if (api_type == APIType::D3D)
{
out.Write("Texture2DArray tex0 : register(t0);\n"
"SamplerState samp0 : register(s0);\n"
"uniform float3 filter_coefficients;\n"
"uniform float gamma_rcp;\n"
"uniform float2 clamp_tb;\n"
"uniform float pixel_height;\n\n");
out.Write("float4 SampleEFB(float3 uv, float y_offset) {\n"
" return tex0.Sample(samp0, float3(uv.x, clamp(uv.y + (y_offset * pixel_height), "
"clamp_tb.x, clamp_tb.y), %s));\n"
"}\n",
mono_depth ? "0.0" : "uv.z");
out.Write("void main(out float4 ocol0 : SV_Target,\n"
" in float4 pos : SV_Position,\n"
" in float3 uv0 : TEXCOORD0) {\n");
}
// The copy filter applies to both color and depth copies. This has been verified on hardware.
// The filter is only applied to the RGB channels, the alpha channel is left intact.
if (uid_data->copy_filter)
{
out.Write(" float4 prev_row = SampleEFB(uv0, -1.0f);\n"
" float4 current_row = SampleEFB(uv0, 0.0f);\n"
" float4 next_row = SampleEFB(uv0, 1.0f);\n"
" float4 texcol = float4(min(prev_row.rgb * filter_coefficients[0] +\n"
" current_row.rgb * filter_coefficients[1] +\n"
" next_row.rgb * filter_coefficients[2], \n"
" float3(1, 1, 1)), current_row.a);\n");
}
else
{
out.Write(
" float4 current_row = SampleEFB(uv0, 0.0f);\n"
" float4 texcol = float4(min(current_row.rgb * filter_coefficients[1], float3(1, 1, 1)),\n"
" current_row.a);\n");
}
if (uid_data->is_depth_copy)
{
if (api_type == APIType::D3D || api_type == APIType::Vulkan)
out.Write("texcol.x = 1.0 - texcol.x;\n");
out.Write(" int depth = int(texcol.x * 16777216.0);\n"
// Convert to Z24 format
" int4 workspace;\n"
" workspace.r = (depth >> 16) & 255;\n"
" workspace.g = (depth >> 8) & 255;\n"
" workspace.b = depth & 255;\n"
// Convert to Z4 format
" workspace.a = (depth >> 16) & 0xF0;\n"
// Normalize components to [0.0..1.0]
" texcol = float4(workspace) / 255.0;\n");
switch (uid_data->dst_format)
{
case EFBCopyFormat::R4: // Z4
out.Write(" ocol0 = texcol.aaaa;\n");
break;
case EFBCopyFormat::R8_0x1: // Z8
case EFBCopyFormat::R8: // Z8H
out.Write(" ocol0 = texcol.rrrr;\n");
break;
case EFBCopyFormat::RA8: // Z16
out.Write(" ocol0 = texcol.gggr;\n");
break;
case EFBCopyFormat::RG8: // Z16 (reverse order)
out.Write(" ocol0 = texcol.rrrg;\n");
break;
case EFBCopyFormat::RGBA8: // Z24X8
out.Write(" ocol0 = float4(texcol.rgb, 0.0);\n");
break;
case EFBCopyFormat::G8: // Z8M
out.Write(" ocol0 = texcol.gggg;\n");
break;
case EFBCopyFormat::B8: // Z8L
out.Write(" ocol0 = texcol.bbbb;\n");
break;
case EFBCopyFormat::GB8: // Z16L - copy lower 16 depth bits
// expected to be used as an IA8 texture (upper 8 bits stored as intensity, lower 8 bits
// stored as alpha)
// Used e.g. in Zelda: Skyward Sword
out.Write(" ocol0 = texcol.gggb;\n");
break;
default:
ERROR_LOG(VIDEO, "Unknown copy zbuf format: 0x%X", static_cast<int>(uid_data->dst_format));
out.Write(" ocol0 = float4(texcol.bgr, 0.0);\n");
break;
}
}
else if (uid_data->is_intensity)
{
bool has_four_bits =
(uid_data->dst_format == EFBCopyFormat::R4 || uid_data->dst_format == EFBCopyFormat::RA4);
bool has_alpha =
(uid_data->dst_format == EFBCopyFormat::RA4 || uid_data->dst_format == EFBCopyFormat::RA8);
switch (uid_data->dst_format)
{
case EFBCopyFormat::R4: // I4
case EFBCopyFormat::R8_0x1: // I8
case EFBCopyFormat::R8: // I8
case EFBCopyFormat::RA4: // IA4
case EFBCopyFormat::RA8: // IA8
if (has_four_bits)
out.Write(" texcol = float4(int4(texcol * 255.0) & 0xF0) * (1.0 / 240.0);\n");
// TODO - verify these coefficients
out.Write(" const float3 coefficients = float3(0.257, 0.504, 0.098);\n"
" float intensity = dot(texcol.rgb, coefficients) + 16.0 / 255.0;\n"
" ocol0 = float4(intensity, intensity, intensity, %s);\n",
has_alpha ? "texcol.a" : "intensity");
break;
default:
ERROR_LOG(VIDEO, "Unknown copy intensity format: 0x%X",
static_cast<int>(uid_data->dst_format));
out.Write(" ocol0 = texcol;\n");
break;
}
}
else
{
if (!uid_data->efb_has_alpha)
out.Write(" texcol.a = 1.0;\n");
switch (uid_data->dst_format)
{
case EFBCopyFormat::R4: // R4
out.Write(" float red = float(int(texcol.r * 255.0) & 0xF0) * (1.0 / 240.0);\n"
" ocol0 = float4(red, red, red, red);\n");
break;
case EFBCopyFormat::R8_0x1: // R8
case EFBCopyFormat::R8: // R8
out.Write(" ocol0 = texcol.rrrr;\n");
break;
case EFBCopyFormat::RA4: // RA4
out.Write(" float2 red_alpha = float2(int2(texcol.ra * 255.0) & 0xF0) * (1.0 / 240.0);\n"
" ocol0 = red_alpha.rrrg;\n");
break;
case EFBCopyFormat::RA8: // RA8
out.Write(" ocol0 = texcol.rrra;\n");
break;
case EFBCopyFormat::A8: // A8
out.Write(" ocol0 = texcol.aaaa;\n");
break;
case EFBCopyFormat::G8: // G8
out.Write(" ocol0 = texcol.gggg;\n");
break;
case EFBCopyFormat::B8: // B8
out.Write(" ocol0 = texcol.bbbb;\n");
break;
case EFBCopyFormat::RG8: // RG8
out.Write(" ocol0 = texcol.rrrg;\n");
break;
case EFBCopyFormat::GB8: // GB8
out.Write(" ocol0 = texcol.gggb;\n");
break;
case EFBCopyFormat::RGB565: // RGB565
out.Write(" float2 red_blue = float2(int2(texcol.rb * 255.0) & 0xF8) * (1.0 / 248.0);\n"
" float green = float(int(texcol.g * 255.0) & 0xFC) * (1.0 / 252.0);\n"
" ocol0 = float4(red_blue.r, green, red_blue.g, 1.0);\n");
break;
case EFBCopyFormat::RGB5A3: // RGB5A3
// TODO: The MSB controls whether we have RGB5 or RGB4A3, this selection
// will need to be implemented once we move away from floats.
out.Write(" float3 color = float3(int3(texcol.rgb * 255.0) & 0xF8) * (1.0 / 248.0);\n"
" float alpha = float(int(texcol.a * 255.0) & 0xE0) * (1.0 / 224.0);\n"
" ocol0 = float4(color, alpha);\n");
break;
case EFBCopyFormat::RGBA8: // RGBA8
out.Write(" ocol0 = texcol;\n");
break;
case EFBCopyFormat::XFB:
out.Write(" ocol0 = float4(pow(texcol.rgb, float3(gamma_rcp, gamma_rcp, gamma_rcp)), "
"texcol.a);\n");
break;
default:
ERROR_LOG(VIDEO, "Unknown copy color format: 0x%X", static_cast<int>(uid_data->dst_format));
out.Write(" ocol0 = texcol;\n");
break;
}
}
out.Write("}\n");
return out;
}
} // namespace TextureConversionShaderGen