1
0
mirror of https://github.com/TomHarte/CLK.git synced 2026-03-13 19:16:40 +00:00
Files
CLK/Outputs/OpenGL/Shaders/KernelShaders.cpp
2026-02-07 22:39:15 -05:00

359 lines
9.8 KiB
C++

//
// KernelShaders.cpp
// Clock Signal Kiosk
//
// Created by Thomas Harte on 03/02/2026.
// Copyright © 2026 Thomas Harte. All rights reserved.
//
#include "KernelShaders.hpp"
#include "CommonAtrributes.hpp"
#include "DirtyZone.hpp"
#include "Outputs/ScanTargets/FilterGenerator.hpp"
#include <cstring>
namespace {
constexpr char vertex_shader[] = R"glsl(
uniform float samplesPerLine;
uniform vec2 bufferSize;
in float zoneBegin;
in float zoneEnd;
#ifdef USES_COORDINATES
out vec2 coordinates[11];
#endif
void main(void) {
float lateral = float(gl_VertexID & 1);
float longitudinal = float((gl_VertexID & 2) >> 1);
float sampleY = bufferSize.y - mix(zoneBegin, zoneEnd, longitudinal);
float centreX = lateral * samplesPerLine;
// Factors here:
//
// (1) only 8 vec4 varyings are guaranteed to exist, which can be utilised as 16 vec2s.
// So there aren't enough to guarantee one varying per sample location;
// (2) the cost of dependent reads is negligible nowadays unless and until it obviates
// the cache.
//
// So the coordinates picked are a spread across the area being sampled to provide enough
// information that the GPU should be able to cache efficiently.
#ifdef USES_COORDINATES
coordinates[0] = vec2(centreX - 14.0, sampleY) / bufferSize; // for 15, 14, 13 [0, 1, 2]
coordinates[1] = vec2(centreX - 11.0, sampleY) / bufferSize; // for 12, 11, 10 [3, 4, 5]
coordinates[2] = vec2(centreX - 8.0, sampleY) / bufferSize; // for 9, 8, 7 [6, 7, 8]
coordinates[3] = vec2(centreX - 5.0, sampleY) / bufferSize; // for 6, 5, 4 [9, 10, 11]
coordinates[4] = vec2(centreX - 2.0, sampleY) / bufferSize; // for 3, 2, 1 [12, 13, 14]
coordinates[5] = vec2(centreX + 0.0, sampleY) / bufferSize; // Centre. [15]
coordinates[6] = vec2(centreX + 2.0, sampleY) / bufferSize; // 1, 2, 3 [16, 17, 18]
coordinates[7] = vec2(centreX + 5.0, sampleY) / bufferSize; // 4, 5, 6 [19, 20, 21]
coordinates[8] = vec2(centreX + 8.0, sampleY) / bufferSize; // 7, 8, 9 [22, 23, 24]
coordinates[9] = vec2(centreX + 11.0, sampleY) / bufferSize; // 10, 11, 12 [25, 26, 27]
coordinates[10] = vec2(centreX + 14.0, sampleY) / bufferSize; // 13, 14, 15 [28, 29, 30]
#endif
gl_Position = vec4(
(vec2(centreX, sampleY) / bufferSize - vec2(0.5)) * vec2(2.0),
0.0,
1.0
);
}
)glsl";
constexpr char coordinate_indexer[] = R"glsl(
#define KernelCentre 15
in vec2 coordinates[11];
uniform vec2 bufferSize;
#define offset(i) ((float(i) - 15.0) / bufferSize.x)
#define coordinate(i) (\
(i) == 1 ? coordinates[0] : \
(i) == 4 ? coordinates[1] : \
(i) == 7 ? coordinates[2] : \
(i) == 10 ? coordinates[3] : \
(i) == 13 ? coordinates[4] : \
(i) == 15 ? coordinates[5] : \
(i) == 17 ? coordinates[6] : \
(i) == 20 ? coordinates[7] : \
(i) == 23 ? coordinates[8] : \
(i) == 26 ? coordinates[9] : \
(i) == 29 ? coordinates[10] : \
coordinates[5] + vec2(offset(i), 0.0) \
)
#define coefficient(x) filterCoefficients[x > KernelCentre ? KernelCentre - (x - KernelCentre) : x]
)glsl";
constexpr char separation_fragment_shader[] = R"glsl(
uniform sampler2D source;
uniform vec2 filterCoefficients[16];
out vec4 outputColour;
void main(void) {
vec4 centre = texture(source, coordinate(15));
#define Sample(x) texture(source, coordinate(x)).r * coefficient(x)
vec2 channels =
Sample(0) + Sample(1) + Sample(2) + Sample(3) +
Sample(4) + Sample(5) + Sample(6) + Sample(7) +
Sample(8) + Sample(9) + Sample(10) + Sample(11) +
Sample(12) + Sample(13) + Sample(14) +
filterCoefficients[15] * centre.x +
Sample(16) + Sample(17) + Sample(18) +
Sample(19) + Sample(20) + Sample(21) + Sample(22) +
Sample(23) + Sample(24) + Sample(25) + Sample(26) +
Sample(27) + Sample(28) + Sample(29) + Sample(30);
#undef Sample
float colourAmplitude = centre.a;
float isColour = step(0.01, colourAmplitude);
float chromaScale = mix(1.0, colourAmplitude, isColour);
float lumaScale = mix(1.0, 1.0 - colourAmplitude * 2.0, isColour);
outputColour = vec4(
(channels.x - colourAmplitude) / lumaScale,
isColour * channels.y * (centre.yz / chromaScale) + vec2(0.5),
1.0
);
}
)glsl";
constexpr char demodulation_fragment_shader[] = R"glsl(
uniform sampler2D source;
uniform vec3 filterCoefficients[16];
uniform mat3 toRGB;
out vec4 outputColour;
void main(void) {
vec4 centre = texture(source, coordinate(15));
#define Sample(x) (texture(source, coordinate(x)).rgb - vec3(0.0, 0.5, 0.5)) * coefficient(x)
vec3 channels =
Sample(0) + Sample(1) + Sample(2) + Sample(3) +
Sample(4) + Sample(5) + Sample(6) + Sample(7) +
Sample(8) + Sample(9) + Sample(10) + Sample(11) +
Sample(12) + Sample(13) + Sample(14) +
filterCoefficients[15] * (centre.rgb - vec3(0.0, 0.5, 0.5)) +
Sample(16) + Sample(17) + Sample(18) +
Sample(19) + Sample(20) + Sample(21) + Sample(22) +
Sample(23) + Sample(24) + Sample(25) + Sample(26) +
Sample(27) + Sample(28) + Sample(29) + Sample(30);
#undef Sample
outputColour = vec4(
toRGB * channels,
1.0
);
}
)glsl";
constexpr char fill_fragment_shader[] = R"glsl(
uniform vec4 colour;
out vec4 outputColour;
void main(void) {
outputColour = colour;
}
)glsl";
}
using namespace Outputs::Display;
namespace {
void enable_vertex_attributes(
OpenGL::Shader &shader,
const OpenGL::VertexArray &vertex_array
) {
OpenGL::DirtyZone zone;
vertex_array.bind_all();
const auto enable = [&](const std::string &name, uint16_t &element) {
shader.enable_vertex_attribute_with_pointer(
name,
1,
GL_UNSIGNED_SHORT,
GL_FALSE,
sizeof(zone),
reinterpret_cast<void *>((reinterpret_cast<uint8_t *>(&element) - reinterpret_cast<uint8_t *>(&zone))),
1
);
};
enable("zoneBegin", zone.begin);
enable("zoneEnd", zone.end);
}
template <size_t> struct FilterElement;
template <> struct FilterElement<2> {
void set_luma(const float luma) { x = luma; }
void set_chroma(const float chroma) { y = chroma; }
float x, y;
};
template <> struct FilterElement<3> {
void set_luma(const float luma) { x = luma; }
void set_chroma(const float chroma) { y = z = chroma; }
float x, y, z;
};
void set_size_uniforms(
OpenGL::Shader &shader,
const int samples_per_line,
const int buffer_width,
const int buffer_height
) {
shader.set_uniform("samplesPerLine", float(samples_per_line));
shader.set_uniform("bufferSize", float(buffer_width), float(buffer_height));
}
template <size_t FilterSize>
void set_filter_uniforms(
OpenGL::Shader &shader,
const int samples_per_line,
const int buffer_width,
const int buffer_height,
const GLenum source_texture_unit,
const FilterGenerator::FilterPair filter
) {
set_size_uniforms(shader, samples_per_line, buffer_width, buffer_height);
shader.set_uniform("source", GLint(source_texture_unit - GL_TEXTURE0));
// Zip and provide the filter coefficients.
static_assert(FilterGenerator::MaxKernelSize <= 31);
FilterElement<FilterSize> elements[31]{};
filter.luma.copy_to(std::begin(elements), std::end(elements),
[](const auto iterator, const float coefficient) {
iterator->set_luma(coefficient);
}
);
filter.chroma.copy_to(std::begin(elements), std::end(elements),
[](const auto iterator, const float coefficient) {
iterator->set_chroma(coefficient);
}
);
float packaged_elements[31 * FilterSize];
static_assert(sizeof(packaged_elements) == sizeof(elements));
std::memcpy(packaged_elements, elements, sizeof(elements));
shader.set_uniform("filterCoefficients", FilterSize, 16, packaged_elements);
}
}
OpenGL::Shader OpenGL::separation_shader(
const OpenGL::API api,
const float per_line_subcarrier_frequency,
const int samples_per_line,
const int buffer_width,
const int buffer_height,
const VertexArray &vertex_array,
const GLenum source_texture_unit
) {
auto shader = OpenGL::Shader(
api,
std::string("#define USES_COORDINATES\n") + vertex_shader,
std::string(coordinate_indexer) + separation_fragment_shader,
dirty_zone_attributes()
);
enable_vertex_attributes(shader, vertex_array);
set_filter_uniforms<2>(
shader,
samples_per_line,
buffer_width,
buffer_height,
source_texture_unit,
FilterGenerator(
samples_per_line,
per_line_subcarrier_frequency,
FilterGenerator::DecodingPath::Composite
).separation_filter()
);
return shader;
}
OpenGL::Shader OpenGL::demodulation_shader(
const OpenGL::API api,
const ColourSpace colour_space,
const DisplayType display_type,
const float per_line_subcarrier_frequency,
const int samples_per_line,
const int buffer_width,
const int buffer_height,
const VertexArray &vertex_array,
const GLenum source_texture_unit
) {
auto shader = OpenGL::Shader(
api,
std::string("#define USES_COORDINATES\n") + vertex_shader,
std::string(coordinate_indexer) + demodulation_fragment_shader,
dirty_zone_attributes()
);
enable_vertex_attributes(shader, vertex_array);
set_filter_uniforms<3>(
shader,
samples_per_line,
buffer_width,
buffer_height,
source_texture_unit,
FilterGenerator(
samples_per_line,
per_line_subcarrier_frequency,
is_composite(display_type) ?
FilterGenerator::DecodingPath::Composite :
FilterGenerator::DecodingPath::SVideo
).demouldation_filter()
);
shader.set_uniform_matrix("toRGB", 3, false, to_rgb_matrix(colour_space).data());
return shader;
}
OpenGL::FillShader::FillShader(
const API api,
const int samples_per_line,
const int buffer_width,
const int buffer_height,
const VertexArray &vertex_array
) {
shader_ = OpenGL::Shader(
api,
vertex_shader,
fill_fragment_shader,
dirty_zone_attributes()
);
enable_vertex_attributes(shader_, vertex_array);
set_size_uniforms(shader_, samples_per_line, buffer_width, buffer_height);
}
void OpenGL::FillShader::bind(const float r, const float g, const float b, const float a) {
shader_.bind();
if(colour_[0] != r || colour_[1] != g || colour_[2] != b || colour_[3] != a) {
colour_[0] = r;
colour_[1] = g;
colour_[2] = b;
colour_[3] = a;
shader_.set_uniform("colour", r, g, b, a);
}
}