1
0
Fork 0
forked from suyu/suyu

renderer_vulkan: isolate FXAA from blit screen

This commit is contained in:
Liam 2024-01-13 22:52:04 -05:00
parent 2b1dd3bef5
commit 9568b310be
12 changed files with 590 additions and 651 deletions

View file

@ -158,6 +158,14 @@ add_library(video_core STATIC
renderer_opengl/renderer_opengl.h
renderer_opengl/util_shaders.cpp
renderer_opengl/util_shaders.h
renderer_vulkan/present/fsr.cpp
renderer_vulkan/present/fsr.h
renderer_vulkan/present/fxaa.cpp
renderer_vulkan/present/fxaa.h
renderer_vulkan/present/smaa.cpp
renderer_vulkan/present/smaa.h
renderer_vulkan/present/util.cpp
renderer_vulkan/present/util.h
renderer_vulkan/blit_image.cpp
renderer_vulkan/blit_image.h
renderer_vulkan/fixed_pipeline_state.cpp
@ -184,8 +192,6 @@ add_library(video_core STATIC
renderer_vulkan/vk_descriptor_pool.h
renderer_vulkan/vk_fence_manager.cpp
renderer_vulkan/vk_fence_manager.h
renderer_vulkan/vk_fsr.cpp
renderer_vulkan/vk_fsr.h
renderer_vulkan/vk_graphics_pipeline.cpp
renderer_vulkan/vk_graphics_pipeline.h
renderer_vulkan/vk_master_semaphore.cpp
@ -206,8 +212,6 @@ add_library(video_core STATIC
renderer_vulkan/vk_scheduler.h
renderer_vulkan/vk_shader_util.cpp
renderer_vulkan/vk_shader_util.h
renderer_vulkan/vk_smaa.cpp
renderer_vulkan/vk_smaa.h
renderer_vulkan/vk_staging_buffer_pool.cpp
renderer_vulkan/vk_staging_buffer_pool.h
renderer_vulkan/vk_state_tracker.cpp

View file

@ -0,0 +1,27 @@
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
class Scheduler;
class AntiAliasPass {
public:
virtual ~AntiAliasPass() = default;
virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) = 0;
};
class NoAA final : public AntiAliasPass {
public:
virtual VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) {
return source_image_view;
}
};
} // namespace Vulkan

View file

@ -10,7 +10,7 @@
#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32_comp_spv.h"
#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16_comp_spv.h"
#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32_comp_spv.h"
#include "video_core/renderer_vulkan/vk_fsr.h"
#include "video_core/renderer_vulkan/present/fsr.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/vulkan_common/vulkan_device.h"

View file

@ -0,0 +1,144 @@
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include "common/common_types.h"
#include "video_core/host_shaders/fxaa_frag_spv.h"
#include "video_core/host_shaders/fxaa_vert_spv.h"
#include "video_core/renderer_vulkan/present/fxaa.h"
#include "video_core/renderer_vulkan/present/util.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/vulkan_common/vulkan_device.h"
namespace Vulkan {
FXAA::FXAA(const Device& device, MemoryAllocator& allocator, size_t image_count, VkExtent2D extent)
: m_device(device), m_allocator(allocator), m_extent(extent),
m_image_count(static_cast<u32>(image_count)) {
CreateImages();
CreateRenderPasses();
CreateSampler();
CreateShaders();
CreateDescriptorPool();
CreateDescriptorSetLayouts();
CreateDescriptorSets();
CreatePipelineLayouts();
CreatePipelines();
}
FXAA::~FXAA() = default;
void FXAA::CreateImages() {
for (u32 i = 0; i < m_image_count; i++) {
Image& image = m_dynamic_images.emplace_back();
image.image = CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT);
image.image_view =
CreateWrappedImageView(m_device, image.image, VK_FORMAT_R16G16B16A16_SFLOAT);
}
}
void FXAA::CreateRenderPasses() {
m_renderpass = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT);
for (auto& image : m_dynamic_images) {
image.framebuffer =
CreateWrappedFramebuffer(m_device, m_renderpass, image.image_view, m_extent);
}
}
void FXAA::CreateSampler() {
m_sampler = CreateWrappedSampler(m_device);
}
void FXAA::CreateShaders() {
m_vertex_shader = CreateWrappedShaderModule(m_device, FXAA_VERT_SPV);
m_fragment_shader = CreateWrappedShaderModule(m_device, FXAA_FRAG_SPV);
}
void FXAA::CreateDescriptorPool() {
// 2 descriptors, 1 descriptor set per image
m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 2 * m_image_count, m_image_count);
}
void FXAA::CreateDescriptorSetLayouts() {
m_descriptor_set_layout = CreateWrappedDescriptorSetLayout(m_device, 2);
}
void FXAA::CreateDescriptorSets() {
VkDescriptorSetLayout layout = *m_descriptor_set_layout;
for (auto& images : m_dynamic_images) {
images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, {layout});
}
}
void FXAA::CreatePipelineLayouts() {
m_pipeline_layout = CreateWrappedPipelineLayout(m_device, m_descriptor_set_layout);
}
void FXAA::CreatePipelines() {
m_pipeline = CreateWrappedPipeline(m_device, m_renderpass, m_pipeline_layout,
std::tie(m_vertex_shader, m_fragment_shader));
}
void FXAA::UpdateDescriptorSets(VkImageView image_view, size_t image_index) {
Image& image = m_dynamic_images[image_index];
std::vector<VkDescriptorImageInfo> image_infos;
std::vector<VkWriteDescriptorSet> updates;
image_infos.reserve(2);
updates.push_back(
CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, image.descriptor_sets[0], 0));
updates.push_back(
CreateWriteDescriptorSet(image_infos, *m_sampler, image_view, image.descriptor_sets[0], 1));
m_device.GetLogical().UpdateDescriptorSets(updates, {});
}
void FXAA::UploadImages(Scheduler& scheduler) {
if (m_images_ready) {
return;
}
scheduler.Record([&](vk::CommandBuffer cmdbuf) {
for (auto& image : m_dynamic_images) {
ClearColorImage(cmdbuf, *image.image);
}
});
scheduler.Finish();
m_images_ready = true;
}
VkImageView FXAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) {
const Image& image{m_dynamic_images[image_index]};
const VkImage output_image{*image.image};
const VkDescriptorSet descriptor_set{image.descriptor_sets[0]};
const VkFramebuffer framebuffer{*image.framebuffer};
const VkRenderPass renderpass{*m_renderpass};
const VkPipeline pipeline{*m_pipeline};
const VkPipelineLayout layout{*m_pipeline_layout};
const VkExtent2D extent{m_extent};
UploadImages(scheduler);
UpdateDescriptorSets(source_image_view, image_index);
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([=](vk::CommandBuffer cmdbuf) {
TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, renderpass, framebuffer, extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, pipeline);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, layout, 0, descriptor_set, {});
cmdbuf.Draw(4, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
});
return *image.image_view;
}
} // namespace Vulkan

View file

@ -0,0 +1,63 @@
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include "video_core/renderer_vulkan/present/anti_alias_pass.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
class Device;
class Scheduler;
class StagingBufferPool;
class FXAA final : public AntiAliasPass {
public:
explicit FXAA(const Device& device, MemoryAllocator& allocator, size_t image_count,
VkExtent2D extent);
~FXAA() override;
VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) override;
private:
void CreateImages();
void CreateRenderPasses();
void CreateSampler();
void CreateShaders();
void CreateDescriptorPool();
void CreateDescriptorSetLayouts();
void CreateDescriptorSets();
void CreatePipelineLayouts();
void CreatePipelines();
void UpdateDescriptorSets(VkImageView image_view, size_t image_index);
void UploadImages(Scheduler& scheduler);
const Device& m_device;
MemoryAllocator& m_allocator;
const VkExtent2D m_extent;
const u32 m_image_count;
vk::ShaderModule m_vertex_shader{};
vk::ShaderModule m_fragment_shader{};
vk::DescriptorPool m_descriptor_pool{};
vk::DescriptorSetLayout m_descriptor_set_layout{};
vk::PipelineLayout m_pipeline_layout{};
vk::Pipeline m_pipeline{};
vk::RenderPass m_renderpass{};
struct Image {
vk::DescriptorSets descriptor_sets{};
vk::Framebuffer framebuffer{};
vk::Image image{};
vk::ImageView image_view{};
};
std::vector<Image> m_dynamic_images{};
bool m_images_ready{};
vk::Sampler m_sampler{};
};
} // namespace Vulkan

View file

@ -0,0 +1,270 @@
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include <list>
#include "common/assert.h"
#include "common/polyfill_ranges.h"
#include "video_core/renderer_vulkan/present/smaa.h"
#include "video_core/renderer_vulkan/present/util.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/smaa_area_tex.h"
#include "video_core/smaa_search_tex.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/host_shaders/smaa_blending_weight_calculation_frag_spv.h"
#include "video_core/host_shaders/smaa_blending_weight_calculation_vert_spv.h"
#include "video_core/host_shaders/smaa_edge_detection_frag_spv.h"
#include "video_core/host_shaders/smaa_edge_detection_vert_spv.h"
#include "video_core/host_shaders/smaa_neighborhood_blending_frag_spv.h"
#include "video_core/host_shaders/smaa_neighborhood_blending_vert_spv.h"
namespace Vulkan {
SMAA::SMAA(const Device& device, MemoryAllocator& allocator, size_t image_count, VkExtent2D extent)
: m_device(device), m_allocator(allocator), m_extent(extent),
m_image_count(static_cast<u32>(image_count)) {
CreateImages();
CreateRenderPasses();
CreateSampler();
CreateShaders();
CreateDescriptorPool();
CreateDescriptorSetLayouts();
CreateDescriptorSets();
CreatePipelineLayouts();
CreatePipelines();
}
SMAA::~SMAA() = default;
void SMAA::CreateImages() {
static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT};
static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT};
m_static_images[Area] = CreateWrappedImage(m_allocator, area_extent, VK_FORMAT_R8G8_UNORM);
m_static_images[Search] = CreateWrappedImage(m_allocator, search_extent, VK_FORMAT_R8_UNORM);
m_static_image_views[Area] =
CreateWrappedImageView(m_device, m_static_images[Area], VK_FORMAT_R8G8_UNORM);
m_static_image_views[Search] =
CreateWrappedImageView(m_device, m_static_images[Search], VK_FORMAT_R8_UNORM);
for (u32 i = 0; i < m_image_count; i++) {
Images& images = m_dynamic_images.emplace_back();
images.images[Blend] =
CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT);
images.images[Edges] = CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16_SFLOAT);
images.images[Output] =
CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT);
images.image_views[Blend] =
CreateWrappedImageView(m_device, images.images[Blend], VK_FORMAT_R16G16B16A16_SFLOAT);
images.image_views[Edges] =
CreateWrappedImageView(m_device, images.images[Edges], VK_FORMAT_R16G16_SFLOAT);
images.image_views[Output] =
CreateWrappedImageView(m_device, images.images[Output], VK_FORMAT_R16G16B16A16_SFLOAT);
}
}
void SMAA::CreateRenderPasses() {
m_renderpasses[EdgeDetection] = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16_SFLOAT);
m_renderpasses[BlendingWeightCalculation] =
CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT);
m_renderpasses[NeighborhoodBlending] =
CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT);
for (auto& images : m_dynamic_images) {
images.framebuffers[EdgeDetection] = CreateWrappedFramebuffer(
m_device, m_renderpasses[EdgeDetection], images.image_views[Edges], m_extent);
images.framebuffers[BlendingWeightCalculation] =
CreateWrappedFramebuffer(m_device, m_renderpasses[BlendingWeightCalculation],
images.image_views[Blend], m_extent);
images.framebuffers[NeighborhoodBlending] = CreateWrappedFramebuffer(
m_device, m_renderpasses[NeighborhoodBlending], images.image_views[Output], m_extent);
}
}
void SMAA::CreateSampler() {
m_sampler = CreateWrappedSampler(m_device);
}
void SMAA::CreateShaders() {
// These match the order of the SMAAStage enum
static constexpr std::array vert_shader_sources{
ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_VERT_SPV),
ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_VERT_SPV),
ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_VERT_SPV),
};
static constexpr std::array frag_shader_sources{
ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_FRAG_SPV),
ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_FRAG_SPV),
ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_FRAG_SPV),
};
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_vertex_shaders[i] = CreateWrappedShaderModule(m_device, vert_shader_sources[i]);
m_fragment_shaders[i] = CreateWrappedShaderModule(m_device, frag_shader_sources[i]);
}
}
void SMAA::CreateDescriptorPool() {
// Edge detection: 1 descriptor
// Blending weight calculation: 3 descriptors
// Neighborhood blending: 2 descriptors
// 6 descriptors, 3 descriptor sets per image
m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 6 * m_image_count, 3 * m_image_count);
}
void SMAA::CreateDescriptorSetLayouts() {
m_descriptor_set_layouts[EdgeDetection] = CreateWrappedDescriptorSetLayout(m_device, 1);
m_descriptor_set_layouts[BlendingWeightCalculation] =
CreateWrappedDescriptorSetLayout(m_device, 3);
m_descriptor_set_layouts[NeighborhoodBlending] = CreateWrappedDescriptorSetLayout(m_device, 2);
}
void SMAA::CreateDescriptorSets() {
std::vector<VkDescriptorSetLayout> layouts(m_descriptor_set_layouts.size());
std::ranges::transform(m_descriptor_set_layouts, layouts.begin(),
[](auto& layout) { return *layout; });
for (auto& images : m_dynamic_images) {
images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, layouts);
}
}
void SMAA::CreatePipelineLayouts() {
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_pipeline_layouts[i] = CreateWrappedPipelineLayout(m_device, m_descriptor_set_layouts[i]);
}
}
void SMAA::CreatePipelines() {
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_pipelines[i] =
CreateWrappedPipeline(m_device, m_renderpasses[i], m_pipeline_layouts[i],
std::tie(m_vertex_shaders[i], m_fragment_shaders[i]));
}
}
void SMAA::UpdateDescriptorSets(VkImageView image_view, size_t image_index) {
Images& images = m_dynamic_images[image_index];
std::vector<VkDescriptorImageInfo> image_infos;
std::vector<VkWriteDescriptorSet> updates;
image_infos.reserve(6);
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view,
images.descriptor_sets[EdgeDetection], 0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Edges],
images.descriptor_sets[BlendingWeightCalculation],
0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Area],
images.descriptor_sets[BlendingWeightCalculation],
1));
updates.push_back(
CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Search],
images.descriptor_sets[BlendingWeightCalculation], 2));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view,
images.descriptor_sets[NeighborhoodBlending], 0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Blend],
images.descriptor_sets[NeighborhoodBlending], 1));
m_device.GetLogical().UpdateDescriptorSets(updates, {});
}
void SMAA::UploadImages(Scheduler& scheduler) {
if (m_images_ready) {
return;
}
static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT};
static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT};
UploadImage(m_device, m_allocator, scheduler, m_static_images[Area], area_extent,
VK_FORMAT_R8G8_UNORM, ARRAY_TO_SPAN(areaTexBytes));
UploadImage(m_device, m_allocator, scheduler, m_static_images[Search], search_extent,
VK_FORMAT_R8_UNORM, ARRAY_TO_SPAN(searchTexBytes));
scheduler.Record([&](vk::CommandBuffer cmdbuf) {
for (auto& images : m_dynamic_images) {
for (size_t i = 0; i < MaxDynamicImage; i++) {
ClearColorImage(cmdbuf, *images.images[i]);
}
}
});
scheduler.Finish();
m_images_ready = true;
}
VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) {
Images& images = m_dynamic_images[image_index];
VkImage output_image = *images.images[Output];
VkImage edges_image = *images.images[Edges];
VkImage blend_image = *images.images[Blend];
VkDescriptorSet edge_detection_descriptor_set = images.descriptor_sets[EdgeDetection];
VkDescriptorSet blending_weight_calculation_descriptor_set =
images.descriptor_sets[BlendingWeightCalculation];
VkDescriptorSet neighborhood_blending_descriptor_set =
images.descriptor_sets[NeighborhoodBlending];
VkFramebuffer edge_detection_framebuffer = *images.framebuffers[EdgeDetection];
VkFramebuffer blending_weight_calculation_framebuffer =
*images.framebuffers[BlendingWeightCalculation];
VkFramebuffer neighborhood_blending_framebuffer = *images.framebuffers[NeighborhoodBlending];
UploadImages(scheduler);
UpdateDescriptorSets(source_image_view, image_index);
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([=, this](vk::CommandBuffer cmdbuf) {
TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, *m_renderpasses[EdgeDetection], edge_detection_framebuffer,
m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[EdgeDetection]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[EdgeDetection], 0,
edge_detection_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, *m_renderpasses[BlendingWeightCalculation],
blending_weight_calculation_framebuffer, m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipelines[BlendingWeightCalculation]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[BlendingWeightCalculation], 0,
blending_weight_calculation_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, *m_renderpasses[NeighborhoodBlending],
neighborhood_blending_framebuffer, m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[NeighborhoodBlending]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[NeighborhoodBlending], 0,
neighborhood_blending_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
});
return *images.image_views[Output];
}
} // namespace Vulkan

View file

@ -4,6 +4,7 @@
#pragma once
#include <array>
#include "video_core/renderer_vulkan/present/anti_alias_pass.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
@ -13,12 +14,14 @@ class Device;
class Scheduler;
class StagingBufferPool;
class SMAA {
class SMAA final : public AntiAliasPass {
public:
explicit SMAA(const Device& device, MemoryAllocator& allocator, size_t image_count,
VkExtent2D extent);
~SMAA() override;
VkImageView Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view);
VkImageView source_image_view) override;
private:
enum SMAAStage {

View file

@ -1,29 +1,11 @@
// SPDX-FileCopyrightText: Copyright 2022 yuzu Emulator Project
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#include <list>
#include "common/assert.h"
#include "common/polyfill_ranges.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/renderer_vulkan/vk_smaa.h"
#include "video_core/smaa_area_tex.h"
#include "video_core/smaa_search_tex.h"
#include "video_core/vulkan_common/vulkan_device.h"
#include "video_core/host_shaders/smaa_blending_weight_calculation_frag_spv.h"
#include "video_core/host_shaders/smaa_blending_weight_calculation_vert_spv.h"
#include "video_core/host_shaders/smaa_edge_detection_frag_spv.h"
#include "video_core/host_shaders/smaa_edge_detection_vert_spv.h"
#include "video_core/host_shaders/smaa_neighborhood_blending_frag_spv.h"
#include "video_core/host_shaders/smaa_neighborhood_blending_vert_spv.h"
#include "video_core/renderer_vulkan/present/util.h"
namespace Vulkan {
namespace {
#define ARRAY_TO_SPAN(a) std::span(a, (sizeof(a) / sizeof(a[0])))
vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format) {
const VkImageCreateInfo image_ci{
@ -48,7 +30,7 @@ vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions,
}
void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayout target_layout,
VkImageLayout source_layout = VK_IMAGE_LAYOUT_GENERAL) {
VkImageLayout source_layout) {
constexpr VkFlags flags{VK_ACCESS_COLOR_ATTACHMENT_READ_BIT |
VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT | VK_ACCESS_SHADER_READ_BIT};
const VkImageMemoryBarrier barrier{
@ -75,7 +57,7 @@ void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayo
void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& scheduler,
vk::Image& image, VkExtent2D dimensions, VkFormat format,
std::span<const u8> initial_contents = {}) {
std::span<const u8> initial_contents) {
const VkBufferCreateInfo upload_ci = {
.sType = VK_STRUCTURE_TYPE_BUFFER_CREATE_INFO,
.pNext = nullptr,
@ -200,13 +182,13 @@ vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& r
});
}
vk::Sampler CreateWrappedSampler(const Device& device) {
vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter) {
return device.GetLogical().CreateSampler(VkSamplerCreateInfo{
.sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.magFilter = VK_FILTER_LINEAR,
.minFilter = VK_FILTER_LINEAR,
.magFilter = filter,
.minFilter = filter,
.mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR,
.addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE,
.addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE,
@ -471,12 +453,12 @@ void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image) {
cmdbuf.ClearColorImage(image, VK_IMAGE_LAYOUT_GENERAL, {}, subresources);
}
void BeginRenderPass(vk::CommandBuffer& cmdbuf, vk::RenderPass& render_pass,
VkFramebuffer framebuffer, VkExtent2D extent) {
void BeginRenderPass(vk::CommandBuffer& cmdbuf, VkRenderPass render_pass, VkFramebuffer framebuffer,
VkExtent2D extent) {
const VkRenderPassBeginInfo renderpass_bi{
.sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO,
.pNext = nullptr,
.renderPass = *render_pass,
.renderPass = render_pass,
.framebuffer = framebuffer,
.renderArea{
.offset{},
@ -503,248 +485,4 @@ void BeginRenderPass(vk::CommandBuffer& cmdbuf, vk::RenderPass& render_pass,
cmdbuf.SetScissor(0, scissor);
}
} // Anonymous namespace
SMAA::SMAA(const Device& device, MemoryAllocator& allocator, size_t image_count, VkExtent2D extent)
: m_device(device), m_allocator(allocator), m_extent(extent),
m_image_count(static_cast<u32>(image_count)) {
CreateImages();
CreateRenderPasses();
CreateSampler();
CreateShaders();
CreateDescriptorPool();
CreateDescriptorSetLayouts();
CreateDescriptorSets();
CreatePipelineLayouts();
CreatePipelines();
}
void SMAA::CreateImages() {
static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT};
static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT};
m_static_images[Area] = CreateWrappedImage(m_allocator, area_extent, VK_FORMAT_R8G8_UNORM);
m_static_images[Search] = CreateWrappedImage(m_allocator, search_extent, VK_FORMAT_R8_UNORM);
m_static_image_views[Area] =
CreateWrappedImageView(m_device, m_static_images[Area], VK_FORMAT_R8G8_UNORM);
m_static_image_views[Search] =
CreateWrappedImageView(m_device, m_static_images[Search], VK_FORMAT_R8_UNORM);
for (u32 i = 0; i < m_image_count; i++) {
Images& images = m_dynamic_images.emplace_back();
images.images[Blend] =
CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT);
images.images[Edges] = CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16_SFLOAT);
images.images[Output] =
CreateWrappedImage(m_allocator, m_extent, VK_FORMAT_R16G16B16A16_SFLOAT);
images.image_views[Blend] =
CreateWrappedImageView(m_device, images.images[Blend], VK_FORMAT_R16G16B16A16_SFLOAT);
images.image_views[Edges] =
CreateWrappedImageView(m_device, images.images[Edges], VK_FORMAT_R16G16_SFLOAT);
images.image_views[Output] =
CreateWrappedImageView(m_device, images.images[Output], VK_FORMAT_R16G16B16A16_SFLOAT);
}
}
void SMAA::CreateRenderPasses() {
m_renderpasses[EdgeDetection] = CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16_SFLOAT);
m_renderpasses[BlendingWeightCalculation] =
CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT);
m_renderpasses[NeighborhoodBlending] =
CreateWrappedRenderPass(m_device, VK_FORMAT_R16G16B16A16_SFLOAT);
for (auto& images : m_dynamic_images) {
images.framebuffers[EdgeDetection] = CreateWrappedFramebuffer(
m_device, m_renderpasses[EdgeDetection], images.image_views[Edges], m_extent);
images.framebuffers[BlendingWeightCalculation] =
CreateWrappedFramebuffer(m_device, m_renderpasses[BlendingWeightCalculation],
images.image_views[Blend], m_extent);
images.framebuffers[NeighborhoodBlending] = CreateWrappedFramebuffer(
m_device, m_renderpasses[NeighborhoodBlending], images.image_views[Output], m_extent);
}
}
void SMAA::CreateSampler() {
m_sampler = CreateWrappedSampler(m_device);
}
void SMAA::CreateShaders() {
// These match the order of the SMAAStage enum
static constexpr std::array vert_shader_sources{
ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_VERT_SPV),
ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_VERT_SPV),
ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_VERT_SPV),
};
static constexpr std::array frag_shader_sources{
ARRAY_TO_SPAN(SMAA_EDGE_DETECTION_FRAG_SPV),
ARRAY_TO_SPAN(SMAA_BLENDING_WEIGHT_CALCULATION_FRAG_SPV),
ARRAY_TO_SPAN(SMAA_NEIGHBORHOOD_BLENDING_FRAG_SPV),
};
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_vertex_shaders[i] = CreateWrappedShaderModule(m_device, vert_shader_sources[i]);
m_fragment_shaders[i] = CreateWrappedShaderModule(m_device, frag_shader_sources[i]);
}
}
void SMAA::CreateDescriptorPool() {
// Edge detection: 1 descriptor
// Blending weight calculation: 3 descriptors
// Neighborhood blending: 2 descriptors
// 6 descriptors, 3 descriptor sets per image
m_descriptor_pool = CreateWrappedDescriptorPool(m_device, 6 * m_image_count, 3 * m_image_count);
}
void SMAA::CreateDescriptorSetLayouts() {
m_descriptor_set_layouts[EdgeDetection] = CreateWrappedDescriptorSetLayout(m_device, 1);
m_descriptor_set_layouts[BlendingWeightCalculation] =
CreateWrappedDescriptorSetLayout(m_device, 3);
m_descriptor_set_layouts[NeighborhoodBlending] = CreateWrappedDescriptorSetLayout(m_device, 2);
}
void SMAA::CreateDescriptorSets() {
std::vector<VkDescriptorSetLayout> layouts(m_descriptor_set_layouts.size());
std::ranges::transform(m_descriptor_set_layouts, layouts.begin(),
[](auto& layout) { return *layout; });
for (auto& images : m_dynamic_images) {
images.descriptor_sets = CreateWrappedDescriptorSets(m_descriptor_pool, layouts);
}
}
void SMAA::CreatePipelineLayouts() {
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_pipeline_layouts[i] = CreateWrappedPipelineLayout(m_device, m_descriptor_set_layouts[i]);
}
}
void SMAA::CreatePipelines() {
for (size_t i = 0; i < MaxSMAAStage; i++) {
m_pipelines[i] =
CreateWrappedPipeline(m_device, m_renderpasses[i], m_pipeline_layouts[i],
std::tie(m_vertex_shaders[i], m_fragment_shaders[i]));
}
}
void SMAA::UpdateDescriptorSets(VkImageView image_view, size_t image_index) {
Images& images = m_dynamic_images[image_index];
std::vector<VkDescriptorImageInfo> image_infos;
std::vector<VkWriteDescriptorSet> updates;
image_infos.reserve(6);
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view,
images.descriptor_sets[EdgeDetection], 0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Edges],
images.descriptor_sets[BlendingWeightCalculation],
0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Area],
images.descriptor_sets[BlendingWeightCalculation],
1));
updates.push_back(
CreateWriteDescriptorSet(image_infos, *m_sampler, *m_static_image_views[Search],
images.descriptor_sets[BlendingWeightCalculation], 2));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, image_view,
images.descriptor_sets[NeighborhoodBlending], 0));
updates.push_back(CreateWriteDescriptorSet(image_infos, *m_sampler, *images.image_views[Blend],
images.descriptor_sets[NeighborhoodBlending], 1));
m_device.GetLogical().UpdateDescriptorSets(updates, {});
}
void SMAA::UploadImages(Scheduler& scheduler) {
if (m_images_ready) {
return;
}
static constexpr VkExtent2D area_extent{AREATEX_WIDTH, AREATEX_HEIGHT};
static constexpr VkExtent2D search_extent{SEARCHTEX_WIDTH, SEARCHTEX_HEIGHT};
UploadImage(m_device, m_allocator, scheduler, m_static_images[Area], area_extent,
VK_FORMAT_R8G8_UNORM, ARRAY_TO_SPAN(areaTexBytes));
UploadImage(m_device, m_allocator, scheduler, m_static_images[Search], search_extent,
VK_FORMAT_R8_UNORM, ARRAY_TO_SPAN(searchTexBytes));
scheduler.Record([&](vk::CommandBuffer cmdbuf) {
for (auto& images : m_dynamic_images) {
for (size_t i = 0; i < MaxDynamicImage; i++) {
ClearColorImage(cmdbuf, *images.images[i]);
}
}
});
scheduler.Finish();
m_images_ready = true;
}
VkImageView SMAA::Draw(Scheduler& scheduler, size_t image_index, VkImage source_image,
VkImageView source_image_view) {
Images& images = m_dynamic_images[image_index];
VkImage output_image = *images.images[Output];
VkImage edges_image = *images.images[Edges];
VkImage blend_image = *images.images[Blend];
VkDescriptorSet edge_detection_descriptor_set = images.descriptor_sets[EdgeDetection];
VkDescriptorSet blending_weight_calculation_descriptor_set =
images.descriptor_sets[BlendingWeightCalculation];
VkDescriptorSet neighborhood_blending_descriptor_set =
images.descriptor_sets[NeighborhoodBlending];
VkFramebuffer edge_detection_framebuffer = *images.framebuffers[EdgeDetection];
VkFramebuffer blending_weight_calculation_framebuffer =
*images.framebuffers[BlendingWeightCalculation];
VkFramebuffer neighborhood_blending_framebuffer = *images.framebuffers[NeighborhoodBlending];
UploadImages(scheduler);
UpdateDescriptorSets(source_image_view, image_index);
scheduler.RequestOutsideRenderPassOperationContext();
scheduler.Record([=, this](vk::CommandBuffer cmdbuf) {
TransitionImageLayout(cmdbuf, source_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, m_renderpasses[EdgeDetection], edge_detection_framebuffer,
m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[EdgeDetection]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[EdgeDetection], 0,
edge_detection_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, edges_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, m_renderpasses[BlendingWeightCalculation],
blending_weight_calculation_framebuffer, m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipelines[BlendingWeightCalculation]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[BlendingWeightCalculation], 0,
blending_weight_calculation_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, blend_image, VK_IMAGE_LAYOUT_GENERAL);
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
BeginRenderPass(cmdbuf, m_renderpasses[NeighborhoodBlending],
neighborhood_blending_framebuffer, m_extent);
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *m_pipelines[NeighborhoodBlending]);
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS,
*m_pipeline_layouts[NeighborhoodBlending], 0,
neighborhood_blending_descriptor_set, {});
cmdbuf.Draw(3, 1, 0, 0);
cmdbuf.EndRenderPass();
TransitionImageLayout(cmdbuf, output_image, VK_IMAGE_LAYOUT_GENERAL);
});
return *images.image_views[Output];
}
} // namespace Vulkan

View file

@ -0,0 +1,46 @@
// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project
// SPDX-License-Identifier: GPL-2.0-or-later
#pragma once
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/vulkan_common/vulkan_memory_allocator.h"
#include "video_core/vulkan_common/vulkan_wrapper.h"
namespace Vulkan {
#define ARRAY_TO_SPAN(a) std::span(a, (sizeof(a) / sizeof(a[0])))
vk::Image CreateWrappedImage(MemoryAllocator& allocator, VkExtent2D dimensions, VkFormat format);
void TransitionImageLayout(vk::CommandBuffer& cmdbuf, VkImage image, VkImageLayout target_layout,
VkImageLayout source_layout = VK_IMAGE_LAYOUT_GENERAL);
void UploadImage(const Device& device, MemoryAllocator& allocator, Scheduler& scheduler,
vk::Image& image, VkExtent2D dimensions, VkFormat format,
std::span<const u8> initial_contents = {});
void ClearColorImage(vk::CommandBuffer& cmdbuf, VkImage image);
vk::ImageView CreateWrappedImageView(const Device& device, vk::Image& image, VkFormat format);
vk::RenderPass CreateWrappedRenderPass(const Device& device, VkFormat format);
vk::Framebuffer CreateWrappedFramebuffer(const Device& device, vk::RenderPass& render_pass,
vk::ImageView& dest_image, VkExtent2D extent);
vk::Sampler CreateWrappedSampler(const Device& device, VkFilter filter = VK_FILTER_LINEAR);
vk::ShaderModule CreateWrappedShaderModule(const Device& device, std::span<const u32> code);
vk::DescriptorPool CreateWrappedDescriptorPool(const Device& device, u32 max_sampler_bindings,
u32 max_sets);
vk::DescriptorSetLayout CreateWrappedDescriptorSetLayout(const Device& device,
u32 max_sampler_bindings);
vk::DescriptorSets CreateWrappedDescriptorSets(vk::DescriptorPool& pool,
vk::Span<VkDescriptorSetLayout> layouts);
vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device,
vk::DescriptorSetLayout& layout);
vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass,
vk::PipelineLayout& layout,
std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders);
VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector<VkDescriptorImageInfo>& images,
VkSampler sampler, VkImageView view,
VkDescriptorSet set, u32 binding);
void BeginRenderPass(vk::CommandBuffer& cmdbuf, VkRenderPass render_pass, VkFramebuffer framebuffer,
VkExtent2D extent);
} // namespace Vulkan

View file

@ -16,20 +16,19 @@
#include "core/frontend/emu_window.h"
#include "video_core/gpu.h"
#include "video_core/host1x/gpu_device_memory_manager.h"
#include "video_core/host_shaders/fxaa_frag_spv.h"
#include "video_core/host_shaders/fxaa_vert_spv.h"
#include "video_core/host_shaders/present_bicubic_frag_spv.h"
#include "video_core/host_shaders/present_gaussian_frag_spv.h"
#include "video_core/host_shaders/vulkan_present_frag_spv.h"
#include "video_core/host_shaders/vulkan_present_scaleforce_fp16_frag_spv.h"
#include "video_core/host_shaders/vulkan_present_scaleforce_fp32_frag_spv.h"
#include "video_core/host_shaders/vulkan_present_vert_spv.h"
#include "video_core/renderer_vulkan/present/fsr.h"
#include "video_core/renderer_vulkan/present/fxaa.h"
#include "video_core/renderer_vulkan/present/smaa.h"
#include "video_core/renderer_vulkan/renderer_vulkan.h"
#include "video_core/renderer_vulkan/vk_blit_screen.h"
#include "video_core/renderer_vulkan/vk_fsr.h"
#include "video_core/renderer_vulkan/vk_scheduler.h"
#include "video_core/renderer_vulkan/vk_shader_util.h"
#include "video_core/renderer_vulkan/vk_smaa.h"
#include "video_core/renderer_vulkan/vk_swapchain.h"
#include "video_core/surface.h"
#include "video_core/textures/decoders.h"
@ -252,103 +251,17 @@ void BlitScreen::Draw(RasterizerVulkan& rasterizer, const Tegra::FramebufferConf
const auto anti_alias_pass = Settings::values.anti_aliasing.GetValue();
if (use_accelerated && anti_alias_pass == Settings::AntiAliasing::Fxaa) {
UpdateAADescriptorSet(source_image_view, false);
if (!fxaa) {
const u32 up_scale = Settings::values.resolution_info.up_scale;
const u32 down_shift = Settings::values.resolution_info.down_shift;
VkExtent2D size{
const VkExtent2D fxaa_size{
.width = (up_scale * framebuffer.width) >> down_shift,
.height = (up_scale * framebuffer.height) >> down_shift,
};
scheduler.Record([this, index = image_index, size,
anti_alias_pass](vk::CommandBuffer cmdbuf) {
const VkImageMemoryBarrier base_barrier{
.sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
.pNext = nullptr,
.srcAccessMask = 0,
.dstAccessMask = 0,
.oldLayout = VK_IMAGE_LAYOUT_GENERAL,
.newLayout = VK_IMAGE_LAYOUT_GENERAL,
.srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
.image = {},
.subresourceRange =
{
.aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
.baseMipLevel = 0,
.levelCount = 1,
.baseArrayLayer = 0,
.layerCount = 1,
},
};
{
VkImageMemoryBarrier fsr_write_barrier = base_barrier;
fsr_write_barrier.image = *aa_image;
fsr_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED;
cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, fsr_write_barrier);
fxaa = std::make_unique<FXAA>(device, memory_allocator, image_count, fxaa_size);
}
const f32 bg_red = Settings::values.bg_red.GetValue() / 255.0f;
const f32 bg_green = Settings::values.bg_green.GetValue() / 255.0f;
const f32 bg_blue = Settings::values.bg_blue.GetValue() / 255.0f;
const VkClearValue clear_color{
.color = {.float32 = {bg_red, bg_green, bg_blue, 1.0f}},
};
const VkRenderPassBeginInfo renderpass_bi{
.sType = VK_STRUCTURE_TYPE_RENDER_PASS_BEGIN_INFO,
.pNext = nullptr,
.renderPass = *aa_renderpass,
.framebuffer = *aa_framebuffer,
.renderArea =
{
.offset = {0, 0},
.extent = size,
},
.clearValueCount = 1,
.pClearValues = &clear_color,
};
const VkViewport viewport{
.x = 0.0f,
.y = 0.0f,
.width = static_cast<float>(size.width),
.height = static_cast<float>(size.height),
.minDepth = 0.0f,
.maxDepth = 1.0f,
};
const VkRect2D scissor{
.offset = {0, 0},
.extent = size,
};
cmdbuf.BeginRenderPass(renderpass_bi, VK_SUBPASS_CONTENTS_INLINE);
switch (anti_alias_pass) {
case Settings::AntiAliasing::Fxaa:
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *aa_pipeline);
break;
default:
cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, *aa_pipeline);
break;
}
cmdbuf.SetViewport(0, viewport);
cmdbuf.SetScissor(0, scissor);
cmdbuf.BindVertexBuffer(0, *buffer, offsetof(BufferData, vertices));
cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, *aa_pipeline_layout, 0,
aa_descriptor_sets[index], {});
cmdbuf.Draw(4, 1, 0, 0);
cmdbuf.EndRenderPass();
{
VkImageMemoryBarrier blit_read_barrier = base_barrier;
blit_read_barrier.image = *aa_image;
blit_read_barrier.srcAccessMask = VK_ACCESS_COLOR_ATTACHMENT_WRITE_BIT;
blit_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT;
cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COLOR_ATTACHMENT_OUTPUT_BIT,
VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, blit_read_barrier);
}
});
source_image_view = *aa_image_view;
source_image_view = fxaa->Draw(scheduler, image_index, source_image, source_image_view);
}
if (use_accelerated && anti_alias_pass == Settings::AntiAliasing::Smaa) {
if (!smaa) {
@ -496,6 +409,7 @@ void BlitScreen::CreateDynamicResources() {
CreateRenderPass();
CreateGraphicsPipeline();
fsr.reset();
fxaa.reset();
smaa.reset();
if (Settings::values.scaling_filter.GetValue() == Settings::ScalingFilter::Fsr) {
CreateFSR();
@ -520,6 +434,7 @@ void BlitScreen::RefreshResources(const Tegra::FramebufferConfig& framebuffer) {
raw_height = framebuffer.height;
pixel_format = framebuffer.pixel_format;
fxaa.reset();
smaa.reset();
ReleaseRawImages();
@ -529,8 +444,6 @@ void BlitScreen::RefreshResources(const Tegra::FramebufferConfig& framebuffer) {
void BlitScreen::CreateShaders() {
vertex_shader = BuildShader(device, VULKAN_PRESENT_VERT_SPV);
fxaa_vertex_shader = BuildShader(device, FXAA_VERT_SPV);
fxaa_fragment_shader = BuildShader(device, FXAA_FRAG_SPV);
bilinear_fragment_shader = BuildShader(device, VULKAN_PRESENT_FRAG_SPV);
bicubic_fragment_shader = BuildShader(device, PRESENT_BICUBIC_FRAG_SPV);
gaussian_fragment_shader = BuildShader(device, PRESENT_GAUSSIAN_FRAG_SPV);
@ -553,13 +466,6 @@ void BlitScreen::CreateDescriptorPool() {
},
}};
const std::array<VkDescriptorPoolSize, 1> pool_sizes_aa{{
{
.type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
.descriptorCount = static_cast<u32>(image_count * 2),
},
}};
const VkDescriptorPoolCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO,
.pNext = nullptr,
@ -569,16 +475,6 @@ void BlitScreen::CreateDescriptorPool() {
.pPoolSizes = pool_sizes.data(),
};
descriptor_pool = device.GetLogical().CreateDescriptorPool(ci);
const VkDescriptorPoolCreateInfo ci_aa{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.maxSets = static_cast<u32>(image_count),
.poolSizeCount = static_cast<u32>(pool_sizes_aa.size()),
.pPoolSizes = pool_sizes_aa.data(),
};
aa_descriptor_pool = device.GetLogical().CreateDescriptorPool(ci_aa);
}
void BlitScreen::CreateRenderPass() {
@ -659,23 +555,6 @@ void BlitScreen::CreateDescriptorSetLayout() {
},
}};
const std::array<VkDescriptorSetLayoutBinding, 2> layout_bindings_aa{{
{
.binding = 0,
.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_VERTEX_BIT,
.pImmutableSamplers = nullptr,
},
{
.binding = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
.descriptorCount = 1,
.stageFlags = VK_SHADER_STAGE_FRAGMENT_BIT,
.pImmutableSamplers = nullptr,
},
}};
const VkDescriptorSetLayoutCreateInfo ci{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
.pNext = nullptr,
@ -684,21 +563,11 @@ void BlitScreen::CreateDescriptorSetLayout() {
.pBindings = layout_bindings.data(),
};
const VkDescriptorSetLayoutCreateInfo ci_aa{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.bindingCount = static_cast<u32>(layout_bindings_aa.size()),
.pBindings = layout_bindings_aa.data(),
};
descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci);
aa_descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci_aa);
}
void BlitScreen::CreateDescriptorSets() {
const std::vector layouts(image_count, *descriptor_set_layout);
const std::vector layouts_aa(image_count, *aa_descriptor_set_layout);
const VkDescriptorSetAllocateInfo ai{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO,
@ -708,16 +577,7 @@ void BlitScreen::CreateDescriptorSets() {
.pSetLayouts = layouts.data(),
};
const VkDescriptorSetAllocateInfo ai_aa{
.sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO,
.pNext = nullptr,
.descriptorPool = *aa_descriptor_pool,
.descriptorSetCount = static_cast<u32>(image_count),
.pSetLayouts = layouts_aa.data(),
};
descriptor_sets = descriptor_pool.Allocate(ai);
aa_descriptor_sets = aa_descriptor_pool.Allocate(ai_aa);
}
void BlitScreen::CreatePipelineLayout() {
@ -730,17 +590,7 @@ void BlitScreen::CreatePipelineLayout() {
.pushConstantRangeCount = 0,
.pPushConstantRanges = nullptr,
};
const VkPipelineLayoutCreateInfo ci_aa{
.sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.setLayoutCount = 1,
.pSetLayouts = aa_descriptor_set_layout.address(),
.pushConstantRangeCount = 0,
.pPushConstantRanges = nullptr,
};
pipeline_layout = device.GetLogical().CreatePipelineLayout(ci);
aa_pipeline_layout = device.GetLogical().CreatePipelineLayout(ci_aa);
}
void BlitScreen::CreateGraphicsPipeline() {
@ -1068,8 +918,6 @@ void BlitScreen::ReleaseRawImages() {
scheduler.Wait(tick);
}
raw_images.clear();
aa_image_view.reset();
aa_image.reset();
buffer.reset();
}
@ -1150,198 +998,6 @@ void BlitScreen::CreateRawImages(const Tegra::FramebufferConfig& framebuffer) {
raw_images[i] = create_image();
raw_image_views[i] = create_image_view(raw_images[i]);
}
// AA Resources
const u32 up_scale = Settings::values.resolution_info.up_scale;
const u32 down_shift = Settings::values.resolution_info.down_shift;
aa_image = create_image(true, up_scale, down_shift);
aa_image_view = create_image_view(aa_image, true);
VkExtent2D size{
.width = (up_scale * framebuffer.width) >> down_shift,
.height = (up_scale * framebuffer.height) >> down_shift,
};
if (aa_renderpass) {
aa_framebuffer = CreateFramebuffer(*aa_image_view, size, aa_renderpass);
return;
}
aa_renderpass = CreateRenderPassImpl(VK_FORMAT_R16G16B16A16_SFLOAT);
aa_framebuffer = CreateFramebuffer(*aa_image_view, size, aa_renderpass);
const std::array<VkPipelineShaderStageCreateInfo, 2> fxaa_shader_stages{{
{
.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.stage = VK_SHADER_STAGE_VERTEX_BIT,
.module = *fxaa_vertex_shader,
.pName = "main",
.pSpecializationInfo = nullptr,
},
{
.sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.stage = VK_SHADER_STAGE_FRAGMENT_BIT,
.module = *fxaa_fragment_shader,
.pName = "main",
.pSpecializationInfo = nullptr,
},
}};
const auto vertex_binding_description = ScreenRectVertex::GetDescription();
const auto vertex_attrs_description = ScreenRectVertex::GetAttributes();
const VkPipelineVertexInputStateCreateInfo vertex_input_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_VERTEX_INPUT_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.vertexBindingDescriptionCount = 1,
.pVertexBindingDescriptions = &vertex_binding_description,
.vertexAttributeDescriptionCount = u32{vertex_attrs_description.size()},
.pVertexAttributeDescriptions = vertex_attrs_description.data(),
};
const VkPipelineInputAssemblyStateCreateInfo input_assembly_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_INPUT_ASSEMBLY_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.topology = VK_PRIMITIVE_TOPOLOGY_TRIANGLE_STRIP,
.primitiveRestartEnable = VK_FALSE,
};
const VkPipelineViewportStateCreateInfo viewport_state_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_VIEWPORT_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.viewportCount = 1,
.pViewports = nullptr,
.scissorCount = 1,
.pScissors = nullptr,
};
const VkPipelineRasterizationStateCreateInfo rasterization_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_RASTERIZATION_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.depthClampEnable = VK_FALSE,
.rasterizerDiscardEnable = VK_FALSE,
.polygonMode = VK_POLYGON_MODE_FILL,
.cullMode = VK_CULL_MODE_NONE,
.frontFace = VK_FRONT_FACE_CLOCKWISE,
.depthBiasEnable = VK_FALSE,
.depthBiasConstantFactor = 0.0f,
.depthBiasClamp = 0.0f,
.depthBiasSlopeFactor = 0.0f,
.lineWidth = 1.0f,
};
const VkPipelineMultisampleStateCreateInfo multisampling_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_MULTISAMPLE_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.rasterizationSamples = VK_SAMPLE_COUNT_1_BIT,
.sampleShadingEnable = VK_FALSE,
.minSampleShading = 0.0f,
.pSampleMask = nullptr,
.alphaToCoverageEnable = VK_FALSE,
.alphaToOneEnable = VK_FALSE,
};
const VkPipelineColorBlendAttachmentState color_blend_attachment{
.blendEnable = VK_FALSE,
.srcColorBlendFactor = VK_BLEND_FACTOR_ZERO,
.dstColorBlendFactor = VK_BLEND_FACTOR_ZERO,
.colorBlendOp = VK_BLEND_OP_ADD,
.srcAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
.dstAlphaBlendFactor = VK_BLEND_FACTOR_ZERO,
.alphaBlendOp = VK_BLEND_OP_ADD,
.colorWriteMask = VK_COLOR_COMPONENT_R_BIT | VK_COLOR_COMPONENT_G_BIT |
VK_COLOR_COMPONENT_B_BIT | VK_COLOR_COMPONENT_A_BIT,
};
const VkPipelineColorBlendStateCreateInfo color_blend_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_COLOR_BLEND_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.logicOpEnable = VK_FALSE,
.logicOp = VK_LOGIC_OP_COPY,
.attachmentCount = 1,
.pAttachments = &color_blend_attachment,
.blendConstants = {0.0f, 0.0f, 0.0f, 0.0f},
};
static constexpr std::array dynamic_states{
VK_DYNAMIC_STATE_VIEWPORT,
VK_DYNAMIC_STATE_SCISSOR,
};
const VkPipelineDynamicStateCreateInfo dynamic_state_ci{
.sType = VK_STRUCTURE_TYPE_PIPELINE_DYNAMIC_STATE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.dynamicStateCount = static_cast<u32>(dynamic_states.size()),
.pDynamicStates = dynamic_states.data(),
};
const VkGraphicsPipelineCreateInfo fxaa_pipeline_ci{
.sType = VK_STRUCTURE_TYPE_GRAPHICS_PIPELINE_CREATE_INFO,
.pNext = nullptr,
.flags = 0,
.stageCount = static_cast<u32>(fxaa_shader_stages.size()),
.pStages = fxaa_shader_stages.data(),
.pVertexInputState = &vertex_input_ci,
.pInputAssemblyState = &input_assembly_ci,
.pTessellationState = nullptr,
.pViewportState = &viewport_state_ci,
.pRasterizationState = &rasterization_ci,
.pMultisampleState = &multisampling_ci,
.pDepthStencilState = nullptr,
.pColorBlendState = &color_blend_ci,
.pDynamicState = &dynamic_state_ci,
.layout = *aa_pipeline_layout,
.renderPass = *aa_renderpass,
.subpass = 0,
.basePipelineHandle = 0,
.basePipelineIndex = 0,
};
// AA
aa_pipeline = device.GetLogical().CreateGraphicsPipeline(fxaa_pipeline_ci);
}
void BlitScreen::UpdateAADescriptorSet(VkImageView image_view, bool nn) const {
const VkDescriptorImageInfo image_info{
.sampler = nn ? *nn_sampler : *sampler,
.imageView = image_view,
.imageLayout = VK_IMAGE_LAYOUT_GENERAL,
};
const VkWriteDescriptorSet sampler_write{
.sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
.pNext = nullptr,
.dstSet = aa_descriptor_sets[image_index],
.dstBinding = 0,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
.pImageInfo = &image_info,
.pBufferInfo = nullptr,
.pTexelBufferView = nullptr,
};
const VkWriteDescriptorSet sampler_write_2{
.sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
.pNext = nullptr,
.dstSet = aa_descriptor_sets[image_index],
.dstBinding = 1,
.dstArrayElement = 0,
.descriptorCount = 1,
.descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
.pImageInfo = &image_info,
.pBufferInfo = nullptr,
.pTexelBufferView = nullptr,
};
device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, sampler_write_2}, {});
}
void BlitScreen::UpdateDescriptorSet(VkImageView image_view, bool nn) const {

View file

@ -34,6 +34,7 @@ namespace Vulkan {
class Device;
class FSR;
class FXAA;
class RasterizerVulkan;
class Scheduler;
class SMAA;
@ -96,7 +97,6 @@ private:
void CreateRawImages(const Tegra::FramebufferConfig& framebuffer);
void UpdateDescriptorSet(VkImageView image_view, bool nn) const;
void UpdateAADescriptorSet(VkImageView image_view, bool nn) const;
void SetUniformData(BufferData& data, const Layout::FramebufferLayout layout) const;
void SetVertexData(BufferData& data, const Tegra::FramebufferConfig& framebuffer,
const Layout::FramebufferLayout layout, u32 texture_width,
@ -119,8 +119,6 @@ private:
std::size_t image_index{};
vk::ShaderModule vertex_shader;
vk::ShaderModule fxaa_vertex_shader;
vk::ShaderModule fxaa_fragment_shader;
vk::ShaderModule bilinear_fragment_shader;
vk::ShaderModule bicubic_fragment_shader;
vk::ShaderModule gaussian_fragment_shader;
@ -128,7 +126,6 @@ private:
vk::DescriptorPool descriptor_pool;
vk::DescriptorSetLayout descriptor_set_layout;
vk::PipelineLayout pipeline_layout;
vk::Pipeline nearest_neighbor_pipeline;
vk::Pipeline bilinear_pipeline;
vk::Pipeline bicubic_pipeline;
vk::Pipeline gaussian_pipeline;
@ -145,16 +142,6 @@ private:
std::vector<vk::Image> raw_images;
std::vector<vk::ImageView> raw_image_views;
vk::DescriptorPool aa_descriptor_pool;
vk::DescriptorSetLayout aa_descriptor_set_layout;
vk::PipelineLayout aa_pipeline_layout;
vk::Pipeline aa_pipeline;
vk::RenderPass aa_renderpass;
vk::Framebuffer aa_framebuffer;
vk::DescriptorSets aa_descriptor_sets;
vk::Image aa_image;
vk::ImageView aa_image_view;
u32 raw_width = 0;
u32 raw_height = 0;
Service::android::PixelFormat pixel_format{};
@ -163,6 +150,7 @@ private:
std::unique_ptr<FSR> fsr;
std::unique_ptr<SMAA> smaa;
std::unique_ptr<FXAA> fxaa;
};
} // namespace Vulkan