diff options
Diffstat (limited to 'src/video_core')
51 files changed, 566 insertions, 179 deletions
diff --git a/src/video_core/CMakeLists.txt b/src/video_core/CMakeLists.txt index 55180f4b5..2de2beb6e 100644 --- a/src/video_core/CMakeLists.txt +++ b/src/video_core/CMakeLists.txt @@ -18,6 +18,7 @@ add_library(video_core STATIC buffer_cache/usage_tracker.h buffer_cache/word_manager.h cache_types.h + capture.h cdma_pusher.cpp cdma_pusher.h compatible_formats.cpp @@ -101,6 +102,7 @@ add_library(video_core STATIC memory_manager.cpp memory_manager.h precompiled_headers.h + present.h pte_kind.h query_cache/bank_base.h query_cache/query_base.h diff --git a/src/video_core/buffer_cache/buffer_cache.h b/src/video_core/buffer_cache/buffer_cache.h index 3818e00f4..ed7a5b27e 100644 --- a/src/video_core/buffer_cache/buffer_cache.h +++ b/src/video_core/buffer_cache/buffer_cache.h @@ -1130,7 +1130,7 @@ void BufferCache<P>::UpdateVertexBuffer(u32 index) { channel_state->vertex_buffers[index] = NULL_BINDING; return; } - if (!gpu_memory->IsWithinGPUAddressRange(gpu_addr_end)) { + if (!gpu_memory->IsWithinGPUAddressRange(gpu_addr_end) || size >= 64_MiB) { size = static_cast<u32>(gpu_memory->MaxContinuousRange(gpu_addr_begin, size)); } const BufferId buffer_id = FindBuffer(*device_addr, size); diff --git a/src/video_core/capture.h b/src/video_core/capture.h new file mode 100644 index 000000000..8db14a8ec --- /dev/null +++ b/src/video_core/capture.h @@ -0,0 +1,36 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/alignment.h" +#include "common/bit_util.h" +#include "common/common_types.h" +#include "core/frontend/framebuffer_layout.h" +#include "video_core/surface.h" + +namespace VideoCore::Capture { + +constexpr u32 BlockHeight = 4; +constexpr u32 BlockDepth = 0; +constexpr u32 BppLog2 = 2; + +constexpr auto PixelFormat = Surface::PixelFormat::B8G8R8A8_UNORM; + +constexpr auto LinearWidth = Layout::ScreenUndocked::Width; +constexpr auto LinearHeight = Layout::ScreenUndocked::Height; +constexpr auto LinearDepth = 1U; +constexpr auto BytesPerPixel = 4U; + +constexpr auto TiledWidth = LinearWidth; +constexpr auto TiledHeight = Common::AlignUpLog2(LinearHeight, BlockHeight + BlockDepth + BppLog2); +constexpr auto TiledSize = TiledWidth * TiledHeight * (1 << BppLog2); + +constexpr Layout::FramebufferLayout Layout{ + .width = LinearWidth, + .height = LinearHeight, + .screen = {0, 0, LinearWidth, LinearHeight}, + .is_srgb = false, +}; + +} // namespace VideoCore::Capture diff --git a/src/video_core/engines/maxwell_3d.cpp b/src/video_core/engines/maxwell_3d.cpp index a94e1f043..0d47b032c 100644 --- a/src/video_core/engines/maxwell_3d.cpp +++ b/src/video_core/engines/maxwell_3d.cpp @@ -291,7 +291,9 @@ u32 Maxwell3D::ProcessShadowRam(u32 method, u32 argument) { } void Maxwell3D::ConsumeSinkImpl() { - SCOPE_EXIT({ method_sink.clear(); }); + SCOPE_EXIT { + method_sink.clear(); + }; const auto control = shadow_state.shadow_ram_control; if (control == Regs::ShadowRamControl::Track || control == Regs::ShadowRamControl::TrackWithFilter) { diff --git a/src/video_core/fence_manager.h b/src/video_core/fence_manager.h index c3eda6893..2135f1f2d 100644 --- a/src/video_core/fence_manager.h +++ b/src/video_core/fence_manager.h @@ -197,7 +197,9 @@ private: MicroProfileOnThreadCreate(name.c_str()); // Cleanup - SCOPE_EXIT({ MicroProfileOnThreadExit(); }); + SCOPE_EXIT { + MicroProfileOnThreadExit(); + }; Common::SetCurrentThreadName(name.c_str()); Common::SetCurrentThreadPriority(Common::ThreadPriority::High); diff --git a/src/video_core/framebuffer_config.h b/src/video_core/framebuffer_config.h index 6a18b76fb..8b2a49de5 100644 --- a/src/video_core/framebuffer_config.h +++ b/src/video_core/framebuffer_config.h @@ -11,6 +11,12 @@ namespace Tegra { +enum class BlendMode { + Opaque, + Premultiplied, + Coverage, +}; + /** * Struct describing framebuffer configuration */ @@ -23,6 +29,7 @@ struct FramebufferConfig { Service::android::PixelFormat pixel_format{}; Service::android::BufferTransformFlags transform_flags{}; Common::Rectangle<int> crop_rect{}; + BlendMode blending{}; }; Common::Rectangle<f32> NormalizeCrop(const FramebufferConfig& framebuffer, u32 texture_width, diff --git a/src/video_core/gpu.cpp b/src/video_core/gpu.cpp index f4a5d831c..8e663f2a8 100644 --- a/src/video_core/gpu.cpp +++ b/src/video_core/gpu.cpp @@ -347,6 +347,17 @@ struct GPU::Impl { WaitForSyncOperation(wait_fence); } + std::vector<u8> GetAppletCaptureBuffer() { + std::vector<u8> out; + + const auto wait_fence = + RequestSyncOperation([&] { out = renderer->GetAppletCaptureBuffer(); }); + gpu_thread.TickGPU(); + WaitForSyncOperation(wait_fence); + + return out; + } + GPU& gpu; Core::System& system; Host1x::Host1x& host1x; @@ -505,6 +516,10 @@ void GPU::RequestComposite(std::vector<Tegra::FramebufferConfig>&& layers, impl->RequestComposite(std::move(layers), std::move(fences)); } +std::vector<u8> GPU::GetAppletCaptureBuffer() { + return impl->GetAppletCaptureBuffer(); +} + u64 GPU::GetTicks() const { return impl->GetTicks(); } diff --git a/src/video_core/gpu.h b/src/video_core/gpu.h index c4602ca37..ad535512c 100644 --- a/src/video_core/gpu.h +++ b/src/video_core/gpu.h @@ -215,6 +215,8 @@ public: void RequestComposite(std::vector<Tegra::FramebufferConfig>&& layers, std::vector<Service::Nvidia::NvFence>&& fences); + std::vector<u8> GetAppletCaptureBuffer(); + /// Performs any additional setup necessary in order to begin GPU emulation. /// This can be used to launch any necessary threads and register any necessary /// core timing events. diff --git a/src/video_core/gpu_thread.cpp b/src/video_core/gpu_thread.cpp index 58d8110b8..477e11457 100644 --- a/src/video_core/gpu_thread.cpp +++ b/src/video_core/gpu_thread.cpp @@ -22,7 +22,9 @@ static void RunThread(std::stop_token stop_token, Core::System& system, Tegra::Control::Scheduler& scheduler, SynchState& state) { std::string name = "GPU"; MicroProfileOnThreadCreate(name.c_str()); - SCOPE_EXIT({ MicroProfileOnThreadExit(); }); + SCOPE_EXIT { + MicroProfileOnThreadExit(); + }; Common::SetCurrentThreadName(name.c_str()); Common::SetCurrentThreadPriority(Common::ThreadPriority::Critical); diff --git a/src/video_core/host1x/ffmpeg/ffmpeg.cpp b/src/video_core/host1x/ffmpeg/ffmpeg.cpp index 96686da59..1003cd38d 100644 --- a/src/video_core/host1x/ffmpeg/ffmpeg.cpp +++ b/src/video_core/host1x/ffmpeg/ffmpeg.cpp @@ -273,10 +273,10 @@ DeinterlaceFilter::DeinterlaceFilter(const Frame& frame) { const AVFilter* buffer_sink = avfilter_get_by_name("buffersink"); AVFilterInOut* inputs = avfilter_inout_alloc(); AVFilterInOut* outputs = avfilter_inout_alloc(); - SCOPE_EXIT({ + SCOPE_EXIT { avfilter_inout_free(&inputs); avfilter_inout_free(&outputs); - }); + }; // Don't know how to get the accurate time_base but it doesn't matter for yadif filter // so just use 1/1 to make buffer filter happy diff --git a/src/video_core/host_shaders/fidelityfx_fsr.frag b/src/video_core/host_shaders/fidelityfx_fsr.frag index a266e1c4e..54eedb450 100644 --- a/src/video_core/host_shaders/fidelityfx_fsr.frag +++ b/src/video_core/host_shaders/fidelityfx_fsr.frag @@ -37,6 +37,7 @@ layout(set=0,binding=0) uniform sampler2D InputTexture; #define A_GPU 1 #define A_GLSL 1 +#define FSR_RCAS_PASSTHROUGH_ALPHA 1 #ifndef YUZU_USE_FP16 #include "ffx_a.h" @@ -71,9 +72,7 @@ layout(set=0,binding=0) uniform sampler2D InputTexture; #include "ffx_fsr1.h" -#if USE_RCAS - layout(location = 0) in vec2 frag_texcoord; -#endif +layout (location = 0) in vec2 frag_texcoord; layout (location = 0) out vec4 frag_color; void CurrFilter(AU2 pos) { @@ -81,22 +80,22 @@ void CurrFilter(AU2 pos) { #ifndef YUZU_USE_FP16 AF3 c; FsrEasuF(c, pos, Const0, Const1, Const2, Const3); - frag_color = AF4(c, 1.0); + frag_color = AF4(c, texture(InputTexture, frag_texcoord).a); #else AH3 c; FsrEasuH(c, pos, Const0, Const1, Const2, Const3); - frag_color = AH4(c, 1.0); + frag_color = AH4(c, texture(InputTexture, frag_texcoord).a); #endif #endif #if USE_RCAS #ifndef YUZU_USE_FP16 - AF3 c; - FsrRcasF(c.r, c.g, c.b, pos, Const0); - frag_color = AF4(c, 1.0); + AF4 c; + FsrRcasF(c.r, c.g, c.b, c.a, pos, Const0); + frag_color = c; #else - AH3 c; - FsrRcasH(c.r, c.g, c.b, pos, Const0); - frag_color = AH4(c, 1.0); + AH4 c; + FsrRcasH(c.r, c.g, c.b, c.a, pos, Const0); + frag_color = c; #endif #endif } diff --git a/src/video_core/host_shaders/fxaa.frag b/src/video_core/host_shaders/fxaa.frag index 9bffc20d5..192a602c1 100644 --- a/src/video_core/host_shaders/fxaa.frag +++ b/src/video_core/host_shaders/fxaa.frag @@ -71,5 +71,5 @@ vec3 FxaaPixelShader(vec4 posPos, sampler2D tex) { } void main() { - frag_color = vec4(FxaaPixelShader(posPos, input_texture), 1.0); + frag_color = vec4(FxaaPixelShader(posPos, input_texture), texture(input_texture, posPos.xy).a); } diff --git a/src/video_core/host_shaders/opengl_fidelityfx_fsr.frag b/src/video_core/host_shaders/opengl_fidelityfx_fsr.frag index 16d22f58e..fc47d3810 100644 --- a/src/video_core/host_shaders/opengl_fidelityfx_fsr.frag +++ b/src/video_core/host_shaders/opengl_fidelityfx_fsr.frag @@ -31,6 +31,7 @@ layout (location = 0) uniform uvec4 constants[4]; #define A_GPU 1 #define A_GLSL 1 +#define FSR_RCAS_PASSTHROUGH_ALPHA 1 #ifdef YUZU_USE_FP16 #define A_HALF @@ -67,9 +68,7 @@ layout (location = 0) uniform uvec4 constants[4]; #include "ffx_fsr1.h" -#if USE_RCAS - layout(location = 0) in vec2 frag_texcoord; -#endif +layout (location = 0) in vec2 frag_texcoord; layout (location = 0) out vec4 frag_color; void CurrFilter(AU2 pos) @@ -78,22 +77,22 @@ void CurrFilter(AU2 pos) #ifndef YUZU_USE_FP16 AF3 c; FsrEasuF(c, pos, constants[0], constants[1], constants[2], constants[3]); - frag_color = AF4(c, 1.0); + frag_color = AF4(c, texture(InputTexture, frag_texcoord).a); #else AH3 c; FsrEasuH(c, pos, constants[0], constants[1], constants[2], constants[3]); - frag_color = AH4(c, 1.0); + frag_color = AH4(c, texture(InputTexture, frag_texcoord).a); #endif #endif #if USE_RCAS #ifndef YUZU_USE_FP16 - AF3 c; - FsrRcasF(c.r, c.g, c.b, pos, constants[0]); - frag_color = AF4(c, 1.0); + AF4 c; + FsrRcasF(c.r, c.g, c.b, c.a, pos, constants[0]); + frag_color = c; #else AH3 c; - FsrRcasH(c.r, c.g, c.b, pos, constants[0]); - frag_color = AH4(c, 1.0); + FsrRcasH(c.r, c.g, c.b, c.a, pos, constants[0]); + frag_color = c; #endif #endif } diff --git a/src/video_core/host_shaders/opengl_present.frag b/src/video_core/host_shaders/opengl_present.frag index 5fd7ad297..096b4e4db 100644 --- a/src/video_core/host_shaders/opengl_present.frag +++ b/src/video_core/host_shaders/opengl_present.frag @@ -9,5 +9,5 @@ layout (location = 0) out vec4 color; layout (binding = 0) uniform sampler2D color_texture; void main() { - color = vec4(texture(color_texture, frag_tex_coord).rgb, 1.0f); + color = vec4(texture(color_texture, frag_tex_coord)); } diff --git a/src/video_core/host_shaders/present_bicubic.frag b/src/video_core/host_shaders/present_bicubic.frag index c814629cf..a9d9d40a3 100644 --- a/src/video_core/host_shaders/present_bicubic.frag +++ b/src/video_core/host_shaders/present_bicubic.frag @@ -52,5 +52,5 @@ vec4 textureBicubic( sampler2D textureSampler, vec2 texCoords ) { } void main() { - color = vec4(textureBicubic(color_texture, frag_tex_coord).rgb, 1.0f); + color = textureBicubic(color_texture, frag_tex_coord); } diff --git a/src/video_core/host_shaders/present_gaussian.frag b/src/video_core/host_shaders/present_gaussian.frag index ad9bb76a4..78edeb9b4 100644 --- a/src/video_core/host_shaders/present_gaussian.frag +++ b/src/video_core/host_shaders/present_gaussian.frag @@ -46,14 +46,14 @@ vec4 blurDiagonal(sampler2D textureSampler, vec2 coord, vec2 norm) { } void main() { - vec3 base = texture(color_texture, vec2(frag_tex_coord)).rgb * weight[0]; + vec4 base = texture(color_texture, vec2(frag_tex_coord)) * weight[0]; vec2 tex_offset = 1.0f / textureSize(color_texture, 0); // TODO(Blinkhawk): This code can be optimized through shader group instructions. - vec3 horizontal = blurHorizontal(color_texture, frag_tex_coord, tex_offset).rgb; - vec3 vertical = blurVertical(color_texture, frag_tex_coord, tex_offset).rgb; - vec3 diagonalA = blurDiagonal(color_texture, frag_tex_coord, tex_offset).rgb; - vec3 diagonalB = blurDiagonal(color_texture, frag_tex_coord, tex_offset * vec2(1.0, -1.0)).rgb; - vec3 combination = mix(mix(horizontal, vertical, 0.5f), mix(diagonalA, diagonalB, 0.5f), 0.5f); - color = vec4(combination + base, 1.0f); + vec4 horizontal = blurHorizontal(color_texture, frag_tex_coord, tex_offset); + vec4 vertical = blurVertical(color_texture, frag_tex_coord, tex_offset); + vec4 diagonalA = blurDiagonal(color_texture, frag_tex_coord, tex_offset); + vec4 diagonalB = blurDiagonal(color_texture, frag_tex_coord, tex_offset * vec2(1.0, -1.0)); + vec4 combination = mix(mix(horizontal, vertical, 0.5f), mix(diagonalA, diagonalB, 0.5f), 0.5f); + color = combination + base; } diff --git a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16.frag b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16.frag index d369bef06..05d033310 100644 --- a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16.frag +++ b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp16.frag @@ -6,5 +6,6 @@ #define YUZU_USE_FP16 #define USE_EASU 1 +#define VERSION 1 #include "fidelityfx_fsr.frag" diff --git a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32.frag b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32.frag index 6f25ef00f..7ae11dd66 100644 --- a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32.frag +++ b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu_fp32.frag @@ -5,5 +5,6 @@ #extension GL_GOOGLE_include_directive : enable #define USE_EASU 1 +#define VERSION 1 #include "fidelityfx_fsr.frag" diff --git a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16.frag b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16.frag index 0c953a900..c017214a5 100644 --- a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16.frag +++ b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp16.frag @@ -6,5 +6,6 @@ #define YUZU_USE_FP16 #define USE_RCAS 1 +#define VERSION 1 #include "fidelityfx_fsr.frag" diff --git a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32.frag b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32.frag index 02e9a27c6..976825f4b 100644 --- a/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32.frag +++ b/src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_fp32.frag @@ -5,5 +5,6 @@ #extension GL_GOOGLE_include_directive : enable #define USE_RCAS 1 +#define VERSION 1 #include "fidelityfx_fsr.frag" diff --git a/src/video_core/host_shaders/vulkan_present_scaleforce_fp16.frag b/src/video_core/host_shaders/vulkan_present_scaleforce_fp16.frag index 79ea817c2..cea5dac9d 100644 --- a/src/video_core/host_shaders/vulkan_present_scaleforce_fp16.frag +++ b/src/video_core/host_shaders/vulkan_present_scaleforce_fp16.frag @@ -5,7 +5,7 @@ #extension GL_GOOGLE_include_directive : enable -#define VERSION 1 +#define VERSION 2 #define YUZU_USE_FP16 #include "opengl_present_scaleforce.frag" diff --git a/src/video_core/host_shaders/vulkan_present_scaleforce_fp32.frag b/src/video_core/host_shaders/vulkan_present_scaleforce_fp32.frag index 9605bb58b..10ddf0401 100644 --- a/src/video_core/host_shaders/vulkan_present_scaleforce_fp32.frag +++ b/src/video_core/host_shaders/vulkan_present_scaleforce_fp32.frag @@ -5,6 +5,6 @@ #extension GL_GOOGLE_include_directive : enable -#define VERSION 1 +#define VERSION 2 #include "opengl_present_scaleforce.frag" diff --git a/src/video_core/macro/macro_hle.cpp b/src/video_core/macro/macro_hle.cpp index 46e853e04..fb529f88b 100644 --- a/src/video_core/macro/macro_hle.cpp +++ b/src/video_core/macro/macro_hle.cpp @@ -92,12 +92,12 @@ public: private: void Fallback(const std::vector<u32>& parameters) { - SCOPE_EXIT({ + SCOPE_EXIT { if (extended) { maxwell3d.engine_state = Maxwell3D::EngineHint::None; maxwell3d.replace_table.clear(); } - }); + }; maxwell3d.RefreshParameters(); const u32 instance_count = (maxwell3d.GetRegisterValue(0xD1B) & parameters[2]); @@ -281,12 +281,12 @@ public: private: void Fallback(const std::vector<u32>& parameters) { - SCOPE_EXIT({ + SCOPE_EXIT { // Clean everything. maxwell3d.regs.vertex_id_base = 0x0; maxwell3d.engine_state = Maxwell3D::EngineHint::None; maxwell3d.replace_table.clear(); - }); + }; maxwell3d.RefreshParameters(); const u32 start_indirect = parameters[0]; const u32 end_indirect = parameters[1]; diff --git a/src/video_core/present.h b/src/video_core/present.h new file mode 100644 index 000000000..4fdfcca68 --- /dev/null +++ b/src/video_core/present.h @@ -0,0 +1,37 @@ +// SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project +// SPDX-License-Identifier: GPL-2.0-or-later + +#pragma once + +#include "common/settings.h" + +static inline Settings::ScalingFilter GetScalingFilter() { + return Settings::values.scaling_filter.GetValue(); +} + +static inline Settings::AntiAliasing GetAntiAliasing() { + return Settings::values.anti_aliasing.GetValue(); +} + +static inline Settings::ScalingFilter GetScalingFilterForAppletCapture() { + return Settings::ScalingFilter::Bilinear; +} + +static inline Settings::AntiAliasing GetAntiAliasingForAppletCapture() { + return Settings::AntiAliasing::None; +} + +struct PresentFilters { + Settings::ScalingFilter (*get_scaling_filter)(); + Settings::AntiAliasing (*get_anti_aliasing)(); +}; + +constexpr PresentFilters PresentFiltersForDisplay{ + .get_scaling_filter = &GetScalingFilter, + .get_anti_aliasing = &GetAntiAliasing, +}; + +constexpr PresentFilters PresentFiltersForAppletCapture{ + .get_scaling_filter = &GetScalingFilterForAppletCapture, + .get_anti_aliasing = &GetAntiAliasingForAppletCapture, +}; diff --git a/src/video_core/renderer_base.h b/src/video_core/renderer_base.h index 3ad180f67..67427f937 100644 --- a/src/video_core/renderer_base.h +++ b/src/video_core/renderer_base.h @@ -40,6 +40,9 @@ public: /// Finalize rendering the guest frame and draw into the presentation texture virtual void Composite(std::span<const Tegra::FramebufferConfig> layers) = 0; + /// Get the tiled applet layer capture buffer + virtual std::vector<u8> GetAppletCaptureBuffer() = 0; + [[nodiscard]] virtual RasterizerInterface* ReadRasterizer() = 0; [[nodiscard]] virtual std::string GetDeviceVendor() const = 0; diff --git a/src/video_core/renderer_null/renderer_null.cpp b/src/video_core/renderer_null/renderer_null.cpp index c89daff53..e6147d66c 100644 --- a/src/video_core/renderer_null/renderer_null.cpp +++ b/src/video_core/renderer_null/renderer_null.cpp @@ -3,6 +3,7 @@ #include "core/frontend/emu_window.h" #include "core/frontend/graphics_context.h" +#include "video_core/capture.h" #include "video_core/renderer_null/renderer_null.h" namespace Null { @@ -22,4 +23,8 @@ void RendererNull::Composite(std::span<const Tegra::FramebufferConfig> framebuff render_window.OnFrameDisplayed(); } +std::vector<u8> RendererNull::GetAppletCaptureBuffer() { + return std::vector<u8>(VideoCore::Capture::TiledSize); +} + } // namespace Null diff --git a/src/video_core/renderer_null/renderer_null.h b/src/video_core/renderer_null/renderer_null.h index 063b476bb..34dbe1e4f 100644 --- a/src/video_core/renderer_null/renderer_null.h +++ b/src/video_core/renderer_null/renderer_null.h @@ -19,6 +19,8 @@ public: void Composite(std::span<const Tegra::FramebufferConfig> framebuffer) override; + std::vector<u8> GetAppletCaptureBuffer() override; + VideoCore::RasterizerInterface* ReadRasterizer() override { return &m_rasterizer; } diff --git a/src/video_core/renderer_opengl/gl_blit_screen.cpp b/src/video_core/renderer_opengl/gl_blit_screen.cpp index 6ba8b214b..9260a4dc4 100644 --- a/src/video_core/renderer_opengl/gl_blit_screen.cpp +++ b/src/video_core/renderer_opengl/gl_blit_screen.cpp @@ -2,6 +2,7 @@ // SPDX-License-Identifier: GPL-2.0-or-later #include "common/settings.h" +#include "video_core/present.h" #include "video_core/renderer_opengl/gl_blit_screen.h" #include "video_core/renderer_opengl/gl_state_tracker.h" #include "video_core/renderer_opengl/present/filters.h" @@ -13,14 +14,14 @@ namespace OpenGL { BlitScreen::BlitScreen(RasterizerOpenGL& rasterizer_, Tegra::MaxwellDeviceMemoryManager& device_memory_, StateTracker& state_tracker_, ProgramManager& program_manager_, - Device& device_) + Device& device_, const PresentFilters& filters_) : rasterizer(rasterizer_), device_memory(device_memory_), state_tracker(state_tracker_), - program_manager(program_manager_), device(device_) {} + program_manager(program_manager_), device(device_), filters(filters_) {} BlitScreen::~BlitScreen() = default; void BlitScreen::DrawScreen(std::span<const Tegra::FramebufferConfig> framebuffers, - const Layout::FramebufferLayout& layout) { + const Layout::FramebufferLayout& layout, bool invert_y) { // TODO: Signal state tracker about these changes state_tracker.NotifyScreenDrawVertexArray(); state_tracker.NotifyPolygonModes(); @@ -56,22 +57,22 @@ void BlitScreen::DrawScreen(std::span<const Tegra::FramebufferConfig> framebuffe glDepthRangeIndexed(0, 0.0, 0.0); while (layers.size() < framebuffers.size()) { - layers.emplace_back(rasterizer, device_memory); + layers.emplace_back(rasterizer, device_memory, filters); } CreateWindowAdapt(); - window_adapt->DrawToFramebuffer(program_manager, layers, framebuffers, layout); + window_adapt->DrawToFramebuffer(program_manager, layers, framebuffers, layout, invert_y); // TODO // program_manager.RestoreGuestPipeline(); } void BlitScreen::CreateWindowAdapt() { - if (window_adapt && Settings::values.scaling_filter.GetValue() == current_window_adapt) { + if (window_adapt && filters.get_scaling_filter() == current_window_adapt) { return; } - current_window_adapt = Settings::values.scaling_filter.GetValue(); + current_window_adapt = filters.get_scaling_filter(); switch (current_window_adapt) { case Settings::ScalingFilter::NearestNeighbor: window_adapt = MakeNearestNeighbor(device); diff --git a/src/video_core/renderer_opengl/gl_blit_screen.h b/src/video_core/renderer_opengl/gl_blit_screen.h index 0c3d838f1..df2da9424 100644 --- a/src/video_core/renderer_opengl/gl_blit_screen.h +++ b/src/video_core/renderer_opengl/gl_blit_screen.h @@ -15,6 +15,8 @@ namespace Layout { struct FramebufferLayout; } +struct PresentFilters; + namespace Tegra { struct FramebufferConfig; } @@ -46,12 +48,12 @@ public: explicit BlitScreen(RasterizerOpenGL& rasterizer, Tegra::MaxwellDeviceMemoryManager& device_memory, StateTracker& state_tracker, ProgramManager& program_manager, - Device& device); + Device& device, const PresentFilters& filters); ~BlitScreen(); /// Draws the emulated screens to the emulator window. void DrawScreen(std::span<const Tegra::FramebufferConfig> framebuffers, - const Layout::FramebufferLayout& layout); + const Layout::FramebufferLayout& layout, bool invert_y); private: void CreateWindowAdapt(); @@ -61,6 +63,7 @@ private: StateTracker& state_tracker; ProgramManager& program_manager; Device& device; + const PresentFilters& filters; Settings::ScalingFilter current_window_adapt{}; std::unique_ptr<WindowAdaptPass> window_adapt; diff --git a/src/video_core/renderer_opengl/gl_rasterizer.cpp b/src/video_core/renderer_opengl/gl_rasterizer.cpp index b42fb110c..16af8e6bd 100644 --- a/src/video_core/renderer_opengl/gl_rasterizer.cpp +++ b/src/video_core/renderer_opengl/gl_rasterizer.cpp @@ -230,7 +230,9 @@ template <typename Func> void RasterizerOpenGL::PrepareDraw(bool is_indexed, Func&& draw_func) { MICROPROFILE_SCOPE(OpenGL_Drawing); - SCOPE_EXIT({ gpu.TickWork(); }); + SCOPE_EXIT { + gpu.TickWork(); + }; gpu_memory->FlushCaching(); GraphicsPipeline* const pipeline{shader_cache.CurrentGraphicsPipeline()}; @@ -355,7 +357,9 @@ void RasterizerOpenGL::DrawIndirect() { void RasterizerOpenGL::DrawTexture() { MICROPROFILE_SCOPE(OpenGL_Drawing); - SCOPE_EXIT({ gpu.TickWork(); }); + SCOPE_EXIT { + gpu.TickWork(); + }; texture_cache.SynchronizeGraphicsDescriptors(); texture_cache.UpdateRenderTargets(false); diff --git a/src/video_core/renderer_opengl/present/layer.cpp b/src/video_core/renderer_opengl/present/layer.cpp index 8643e07c6..6c7092d22 100644 --- a/src/video_core/renderer_opengl/present/layer.cpp +++ b/src/video_core/renderer_opengl/present/layer.cpp @@ -2,6 +2,7 @@ // SPDX-License-Identifier: GPL-2.0-or-later #include "video_core/framebuffer_config.h" +#include "video_core/present.h" #include "video_core/renderer_opengl/gl_blit_screen.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/present/fsr.h" @@ -14,8 +15,9 @@ namespace OpenGL { -Layer::Layer(RasterizerOpenGL& rasterizer_, Tegra::MaxwellDeviceMemoryManager& device_memory_) - : rasterizer(rasterizer_), device_memory(device_memory_) { +Layer::Layer(RasterizerOpenGL& rasterizer_, Tegra::MaxwellDeviceMemoryManager& device_memory_, + const PresentFilters& filters_) + : rasterizer(rasterizer_), device_memory(device_memory_), filters(filters_) { // Allocate textures for the screen framebuffer_texture.resource.Create(GL_TEXTURE_2D); @@ -34,12 +36,12 @@ GLuint Layer::ConfigureDraw(std::array<GLfloat, 3 * 2>& out_matrix, std::array<ScreenRectVertex, 4>& out_vertices, ProgramManager& program_manager, const Tegra::FramebufferConfig& framebuffer, - const Layout::FramebufferLayout& layout) { + const Layout::FramebufferLayout& layout, bool invert_y) { FramebufferTextureInfo info = PrepareRenderTarget(framebuffer); auto crop = Tegra::NormalizeCrop(framebuffer, info.width, info.height); GLuint texture = info.display_texture; - auto anti_aliasing = Settings::values.anti_aliasing.GetValue(); + auto anti_aliasing = filters.get_anti_aliasing(); if (anti_aliasing != Settings::AntiAliasing::None) { glEnablei(GL_SCISSOR_TEST, 0); auto viewport_width = Settings::values.resolution_info.ScaleUp(framebuffer_texture.width); @@ -64,7 +66,7 @@ GLuint Layer::ConfigureDraw(std::array<GLfloat, 3 * 2>& out_matrix, glDisablei(GL_SCISSOR_TEST, 0); - if (Settings::values.scaling_filter.GetValue() == Settings::ScalingFilter::Fsr) { + if (filters.get_scaling_filter() == Settings::ScalingFilter::Fsr) { if (!fsr || fsr->NeedsRecreation(layout.screen)) { fsr = std::make_unique<FSR>(layout.screen.GetWidth(), layout.screen.GetHeight()); } @@ -83,10 +85,15 @@ GLuint Layer::ConfigureDraw(std::array<GLfloat, 3 * 2>& out_matrix, const auto w = screen.GetWidth(); const auto h = screen.GetHeight(); - out_vertices[0] = ScreenRectVertex(x, y, crop.left, crop.top); - out_vertices[1] = ScreenRectVertex(x + w, y, crop.right, crop.top); - out_vertices[2] = ScreenRectVertex(x, y + h, crop.left, crop.bottom); - out_vertices[3] = ScreenRectVertex(x + w, y + h, crop.right, crop.bottom); + const auto left = crop.left; + const auto right = crop.right; + const auto top = invert_y ? crop.bottom : crop.top; + const auto bottom = invert_y ? crop.top : crop.bottom; + + out_vertices[0] = ScreenRectVertex(x, y, left, top); + out_vertices[1] = ScreenRectVertex(x + w, y, right, top); + out_vertices[2] = ScreenRectVertex(x, y + h, left, bottom); + out_vertices[3] = ScreenRectVertex(x + w, y + h, right, bottom); return texture; } @@ -131,10 +138,12 @@ FramebufferTextureInfo Layer::LoadFBToScreenInfo(const Tegra::FramebufferConfig& const u64 size_in_bytes{Tegra::Texture::CalculateSize( true, bytes_per_pixel, framebuffer.stride, framebuffer.height, 1, block_height_log2, 0)}; const u8* const host_ptr{device_memory.GetPointer<u8>(framebuffer_addr)}; - const std::span<const u8> input_data(host_ptr, size_in_bytes); - Tegra::Texture::UnswizzleTexture(gl_framebuffer_data, input_data, bytes_per_pixel, - framebuffer.width, framebuffer.height, 1, block_height_log2, - 0); + if (host_ptr) { + const std::span<const u8> input_data(host_ptr, size_in_bytes); + Tegra::Texture::UnswizzleTexture(gl_framebuffer_data, input_data, bytes_per_pixel, + framebuffer.width, framebuffer.height, 1, + block_height_log2, 0); + } glBindBuffer(GL_PIXEL_UNPACK_BUFFER, 0); glPixelStorei(GL_UNPACK_ROW_LENGTH, static_cast<GLint>(framebuffer.stride)); diff --git a/src/video_core/renderer_opengl/present/layer.h b/src/video_core/renderer_opengl/present/layer.h index ef1055abf..5b15b730f 100644 --- a/src/video_core/renderer_opengl/present/layer.h +++ b/src/video_core/renderer_opengl/present/layer.h @@ -13,6 +13,8 @@ namespace Layout { struct FramebufferLayout; } +struct PresentFilters; + namespace Service::android { enum class PixelFormat : u32; }; @@ -44,14 +46,15 @@ struct ScreenRectVertex; class Layer { public: - explicit Layer(RasterizerOpenGL& rasterizer, Tegra::MaxwellDeviceMemoryManager& device_memory); + explicit Layer(RasterizerOpenGL& rasterizer, Tegra::MaxwellDeviceMemoryManager& device_memory, + const PresentFilters& filters); ~Layer(); GLuint ConfigureDraw(std::array<GLfloat, 3 * 2>& out_matrix, std::array<ScreenRectVertex, 4>& out_vertices, ProgramManager& program_manager, const Tegra::FramebufferConfig& framebuffer, - const Layout::FramebufferLayout& layout); + const Layout::FramebufferLayout& layout, bool invert_y); private: /// Loads framebuffer from emulated memory into the active OpenGL texture. @@ -65,6 +68,7 @@ private: private: RasterizerOpenGL& rasterizer; Tegra::MaxwellDeviceMemoryManager& device_memory; + const PresentFilters& filters; /// OpenGL framebuffer data std::vector<u8> gl_framebuffer_data; diff --git a/src/video_core/renderer_opengl/present/window_adapt_pass.cpp b/src/video_core/renderer_opengl/present/window_adapt_pass.cpp index 4d681606b..d8b6a11cb 100644 --- a/src/video_core/renderer_opengl/present/window_adapt_pass.cpp +++ b/src/video_core/renderer_opengl/present/window_adapt_pass.cpp @@ -37,7 +37,7 @@ WindowAdaptPass::~WindowAdaptPass() = default; void WindowAdaptPass::DrawToFramebuffer(ProgramManager& program_manager, std::list<Layer>& layers, std::span<const Tegra::FramebufferConfig> framebuffers, - const Layout::FramebufferLayout& layout) { + const Layout::FramebufferLayout& layout, bool invert_y) { GLint old_read_fb; GLint old_draw_fb; glGetIntegerv(GL_READ_FRAMEBUFFER_BINDING, &old_read_fb); @@ -51,7 +51,7 @@ void WindowAdaptPass::DrawToFramebuffer(ProgramManager& program_manager, std::li auto layer_it = layers.begin(); for (size_t i = 0; i < layer_count; i++) { textures[i] = layer_it->ConfigureDraw(matrices[i], vertices[i], program_manager, - framebuffers[i], layout); + framebuffers[i], layout, invert_y); layer_it++; } @@ -92,6 +92,21 @@ void WindowAdaptPass::DrawToFramebuffer(ProgramManager& program_manager, std::li glClear(GL_COLOR_BUFFER_BIT); for (size_t i = 0; i < layer_count; i++) { + switch (framebuffers[i].blending) { + case Tegra::BlendMode::Opaque: + default: + glDisablei(GL_BLEND, 0); + break; + case Tegra::BlendMode::Premultiplied: + glEnablei(GL_BLEND, 0); + glBlendFuncSeparatei(0, GL_ONE, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ZERO); + break; + case Tegra::BlendMode::Coverage: + glEnablei(GL_BLEND, 0); + glBlendFuncSeparatei(0, GL_SRC_ALPHA, GL_ONE_MINUS_SRC_ALPHA, GL_ONE, GL_ZERO); + break; + } + glBindTextureUnit(0, textures[i]); glProgramUniformMatrix3x2fv(vert.handle, ModelViewMatrixLocation, 1, GL_FALSE, matrices[i].data()); diff --git a/src/video_core/renderer_opengl/present/window_adapt_pass.h b/src/video_core/renderer_opengl/present/window_adapt_pass.h index 00975a9c6..0a8bcef2f 100644 --- a/src/video_core/renderer_opengl/present/window_adapt_pass.h +++ b/src/video_core/renderer_opengl/present/window_adapt_pass.h @@ -31,7 +31,7 @@ public: void DrawToFramebuffer(ProgramManager& program_manager, std::list<Layer>& layers, std::span<const Tegra::FramebufferConfig> framebuffers, - const Layout::FramebufferLayout& layout); + const Layout::FramebufferLayout& layout, bool invert_y); private: const Device& device; diff --git a/src/video_core/renderer_opengl/renderer_opengl.cpp b/src/video_core/renderer_opengl/renderer_opengl.cpp index e33a32592..5fb54635d 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.cpp +++ b/src/video_core/renderer_opengl/renderer_opengl.cpp @@ -16,6 +16,8 @@ #include "core/core_timing.h" #include "core/frontend/emu_window.h" #include "core/telemetry_session.h" +#include "video_core/capture.h" +#include "video_core/present.h" #include "video_core/renderer_opengl/gl_blit_screen.h" #include "video_core/renderer_opengl/gl_rasterizer.h" #include "video_core/renderer_opengl/gl_shader_manager.h" @@ -120,7 +122,15 @@ RendererOpenGL::RendererOpenGL(Core::TelemetrySession& telemetry_session_, glEnableClientState(GL_ELEMENT_ARRAY_UNIFIED_NV); } blit_screen = std::make_unique<BlitScreen>(rasterizer, device_memory, state_tracker, - program_manager, device); + program_manager, device, PresentFiltersForDisplay); + blit_applet = + std::make_unique<BlitScreen>(rasterizer, device_memory, state_tracker, program_manager, + device, PresentFiltersForAppletCapture); + capture_framebuffer.Create(); + capture_renderbuffer.Create(); + glBindRenderbuffer(GL_RENDERBUFFER, capture_renderbuffer.handle); + glRenderbufferStorage(GL_RENDERBUFFER, GL_SRGB8, VideoCore::Capture::LinearWidth, + VideoCore::Capture::LinearHeight); } RendererOpenGL::~RendererOpenGL() = default; @@ -130,10 +140,11 @@ void RendererOpenGL::Composite(std::span<const Tegra::FramebufferConfig> framebu return; } + RenderAppletCaptureLayer(framebuffers); RenderScreenshot(framebuffers); state_tracker.BindFramebuffer(0); - blit_screen->DrawScreen(framebuffers, emu_window.GetFramebufferLayout()); + blit_screen->DrawScreen(framebuffers, emu_window.GetFramebufferLayout(), false); ++m_current_frame; @@ -159,11 +170,8 @@ void RendererOpenGL::AddTelemetryFields() { telemetry_session.AddField(user_system, "GPU_OpenGL_Version", std::string(gl_version)); } -void RendererOpenGL::RenderScreenshot(std::span<const Tegra::FramebufferConfig> framebuffers) { - if (!renderer_settings.screenshot_requested) { - return; - } - +void RendererOpenGL::RenderToBuffer(std::span<const Tegra::FramebufferConfig> framebuffers, + const Layout::FramebufferLayout& layout, void* dst) { GLint old_read_fb; GLint old_draw_fb; glGetIntegerv(GL_READ_FRAMEBUFFER_BINDING, &old_read_fb); @@ -173,29 +181,86 @@ void RendererOpenGL::RenderScreenshot(std::span<const Tegra::FramebufferConfig> screenshot_framebuffer.Create(); glBindFramebuffer(GL_FRAMEBUFFER, screenshot_framebuffer.handle); - const Layout::FramebufferLayout layout{renderer_settings.screenshot_framebuffer_layout}; - GLuint renderbuffer; glGenRenderbuffers(1, &renderbuffer); glBindRenderbuffer(GL_RENDERBUFFER, renderbuffer); glRenderbufferStorage(GL_RENDERBUFFER, GL_SRGB8, layout.width, layout.height); glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, renderbuffer); - blit_screen->DrawScreen(framebuffers, layout); + blit_screen->DrawScreen(framebuffers, layout, false); glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); glPixelStorei(GL_PACK_ROW_LENGTH, 0); - glReadPixels(0, 0, layout.width, layout.height, GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, - renderer_settings.screenshot_bits); + glReadPixels(0, 0, layout.width, layout.height, GL_BGRA, GL_UNSIGNED_INT_8_8_8_8_REV, dst); screenshot_framebuffer.Release(); glDeleteRenderbuffers(1, &renderbuffer); glBindFramebuffer(GL_READ_FRAMEBUFFER, old_read_fb); glBindFramebuffer(GL_DRAW_FRAMEBUFFER, old_draw_fb); +} + +void RendererOpenGL::RenderScreenshot(std::span<const Tegra::FramebufferConfig> framebuffers) { + if (!renderer_settings.screenshot_requested) { + return; + } + + RenderToBuffer(framebuffers, renderer_settings.screenshot_framebuffer_layout, + renderer_settings.screenshot_bits); renderer_settings.screenshot_complete_callback(true); renderer_settings.screenshot_requested = false; } +void RendererOpenGL::RenderAppletCaptureLayer( + std::span<const Tegra::FramebufferConfig> framebuffers) { + GLint old_read_fb; + GLint old_draw_fb; + glGetIntegerv(GL_READ_FRAMEBUFFER_BINDING, &old_read_fb); + glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &old_draw_fb); + + glBindFramebuffer(GL_FRAMEBUFFER, capture_framebuffer.handle); + glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, + capture_renderbuffer.handle); + + blit_applet->DrawScreen(framebuffers, VideoCore::Capture::Layout, true); + + glBindFramebuffer(GL_READ_FRAMEBUFFER, old_read_fb); + glBindFramebuffer(GL_DRAW_FRAMEBUFFER, old_draw_fb); +} + +std::vector<u8> RendererOpenGL::GetAppletCaptureBuffer() { + using namespace VideoCore::Capture; + + std::vector<u8> linear(TiledSize); + std::vector<u8> out(TiledSize); + + GLint old_read_fb; + GLint old_draw_fb; + GLint old_pixel_pack_buffer; + GLint old_pack_row_length; + glGetIntegerv(GL_READ_FRAMEBUFFER_BINDING, &old_read_fb); + glGetIntegerv(GL_DRAW_FRAMEBUFFER_BINDING, &old_draw_fb); + glGetIntegerv(GL_PIXEL_PACK_BUFFER_BINDING, &old_pixel_pack_buffer); + glGetIntegerv(GL_PACK_ROW_LENGTH, &old_pack_row_length); + + glBindFramebuffer(GL_FRAMEBUFFER, capture_framebuffer.handle); + glFramebufferRenderbuffer(GL_FRAMEBUFFER, GL_COLOR_ATTACHMENT0, GL_RENDERBUFFER, + capture_renderbuffer.handle); + glBindBuffer(GL_PIXEL_PACK_BUFFER, 0); + glPixelStorei(GL_PACK_ROW_LENGTH, 0); + glReadPixels(0, 0, LinearWidth, LinearHeight, GL_RGBA, GL_UNSIGNED_INT_8_8_8_8_REV, + linear.data()); + + glBindFramebuffer(GL_READ_FRAMEBUFFER, old_read_fb); + glBindFramebuffer(GL_DRAW_FRAMEBUFFER, old_draw_fb); + glBindBuffer(GL_PIXEL_PACK_BUFFER, old_pixel_pack_buffer); + glPixelStorei(GL_PACK_ROW_LENGTH, old_pack_row_length); + + Tegra::Texture::SwizzleTexture(out, linear, BytesPerPixel, LinearWidth, LinearHeight, + LinearDepth, BlockHeight, BlockDepth); + + return out; +} + } // namespace OpenGL diff --git a/src/video_core/renderer_opengl/renderer_opengl.h b/src/video_core/renderer_opengl/renderer_opengl.h index c4625c96e..60d6a1477 100644 --- a/src/video_core/renderer_opengl/renderer_opengl.h +++ b/src/video_core/renderer_opengl/renderer_opengl.h @@ -42,6 +42,8 @@ public: void Composite(std::span<const Tegra::FramebufferConfig> framebuffers) override; + std::vector<u8> GetAppletCaptureBuffer() override; + VideoCore::RasterizerInterface* ReadRasterizer() override { return &rasterizer; } @@ -52,7 +54,11 @@ public: private: void AddTelemetryFields(); + + void RenderToBuffer(std::span<const Tegra::FramebufferConfig> framebuffers, + const Layout::FramebufferLayout& layout, void* dst); void RenderScreenshot(std::span<const Tegra::FramebufferConfig> framebuffers); + void RenderAppletCaptureLayer(std::span<const Tegra::FramebufferConfig> framebuffers); Core::TelemetrySession& telemetry_session; Core::Frontend::EmuWindow& emu_window; @@ -64,8 +70,11 @@ private: ProgramManager program_manager; RasterizerOpenGL rasterizer; OGLFramebuffer screenshot_framebuffer; + OGLFramebuffer capture_framebuffer; + OGLRenderbuffer capture_renderbuffer; std::unique_ptr<BlitScreen> blit_screen; + std::unique_ptr<BlitScreen> blit_applet; }; } // namespace OpenGL diff --git a/src/video_core/renderer_vulkan/present/layer.cpp b/src/video_core/renderer_vulkan/present/layer.cpp index cfc04be44..4e41afe5b 100644 --- a/src/video_core/renderer_vulkan/present/layer.cpp +++ b/src/video_core/renderer_vulkan/present/layer.cpp @@ -1,6 +1,7 @@ // SPDX-FileCopyrightText: Copyright 2024 yuzu Emulator Project // SPDX-License-Identifier: GPL-2.0-or-later +#include "video_core/present.h" #include "video_core/renderer_vulkan/vk_rasterizer.h" #include "common/settings.h" @@ -48,12 +49,12 @@ VkFormat GetFormat(const Tegra::FramebufferConfig& framebuffer) { Layer::Layer(const Device& device_, MemoryAllocator& memory_allocator_, Scheduler& scheduler_, Tegra::MaxwellDeviceMemoryManager& device_memory_, size_t image_count_, - VkExtent2D output_size, VkDescriptorSetLayout layout) + VkExtent2D output_size, VkDescriptorSetLayout layout, const PresentFilters& filters_) : device(device_), memory_allocator(memory_allocator_), scheduler(scheduler_), - device_memory(device_memory_), image_count(image_count_) { + device_memory(device_memory_), filters(filters_), image_count(image_count_) { CreateDescriptorPool(); CreateDescriptorSets(layout); - if (Settings::values.scaling_filter.GetValue() == Settings::ScalingFilter::Fsr) { + if (filters.get_scaling_filter() == Settings::ScalingFilter::Fsr) { CreateFSR(output_size); } } @@ -81,7 +82,9 @@ void Layer::ConfigureDraw(PresentPushConstants* out_push_constants, // Finish any pending renderpass scheduler.RequestOutsideRenderPassOperationContext(); scheduler.Wait(resource_ticks[image_index]); - SCOPE_EXIT({ resource_ticks[image_index] = scheduler.CurrentTick(); }); + SCOPE_EXIT { + resource_ticks[image_index] = scheduler.CurrentTick(); + }; if (!use_accelerated) { UpdateRawImage(framebuffer, image_index); @@ -171,11 +174,11 @@ void Layer::RefreshResources(const Tegra::FramebufferConfig& framebuffer) { } void Layer::SetAntiAliasPass() { - if (anti_alias && anti_alias_setting == Settings::values.anti_aliasing.GetValue()) { + if (anti_alias && anti_alias_setting == filters.get_anti_aliasing()) { return; } - anti_alias_setting = Settings::values.anti_aliasing.GetValue(); + anti_alias_setting = filters.get_anti_aliasing(); const VkExtent2D render_area{ .width = Settings::values.resolution_info.ScaleUp(raw_width), @@ -270,9 +273,11 @@ void Layer::UpdateRawImage(const Tegra::FramebufferConfig& framebuffer, size_t i const u64 linear_size{GetSizeInBytes(framebuffer)}; const u64 tiled_size{Tegra::Texture::CalculateSize( true, bytes_per_pixel, framebuffer.stride, framebuffer.height, 1, block_height_log2, 0)}; - Tegra::Texture::UnswizzleTexture( - mapped_span.subspan(image_offset, linear_size), std::span(host_ptr, tiled_size), - bytes_per_pixel, framebuffer.width, framebuffer.height, 1, block_height_log2, 0); + if (host_ptr) { + Tegra::Texture::UnswizzleTexture( + mapped_span.subspan(image_offset, linear_size), std::span(host_ptr, tiled_size), + bytes_per_pixel, framebuffer.width, framebuffer.height, 1, block_height_log2, 0); + } const VkBufferImageCopy copy{ .bufferOffset = image_offset, diff --git a/src/video_core/renderer_vulkan/present/layer.h b/src/video_core/renderer_vulkan/present/layer.h index 88d43fc5f..f5effdcd7 100644 --- a/src/video_core/renderer_vulkan/present/layer.h +++ b/src/video_core/renderer_vulkan/present/layer.h @@ -11,6 +11,8 @@ namespace Layout { struct FramebufferLayout; } +struct PresentFilters; + namespace Tegra { struct FramebufferConfig; } @@ -37,7 +39,8 @@ class Layer final { public: explicit Layer(const Device& device, MemoryAllocator& memory_allocator, Scheduler& scheduler, Tegra::MaxwellDeviceMemoryManager& device_memory, size_t image_count, - VkExtent2D output_size, VkDescriptorSetLayout layout); + VkExtent2D output_size, VkDescriptorSetLayout layout, + const PresentFilters& filters); ~Layer(); void ConfigureDraw(PresentPushConstants* out_push_constants, @@ -71,6 +74,7 @@ private: MemoryAllocator& memory_allocator; Scheduler& scheduler; Tegra::MaxwellDeviceMemoryManager& device_memory; + const PresentFilters& filters; const size_t image_count{}; vk::DescriptorPool descriptor_pool{}; vk::DescriptorSets descriptor_sets{}; diff --git a/src/video_core/renderer_vulkan/present/util.cpp b/src/video_core/renderer_vulkan/present/util.cpp index 6ee16595d..7f27c7c1b 100644 --- a/src/video_core/renderer_vulkan/present/util.cpp +++ b/src/video_core/renderer_vulkan/present/util.cpp @@ -362,10 +362,10 @@ vk::PipelineLayout CreateWrappedPipelineLayout(const Device& device, }); } -vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, - vk::PipelineLayout& layout, - std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders, - bool enable_blending) { +static vk::Pipeline CreateWrappedPipelineImpl( + const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders, + VkPipelineColorBlendAttachmentState blending) { const std::array<VkPipelineShaderStageCreateInfo, 2> shader_stages{{ { .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO, @@ -443,30 +443,6 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp .alphaToOneEnable = VK_FALSE, }; - constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_disabled{ - .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, - }; - - constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_enabled{ - .blendEnable = VK_TRUE, - .srcColorBlendFactor = VK_BLEND_FACTOR_SRC_ALPHA, - .dstColorBlendFactor = VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA, - .colorBlendOp = VK_BLEND_OP_ADD, - .srcAlphaBlendFactor = VK_BLEND_FACTOR_ONE, - .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, @@ -474,8 +450,7 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp .logicOpEnable = VK_FALSE, .logicOp = VK_LOGIC_OP_COPY, .attachmentCount = 1, - .pAttachments = - enable_blending ? &color_blend_attachment_enabled : &color_blend_attachment_disabled, + .pAttachments = &blending, .blendConstants = {0.0f, 0.0f, 0.0f, 0.0f}, }; @@ -515,6 +490,63 @@ vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderp }); } +vk::Pipeline CreateWrappedPipeline(const Device& device, vk::RenderPass& renderpass, + vk::PipelineLayout& layout, + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders) { + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_disabled{ + .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, + }; + + return CreateWrappedPipelineImpl(device, renderpass, layout, shaders, + color_blend_attachment_disabled); +} + +vk::Pipeline CreateWrappedPremultipliedBlendingPipeline( + const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders) { + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_premultiplied{ + .blendEnable = VK_TRUE, + .srcColorBlendFactor = VK_BLEND_FACTOR_ONE, + .dstColorBlendFactor = VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA, + .colorBlendOp = VK_BLEND_OP_ADD, + .srcAlphaBlendFactor = VK_BLEND_FACTOR_ONE, + .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, + }; + + return CreateWrappedPipelineImpl(device, renderpass, layout, shaders, + color_blend_attachment_premultiplied); +} + +vk::Pipeline CreateWrappedCoverageBlendingPipeline( + const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders) { + constexpr VkPipelineColorBlendAttachmentState color_blend_attachment_coverage{ + .blendEnable = VK_TRUE, + .srcColorBlendFactor = VK_BLEND_FACTOR_SRC_ALPHA, + .dstColorBlendFactor = VK_BLEND_FACTOR_ONE_MINUS_SRC_ALPHA, + .colorBlendOp = VK_BLEND_OP_ADD, + .srcAlphaBlendFactor = VK_BLEND_FACTOR_ONE, + .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, + }; + + return CreateWrappedPipelineImpl(device, renderpass, layout, shaders, + color_blend_attachment_coverage); +} + VkWriteDescriptorSet CreateWriteDescriptorSet(std::vector<VkDescriptorImageInfo>& images, VkSampler sampler, VkImageView view, VkDescriptorSet set, u32 binding) { diff --git a/src/video_core/renderer_vulkan/present/util.h b/src/video_core/renderer_vulkan/present/util.h index 1104aaa15..5b22f0fa8 100644 --- a/src/video_core/renderer_vulkan/present/util.h +++ b/src/video_core/renderer_vulkan/present/util.h @@ -42,8 +42,13 @@ 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, - bool enable_blending = false); + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders); +vk::Pipeline CreateWrappedPremultipliedBlendingPipeline( + const Device& device, vk::RenderPass& renderpass, vk::PipelineLayout& layout, + std::tuple<vk::ShaderModule&, vk::ShaderModule&> shaders); +vk::Pipeline CreateWrappedCoverageBlendingPipeline( + 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); diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp index c5db0230d..22ffacf11 100644 --- a/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.cpp @@ -22,7 +22,7 @@ WindowAdaptPass::WindowAdaptPass(const Device& device_, VkFormat frame_format, CreatePipelineLayout(); CreateVertexShader(); CreateRenderPass(frame_format); - CreatePipeline(); + CreatePipelines(); } WindowAdaptPass::~WindowAdaptPass() = default; @@ -34,7 +34,6 @@ void WindowAdaptPass::Draw(RasterizerVulkan& rasterizer, Scheduler& scheduler, s const VkFramebuffer host_framebuffer{*dst->framebuffer}; const VkRenderPass renderpass{*render_pass}; - const VkPipeline graphics_pipeline{*pipeline}; const VkPipelineLayout graphics_pipeline_layout{*pipeline_layout}; const VkExtent2D render_area{ .width = dst->width, @@ -44,9 +43,23 @@ void WindowAdaptPass::Draw(RasterizerVulkan& rasterizer, Scheduler& scheduler, s const size_t layer_count = configs.size(); std::vector<PresentPushConstants> push_constants(layer_count); std::vector<VkDescriptorSet> descriptor_sets(layer_count); + std::vector<VkPipeline> graphics_pipelines(layer_count); auto layer_it = layers.begin(); for (size_t i = 0; i < layer_count; i++) { + switch (configs[i].blending) { + case Tegra::BlendMode::Opaque: + default: + graphics_pipelines[i] = *opaque_pipeline; + break; + case Tegra::BlendMode::Premultiplied: + graphics_pipelines[i] = *premultiplied_pipeline; + break; + case Tegra::BlendMode::Coverage: + graphics_pipelines[i] = *coverage_pipeline; + break; + } + layer_it->ConfigureDraw(&push_constants[i], &descriptor_sets[i], rasterizer, *sampler, image_index, configs[i], layout); layer_it++; @@ -77,8 +90,8 @@ void WindowAdaptPass::Draw(RasterizerVulkan& rasterizer, Scheduler& scheduler, s BeginRenderPass(cmdbuf, renderpass, host_framebuffer, render_area); cmdbuf.ClearAttachments({clear_attachment}, {clear_rect}); - cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline); for (size_t i = 0; i < layer_count; i++) { + cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipelines[i]); cmdbuf.PushConstants(graphics_pipeline_layout, VK_SHADER_STAGE_VERTEX_BIT, push_constants[i]); cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_GRAPHICS, graphics_pipeline_layout, 0, @@ -129,9 +142,13 @@ void WindowAdaptPass::CreateRenderPass(VkFormat frame_format) { render_pass = CreateWrappedRenderPass(device, frame_format, VK_IMAGE_LAYOUT_UNDEFINED); } -void WindowAdaptPass::CreatePipeline() { - pipeline = CreateWrappedPipeline(device, render_pass, pipeline_layout, - std::tie(vertex_shader, fragment_shader), false); +void WindowAdaptPass::CreatePipelines() { + opaque_pipeline = CreateWrappedPipeline(device, render_pass, pipeline_layout, + std::tie(vertex_shader, fragment_shader)); + premultiplied_pipeline = CreateWrappedPremultipliedBlendingPipeline( + device, render_pass, pipeline_layout, std::tie(vertex_shader, fragment_shader)); + coverage_pipeline = CreateWrappedCoverageBlendingPipeline( + device, render_pass, pipeline_layout, std::tie(vertex_shader, fragment_shader)); } } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/present/window_adapt_pass.h b/src/video_core/renderer_vulkan/present/window_adapt_pass.h index 0e2edfc31..cf667a4fc 100644 --- a/src/video_core/renderer_vulkan/present/window_adapt_pass.h +++ b/src/video_core/renderer_vulkan/present/window_adapt_pass.h @@ -42,7 +42,7 @@ private: void CreatePipelineLayout(); void CreateVertexShader(); void CreateRenderPass(VkFormat frame_format); - void CreatePipeline(); + void CreatePipelines(); private: const Device& device; @@ -52,7 +52,9 @@ private: vk::ShaderModule vertex_shader; vk::ShaderModule fragment_shader; vk::RenderPass render_pass; - vk::Pipeline pipeline; + vk::Pipeline opaque_pipeline; + vk::Pipeline premultiplied_pipeline; + vk::Pipeline coverage_pipeline; }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.cpp b/src/video_core/renderer_vulkan/renderer_vulkan.cpp index 48a105327..c553f5b3d 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.cpp +++ b/src/video_core/renderer_vulkan/renderer_vulkan.cpp @@ -19,7 +19,9 @@ #include "core/core_timing.h" #include "core/frontend/graphics_context.h" #include "core/telemetry_session.h" +#include "video_core/capture.h" #include "video_core/gpu.h" +#include "video_core/present.h" #include "video_core/renderer_vulkan/present/util.h" #include "video_core/renderer_vulkan/renderer_vulkan.h" #include "video_core/renderer_vulkan/vk_blit_screen.h" @@ -38,6 +40,20 @@ namespace Vulkan { namespace { + +constexpr VkExtent2D CaptureImageSize{ + .width = VideoCore::Capture::LinearWidth, + .height = VideoCore::Capture::LinearHeight, +}; + +constexpr VkExtent3D CaptureImageExtent{ + .width = VideoCore::Capture::LinearWidth, + .height = VideoCore::Capture::LinearHeight, + .depth = VideoCore::Capture::LinearDepth, +}; + +constexpr VkFormat CaptureFormat = VK_FORMAT_A8B8G8R8_UNORM_PACK32; + std::string GetReadableVersion(u32 version) { return fmt::format("{}.{}.{}", VK_VERSION_MAJOR(version), VK_VERSION_MINOR(version), VK_VERSION_PATCH(version)); @@ -99,10 +115,15 @@ RendererVulkan::RendererVulkan(Core::TelemetrySession& telemetry_session_, render_window.GetFramebufferLayout().height), present_manager(instance, render_window, device, memory_allocator, scheduler, swapchain, surface), - blit_swapchain(device_memory, device, memory_allocator, present_manager, scheduler), - blit_screenshot(device_memory, device, memory_allocator, present_manager, scheduler), + blit_swapchain(device_memory, device, memory_allocator, present_manager, scheduler, + PresentFiltersForDisplay), + blit_capture(device_memory, device, memory_allocator, present_manager, scheduler, + PresentFiltersForDisplay), + blit_applet(device_memory, device, memory_allocator, present_manager, scheduler, + PresentFiltersForAppletCapture), rasterizer(render_window, gpu, device_memory, device, memory_allocator, state_tracker, - scheduler) { + scheduler), + applet_frame() { if (Settings::values.renderer_force_max_clock.GetValue() && device.ShouldBoostClocks()) { turbo_mode.emplace(instance, dld); scheduler.RegisterOnSubmit([this] { turbo_mode->QueueSubmitted(); }); @@ -123,7 +144,11 @@ void RendererVulkan::Composite(std::span<const Tegra::FramebufferConfig> framebu return; } - SCOPE_EXIT({ render_window.OnFrameDisplayed(); }); + SCOPE_EXIT { + render_window.OnFrameDisplayed(); + }; + + RenderAppletCaptureLayer(framebuffers); if (!render_window.IsShown()) { return; @@ -167,30 +192,20 @@ void RendererVulkan::Report() const { telemetry_session.AddField(field, "GPU_Vulkan_Extensions", extensions); } -void Vulkan::RendererVulkan::RenderScreenshot( - std::span<const Tegra::FramebufferConfig> framebuffers) { - if (!renderer_settings.screenshot_requested) { - return; - } - - constexpr VkFormat ScreenshotFormat{VK_FORMAT_B8G8R8A8_UNORM}; - const Layout::FramebufferLayout layout{renderer_settings.screenshot_framebuffer_layout}; - +vk::Buffer RendererVulkan::RenderToBuffer(std::span<const Tegra::FramebufferConfig> framebuffers, + const Layout::FramebufferLayout& layout, VkFormat format, + VkDeviceSize buffer_size) { auto frame = [&]() { Frame f{}; - f.image = CreateWrappedImage(memory_allocator, VkExtent2D{layout.width, layout.height}, - ScreenshotFormat); - f.image_view = CreateWrappedImageView(device, f.image, ScreenshotFormat); - f.framebuffer = blit_screenshot.CreateFramebuffer(layout, *f.image_view, ScreenshotFormat); + f.image = + CreateWrappedImage(memory_allocator, VkExtent2D{layout.width, layout.height}, format); + f.image_view = CreateWrappedImageView(device, f.image, format); + f.framebuffer = blit_capture.CreateFramebuffer(layout, *f.image_view, format); return f; }(); - blit_screenshot.DrawToFrame(rasterizer, &frame, framebuffers, layout, 1, - VK_FORMAT_B8G8R8A8_UNORM); - - const auto dst_buffer = CreateWrappedBuffer( - memory_allocator, static_cast<VkDeviceSize>(layout.width * layout.height * 4), - MemoryUsage::Download); + auto dst_buffer = CreateWrappedBuffer(memory_allocator, buffer_size, MemoryUsage::Download); + blit_capture.DrawToFrame(rasterizer, &frame, framebuffers, layout, 1, format); scheduler.RequestOutsideRenderPassOperationContext(); scheduler.Record([&](vk::CommandBuffer cmdbuf) { @@ -198,15 +213,68 @@ void Vulkan::RendererVulkan::RenderScreenshot( VkExtent3D{layout.width, layout.height, 1}); }); - // Ensure the copy is fully completed before saving the screenshot + // Ensure the copy is fully completed before saving the capture scheduler.Finish(); - // Copy backing image data to the QImage screenshot buffer + // Copy backing image data to the capture buffer dst_buffer.Invalidate(); + return dst_buffer; +} + +void RendererVulkan::RenderScreenshot(std::span<const Tegra::FramebufferConfig> framebuffers) { + if (!renderer_settings.screenshot_requested) { + return; + } + + const auto& layout{renderer_settings.screenshot_framebuffer_layout}; + const auto dst_buffer = RenderToBuffer(framebuffers, layout, VK_FORMAT_B8G8R8A8_UNORM, + layout.width * layout.height * 4); + std::memcpy(renderer_settings.screenshot_bits, dst_buffer.Mapped().data(), dst_buffer.Mapped().size()); renderer_settings.screenshot_complete_callback(false); renderer_settings.screenshot_requested = false; } +std::vector<u8> RendererVulkan::GetAppletCaptureBuffer() { + using namespace VideoCore::Capture; + + std::vector<u8> out(VideoCore::Capture::TiledSize); + + if (!applet_frame.image) { + return out; + } + + const auto dst_buffer = + CreateWrappedBuffer(memory_allocator, VideoCore::Capture::TiledSize, MemoryUsage::Download); + + scheduler.RequestOutsideRenderPassOperationContext(); + scheduler.Record([&](vk::CommandBuffer cmdbuf) { + DownloadColorImage(cmdbuf, *applet_frame.image, *dst_buffer, CaptureImageExtent); + }); + + // Ensure the copy is fully completed before writing the capture + scheduler.Finish(); + + // Swizzle image data to the capture buffer + dst_buffer.Invalidate(); + Tegra::Texture::SwizzleTexture(out, dst_buffer.Mapped(), BytesPerPixel, LinearWidth, + LinearHeight, LinearDepth, BlockHeight, BlockDepth); + + return out; +} + +void RendererVulkan::RenderAppletCaptureLayer( + std::span<const Tegra::FramebufferConfig> framebuffers) { + if (!applet_frame.image) { + applet_frame.image = CreateWrappedImage(memory_allocator, CaptureImageSize, CaptureFormat); + applet_frame.image_view = CreateWrappedImageView(device, applet_frame.image, CaptureFormat); + applet_frame.framebuffer = blit_applet.CreateFramebuffer( + VideoCore::Capture::Layout, *applet_frame.image_view, CaptureFormat); + } + + blit_applet.DrawToFrame(rasterizer, &applet_frame, framebuffers, VideoCore::Capture::Layout, 1, + CaptureFormat); +} + } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/renderer_vulkan.h b/src/video_core/renderer_vulkan/renderer_vulkan.h index c6d8a0f21..fb9d83412 100644 --- a/src/video_core/renderer_vulkan/renderer_vulkan.h +++ b/src/video_core/renderer_vulkan/renderer_vulkan.h @@ -48,6 +48,8 @@ public: void Composite(std::span<const Tegra::FramebufferConfig> framebuffers) override; + std::vector<u8> GetAppletCaptureBuffer() override; + VideoCore::RasterizerInterface* ReadRasterizer() override { return &rasterizer; } @@ -59,7 +61,11 @@ public: private: void Report() const; + vk::Buffer RenderToBuffer(std::span<const Tegra::FramebufferConfig> framebuffers, + const Layout::FramebufferLayout& layout, VkFormat format, + VkDeviceSize buffer_size); void RenderScreenshot(std::span<const Tegra::FramebufferConfig> framebuffers); + void RenderAppletCaptureLayer(std::span<const Tegra::FramebufferConfig> framebuffers); Core::TelemetrySession& telemetry_session; Tegra::MaxwellDeviceMemoryManager& device_memory; @@ -79,9 +85,12 @@ private: Swapchain swapchain; PresentManager present_manager; BlitScreen blit_swapchain; - BlitScreen blit_screenshot; + BlitScreen blit_capture; + BlitScreen blit_applet; RasterizerVulkan rasterizer; std::optional<TurboMode> turbo_mode; + + Frame applet_frame; }; } // namespace Vulkan diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.cpp b/src/video_core/renderer_vulkan/vk_blit_screen.cpp index 2275fcc46..b7797f833 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.cpp +++ b/src/video_core/renderer_vulkan/vk_blit_screen.cpp @@ -2,6 +2,7 @@ // SPDX-License-Identifier: GPL-2.0-or-later #include "video_core/framebuffer_config.h" +#include "video_core/present.h" #include "video_core/renderer_vulkan/present/filters.h" #include "video_core/renderer_vulkan/present/layer.h" #include "video_core/renderer_vulkan/vk_blit_screen.h" @@ -12,9 +13,9 @@ namespace Vulkan { BlitScreen::BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory_, const Device& device_, MemoryAllocator& memory_allocator_, PresentManager& present_manager_, - Scheduler& scheduler_) + Scheduler& scheduler_, const PresentFilters& filters_) : device_memory{device_memory_}, device{device_}, memory_allocator{memory_allocator_}, - present_manager{present_manager_}, scheduler{scheduler_}, image_count{1}, + present_manager{present_manager_}, scheduler{scheduler_}, filters{filters_}, image_count{1}, swapchain_view_format{VK_FORMAT_B8G8R8A8_UNORM} {} BlitScreen::~BlitScreen() = default; @@ -27,7 +28,7 @@ void BlitScreen::WaitIdle() { void BlitScreen::SetWindowAdaptPass() { layers.clear(); - scaling_filter = Settings::values.scaling_filter.GetValue(); + scaling_filter = filters.get_scaling_filter(); switch (scaling_filter) { case Settings::ScalingFilter::NearestNeighbor: @@ -59,7 +60,7 @@ void BlitScreen::DrawToFrame(RasterizerVulkan& rasterizer, Frame* frame, bool presentation_recreate_required = false; // Recreate dynamic resources if the adapting filter changed - if (!window_adapt || scaling_filter != Settings::values.scaling_filter.GetValue()) { + if (!window_adapt || scaling_filter != filters.get_scaling_filter()) { resource_update_required = true; } @@ -102,7 +103,7 @@ void BlitScreen::DrawToFrame(RasterizerVulkan& rasterizer, Frame* frame, while (layers.size() < framebuffers.size()) { layers.emplace_back(device, memory_allocator, scheduler, device_memory, image_count, - window_size, window_adapt->GetDescriptorSetLayout()); + window_size, window_adapt->GetDescriptorSetLayout(), filters); } // Perform the draw @@ -119,8 +120,7 @@ vk::Framebuffer BlitScreen::CreateFramebuffer(const Layout::FramebufferLayout& l VkFormat current_view_format) { const bool format_updated = std::exchange(swapchain_view_format, current_view_format) != current_view_format; - if (!window_adapt || scaling_filter != Settings::values.scaling_filter.GetValue() || - format_updated) { + if (!window_adapt || scaling_filter != filters.get_scaling_filter() || format_updated) { WaitIdle(); SetWindowAdaptPass(); } diff --git a/src/video_core/renderer_vulkan/vk_blit_screen.h b/src/video_core/renderer_vulkan/vk_blit_screen.h index cbdf2d5d0..531c57fc5 100644 --- a/src/video_core/renderer_vulkan/vk_blit_screen.h +++ b/src/video_core/renderer_vulkan/vk_blit_screen.h @@ -16,6 +16,8 @@ namespace Core { class System; } +struct PresentFilters; + namespace Tegra { struct FramebufferConfig; } @@ -47,7 +49,7 @@ class BlitScreen { public: explicit BlitScreen(Tegra::MaxwellDeviceMemoryManager& device_memory, const Device& device, MemoryAllocator& memory_allocator, PresentManager& present_manager, - Scheduler& scheduler); + Scheduler& scheduler, const PresentFilters& filters); ~BlitScreen(); void DrawToFrame(RasterizerVulkan& rasterizer, Frame* frame, @@ -70,6 +72,7 @@ private: MemoryAllocator& memory_allocator; PresentManager& present_manager; Scheduler& scheduler; + const PresentFilters& filters; std::size_t image_count{}; std::size_t image_index{}; VkFormat swapchain_view_format{}; diff --git a/src/video_core/renderer_vulkan/vk_rasterizer.cpp b/src/video_core/renderer_vulkan/vk_rasterizer.cpp index aa0a027bb..74f9f099e 100644 --- a/src/video_core/renderer_vulkan/vk_rasterizer.cpp +++ b/src/video_core/renderer_vulkan/vk_rasterizer.cpp @@ -196,7 +196,9 @@ template <typename Func> void RasterizerVulkan::PrepareDraw(bool is_indexed, Func&& draw_func) { MICROPROFILE_SCOPE(Vulkan_Drawing); - SCOPE_EXIT({ gpu.TickWork(); }); + SCOPE_EXIT { + gpu.TickWork(); + }; FlushWork(); gpu_memory->FlushCaching(); @@ -288,7 +290,9 @@ void RasterizerVulkan::DrawIndirect() { void RasterizerVulkan::DrawTexture() { MICROPROFILE_SCOPE(Vulkan_Drawing); - SCOPE_EXIT({ gpu.TickWork(); }); + SCOPE_EXIT { + gpu.TickWork(); + }; FlushWork(); query_cache.NotifySegment(true); diff --git a/src/video_core/texture_cache/image_info.cpp b/src/video_core/texture_cache/image_info.cpp index b72788c6d..9444becce 100644 --- a/src/video_core/texture_cache/image_info.cpp +++ b/src/video_core/texture_cache/image_info.cpp @@ -42,6 +42,7 @@ ImageInfo::ImageInfo(const TICEntry& config) noexcept { }; } rescaleable = false; + is_sparse = config.is_sparse != 0; tile_width_spacing = config.tile_width_spacing; if (config.texture_type != TextureType::Texture2D && config.texture_type != TextureType::Texture2DNoMipmap) { diff --git a/src/video_core/texture_cache/image_info.h b/src/video_core/texture_cache/image_info.h index 8a4cb0cbd..eb490a642 100644 --- a/src/video_core/texture_cache/image_info.h +++ b/src/video_core/texture_cache/image_info.h @@ -41,6 +41,7 @@ struct ImageInfo { bool downscaleable = false; bool forced_flushed = false; bool dma_downloaded = false; + bool is_sparse = false; }; } // namespace VideoCommon diff --git a/src/video_core/texture_cache/texture_cache.h b/src/video_core/texture_cache/texture_cache.h index ca0794214..53b4876f2 100644 --- a/src/video_core/texture_cache/texture_cache.h +++ b/src/video_core/texture_cache/texture_cache.h @@ -600,17 +600,17 @@ void TextureCache<P>::UnmapGPUMemory(size_t as_id, GPUVAddr gpu_addr, size_t siz [&](ImageId id, Image&) { deleted_images.push_back(id); }); for (const ImageId id : deleted_images) { Image& image = slot_images[id]; - if (True(image.flags & ImageFlagBits::CpuModified)) { - continue; + if (False(image.flags & ImageFlagBits::CpuModified)) { + image.flags |= ImageFlagBits::CpuModified; + if (True(image.flags & ImageFlagBits::Tracked)) { + UntrackImage(image, id); + } } - image.flags |= ImageFlagBits::CpuModified; + if (True(image.flags & ImageFlagBits::Remapped)) { continue; } image.flags |= ImageFlagBits::Remapped; - if (True(image.flags & ImageFlagBits::Tracked)) { - UntrackImage(image, id); - } } } @@ -746,7 +746,13 @@ std::pair<typename P::ImageView*, bool> TextureCache<P>::TryFindFramebufferImage }(); const auto GetImageViewForFramebuffer = [&](ImageId image_id) { - const ImageViewInfo info{ImageViewType::e2D, view_format}; + ImageViewInfo info{ImageViewType::e2D, view_format}; + if (config.blending == Tegra::BlendMode::Opaque) { + info.x_source = static_cast<u8>(SwizzleSource::R); + info.y_source = static_cast<u8>(SwizzleSource::G); + info.z_source = static_cast<u8>(SwizzleSource::B); + info.w_source = static_cast<u8>(SwizzleSource::OneFloat); + } return std::make_pair(&slot_image_views[FindOrEmplaceImageView(image_id, info)], slot_images[image_id].IsRescaled()); }; @@ -1463,7 +1469,8 @@ ImageId TextureCache<P>::JoinImages(const ImageInfo& info, GPUVAddr gpu_addr, DA const ImageId new_image_id = slot_images.insert(runtime, new_info, gpu_addr, cpu_addr); Image& new_image = slot_images[new_image_id]; - if (!gpu_memory->IsContinuousRange(new_image.gpu_addr, new_image.guest_size_bytes)) { + if (!gpu_memory->IsContinuousRange(new_image.gpu_addr, new_image.guest_size_bytes) && + new_info.is_sparse) { new_image.flags |= ImageFlagBits::Sparse; } diff --git a/src/video_core/vulkan_common/nsight_aftermath_tracker.cpp b/src/video_core/vulkan_common/nsight_aftermath_tracker.cpp index 5fa0d9620..f41c3e506 100644 --- a/src/video_core/vulkan_common/nsight_aftermath_tracker.cpp +++ b/src/video_core/vulkan_common/nsight_aftermath_tracker.cpp @@ -116,7 +116,9 @@ void NsightAftermathTracker::OnGpuCrashDumpCallback(const void* gpu_crash_dump, LOG_ERROR(Render_Vulkan, "Failed to create decoder"); return; } - SCOPE_EXIT({ GFSDK_Aftermath_GpuCrashDump_DestroyDecoder(decoder); }); + SCOPE_EXIT { + GFSDK_Aftermath_GpuCrashDump_DestroyDecoder(decoder); + }; u32 json_size = 0; if (!GFSDK_Aftermath_SUCCEED(GFSDK_Aftermath_GpuCrashDump_GenerateJSON( |