diff options
Diffstat (limited to 'src')
-rw-r--r-- | src/audio_core/hle/source.cpp | 49 | ||||
-rw-r--r-- | src/audio_core/interpolate.cpp | 86 | ||||
-rw-r--r-- | src/audio_core/interpolate.h | 27 | ||||
-rw-r--r-- | src/video_core/renderer_opengl/gl_shader_gen.cpp | 10 | ||||
-rw-r--r-- | src/video_core/swrasterizer/lighting.cpp | 39 | ||||
-rw-r--r-- | src/video_core/swrasterizer/lighting.h | 3 | ||||
-rw-r--r-- | src/video_core/swrasterizer/rasterizer.cpp | 4 |
7 files changed, 120 insertions, 98 deletions
diff --git a/src/audio_core/hle/source.cpp b/src/audio_core/hle/source.cpp index 92484c526..de4e88cae 100644 --- a/src/audio_core/hle/source.cpp +++ b/src/audio_core/hle/source.cpp @@ -244,17 +244,27 @@ void Source::GenerateFrame() { break; } - const size_t size_to_copy = - std::min(state.current_buffer.size(), current_frame.size() - frame_position); - - std::copy(state.current_buffer.begin(), state.current_buffer.begin() + size_to_copy, - current_frame.begin() + frame_position); - state.current_buffer.erase(state.current_buffer.begin(), - state.current_buffer.begin() + size_to_copy); - - frame_position += size_to_copy; - state.next_sample_number += static_cast<u32>(size_to_copy); + switch (state.interpolation_mode) { + case InterpolationMode::None: + AudioInterp::None(state.interp_state, state.current_buffer, state.rate_multiplier, + current_frame, frame_position); + break; + case InterpolationMode::Linear: + AudioInterp::Linear(state.interp_state, state.current_buffer, state.rate_multiplier, + current_frame, frame_position); + break; + case InterpolationMode::Polyphase: + // TODO(merry): Implement polyphase interpolation + LOG_DEBUG(Audio_DSP, "Polyphase interpolation unimplemented; falling back to linear"); + AudioInterp::Linear(state.interp_state, state.current_buffer, state.rate_multiplier, + current_frame, frame_position); + break; + default: + UNIMPLEMENTED(); + break; + } } + state.next_sample_number += frame_position; state.filters.ProcessFrame(current_frame); } @@ -305,25 +315,6 @@ bool Source::DequeueBuffer() { return true; } - switch (state.interpolation_mode) { - case InterpolationMode::None: - state.current_buffer = - AudioInterp::None(state.interp_state, state.current_buffer, state.rate_multiplier); - break; - case InterpolationMode::Linear: - state.current_buffer = - AudioInterp::Linear(state.interp_state, state.current_buffer, state.rate_multiplier); - break; - case InterpolationMode::Polyphase: - // TODO(merry): Implement polyphase interpolation - state.current_buffer = - AudioInterp::Linear(state.interp_state, state.current_buffer, state.rate_multiplier); - break; - default: - UNIMPLEMENTED(); - break; - } - // the first playthrough starts at play_position, loops start at the beginning of the buffer state.current_sample_number = (!buf.has_played) ? buf.play_position : 0; state.next_sample_number = state.current_sample_number; diff --git a/src/audio_core/interpolate.cpp b/src/audio_core/interpolate.cpp index 8a5d4181a..16e68bc5c 100644 --- a/src/audio_core/interpolate.cpp +++ b/src/audio_core/interpolate.cpp @@ -13,74 +13,64 @@ namespace AudioInterp { constexpr u64 scale_factor = 1 << 24; constexpr u64 scale_mask = scale_factor - 1; -/// Here we step over the input in steps of rate_multiplier, until we consume all of the input. +/// Here we step over the input in steps of rate, until we consume all of the input. /// Three adjacent samples are passed to fn each step. template <typename Function> -static StereoBuffer16 StepOverSamples(State& state, const StereoBuffer16& input, - float rate_multiplier, Function fn) { - ASSERT(rate_multiplier > 0); +static void StepOverSamples(State& state, StereoBuffer16& input, float rate, + DSP::HLE::StereoFrame16& output, size_t& outputi, Function fn) { + ASSERT(rate > 0); - if (input.size() < 2) - return {}; + if (input.empty()) + return; - StereoBuffer16 output; - output.reserve(static_cast<size_t>(input.size() / rate_multiplier)); + input.insert(input.begin(), {state.xn2, state.xn1}); - u64 step_size = static_cast<u64>(rate_multiplier * scale_factor); + const u64 step_size = static_cast<u64>(rate * scale_factor); + u64 fposition = state.fposition; + size_t inputi = 0; - u64 fposition = 0; - const u64 max_fposition = input.size() * scale_factor; + while (outputi < output.size()) { + inputi = static_cast<size_t>(fposition / scale_factor); - while (fposition < 1 * scale_factor) { - u64 fraction = fposition & scale_mask; - - output.push_back(fn(fraction, state.xn2, state.xn1, input[0])); - - fposition += step_size; - } - - while (fposition < 2 * scale_factor) { - u64 fraction = fposition & scale_mask; - - output.push_back(fn(fraction, state.xn1, input[0], input[1])); - - fposition += step_size; - } + if (inputi + 2 >= input.size()) { + inputi = input.size() - 2; + break; + } - while (fposition < max_fposition) { u64 fraction = fposition & scale_mask; - - size_t index = static_cast<size_t>(fposition / scale_factor); - output.push_back(fn(fraction, input[index - 2], input[index - 1], input[index])); + output[outputi++] = fn(fraction, input[inputi], input[inputi + 1], input[inputi + 2]); fposition += step_size; } - state.xn2 = input[input.size() - 2]; - state.xn1 = input[input.size() - 1]; + state.xn2 = input[inputi]; + state.xn1 = input[inputi + 1]; + state.fposition = fposition - inputi * scale_factor; - return output; + input.erase(input.begin(), input.begin() + inputi + 2); } -StereoBuffer16 None(State& state, const StereoBuffer16& input, float rate_multiplier) { - return StepOverSamples( - state, input, rate_multiplier, +void None(State& state, StereoBuffer16& input, float rate, DSP::HLE::StereoFrame16& output, + size_t& outputi) { + StepOverSamples( + state, input, rate, output, outputi, [](u64 fraction, const auto& x0, const auto& x1, const auto& x2) { return x0; }); } -StereoBuffer16 Linear(State& state, const StereoBuffer16& input, float rate_multiplier) { +void Linear(State& state, StereoBuffer16& input, float rate, DSP::HLE::StereoFrame16& output, + size_t& outputi) { // Note on accuracy: Some values that this produces are +/- 1 from the actual firmware. - return StepOverSamples(state, input, rate_multiplier, - [](u64 fraction, const auto& x0, const auto& x1, const auto& x2) { - // This is a saturated subtraction. (Verified by black-box fuzzing.) - s64 delta0 = MathUtil::Clamp<s64>(x1[0] - x0[0], -32768, 32767); - s64 delta1 = MathUtil::Clamp<s64>(x1[1] - x0[1], -32768, 32767); - - return std::array<s16, 2>{ - static_cast<s16>(x0[0] + fraction * delta0 / scale_factor), - static_cast<s16>(x0[1] + fraction * delta1 / scale_factor), - }; - }); + StepOverSamples(state, input, rate, output, outputi, + [](u64 fraction, const auto& x0, const auto& x1, const auto& x2) { + // This is a saturated subtraction. (Verified by black-box fuzzing.) + s64 delta0 = MathUtil::Clamp<s64>(x1[0] - x0[0], -32768, 32767); + s64 delta1 = MathUtil::Clamp<s64>(x1[1] - x0[1], -32768, 32767); + + return std::array<s16, 2>{ + static_cast<s16>(x0[0] + fraction * delta0 / scale_factor), + static_cast<s16>(x0[1] + fraction * delta1 / scale_factor), + }; + }); } } // namespace AudioInterp diff --git a/src/audio_core/interpolate.h b/src/audio_core/interpolate.h index 19a7b66cb..59f59bc14 100644 --- a/src/audio_core/interpolate.h +++ b/src/audio_core/interpolate.h @@ -6,6 +6,7 @@ #include <array> #include <vector> +#include "audio_core/hle/common.h" #include "common/common_types.h" namespace AudioInterp { @@ -14,31 +15,35 @@ namespace AudioInterp { using StereoBuffer16 = std::vector<std::array<s16, 2>>; struct State { - // Two historical samples. + /// Two historical samples. std::array<s16, 2> xn1 = {}; ///< x[n-1] std::array<s16, 2> xn2 = {}; ///< x[n-2] + /// Current fractional position. + u64 fposition = 0; }; /** * No interpolation. This is equivalent to a zero-order hold. There is a two-sample predelay. * @param state Interpolation state. * @param input Input buffer. - * @param rate_multiplier Stretch factor. Must be a positive non-zero value. - * rate_multiplier > 1.0 performs decimation and rate_multipler < 1.0 - * performs upsampling. - * @return The resampled audio buffer. + * @param rate Stretch factor. Must be a positive non-zero value. + * rate > 1.0 performs decimation and rate < 1.0 performs upsampling. + * @param output The resampled audio buffer. + * @param outputi The index of output to start writing to. */ -StereoBuffer16 None(State& state, const StereoBuffer16& input, float rate_multiplier); +void None(State& state, StereoBuffer16& input, float rate, DSP::HLE::StereoFrame16& output, + size_t& outputi); /** * Linear interpolation. This is equivalent to a first-order hold. There is a two-sample predelay. * @param state Interpolation state. * @param input Input buffer. - * @param rate_multiplier Stretch factor. Must be a positive non-zero value. - * rate_multiplier > 1.0 performs decimation and rate_multipler < 1.0 - * performs upsampling. - * @return The resampled audio buffer. + * @param rate Stretch factor. Must be a positive non-zero value. + * rate > 1.0 performs decimation and rate < 1.0 performs upsampling. + * @param output The resampled audio buffer. + * @param outputi The index of output to start writing to. */ -StereoBuffer16 Linear(State& state, const StereoBuffer16& input, float rate_multiplier); +void Linear(State& state, StereoBuffer16& input, float rate, DSP::HLE::StereoFrame16& output, + size_t& outputi); } // namespace AudioInterp diff --git a/src/video_core/renderer_opengl/gl_shader_gen.cpp b/src/video_core/renderer_opengl/gl_shader_gen.cpp index 015e69da9..c8fc7a0ff 100644 --- a/src/video_core/renderer_opengl/gl_shader_gen.cpp +++ b/src/video_core/renderer_opengl/gl_shader_gen.cpp @@ -8,6 +8,7 @@ #include "common/assert.h" #include "common/bit_field.h" #include "common/logging/log.h" +#include "core/core.h" #include "video_core/regs_framebuffer.h" #include "video_core/regs_lighting.h" #include "video_core/regs_rasterizer.h" @@ -594,8 +595,8 @@ static void WriteLighting(std::string& out, const PicaShaderConfig& config) { // Note: even if the normal vector is modified by normal map, which is not the // normal of the tangent plane anymore, the half angle vector is still projected // using the modified normal vector. - std::string half_angle_proj = "normalize(half_vector) - normal / dot(normal, " - "normal) * dot(normal, normalize(half_vector))"; + std::string half_angle_proj = + "normalize(half_vector) - normal * dot(normal, normalize(half_vector))"; // Note: the half angle vector projection is confirmed not normalized before the dot // product. The result is in fact not cos(phi) as the name suggested. index = "dot(" + half_angle_proj + ", tangent)"; @@ -1155,6 +1156,11 @@ vec4 secondary_fragment_color = vec4(0.0); // Blend the fog out += "last_tex_env_out.rgb = mix(fog_color.rgb, last_tex_env_out.rgb, fog_factor);\n"; + } else if (state.fog_mode == TexturingRegs::FogMode::Gas) { + Core::Telemetry().AddField(Telemetry::FieldType::Session, "VideoCore_Pica_UseGasMode", + true); + LOG_CRITICAL(Render_OpenGL, "Unimplemented gas mode"); + UNIMPLEMENTED(); } out += "gl_FragDepth = depth;\n"; diff --git a/src/video_core/swrasterizer/lighting.cpp b/src/video_core/swrasterizer/lighting.cpp index 39a3e396d..b38964530 100644 --- a/src/video_core/swrasterizer/lighting.cpp +++ b/src/video_core/swrasterizer/lighting.cpp @@ -22,18 +22,37 @@ static float LookupLightingLut(const Pica::State::Lighting& lighting, size_t lut std::tuple<Math::Vec4<u8>, Math::Vec4<u8>> ComputeFragmentsColors( const Pica::LightingRegs& lighting, const Pica::State::Lighting& lighting_state, - const Math::Quaternion<float>& normquat, const Math::Vec3<float>& view) { + const Math::Quaternion<float>& normquat, const Math::Vec3<float>& view, + const Math::Vec4<u8> (&texture_color)[4]) { - // TODO(Subv): Bump mapping - Math::Vec3<float> surface_normal = {0.0f, 0.0f, 1.0f}; + Math::Vec3<float> surface_normal; + Math::Vec3<float> surface_tangent; if (lighting.config0.bump_mode != LightingRegs::LightingBumpMode::None) { - LOG_CRITICAL(HW_GPU, "unimplemented bump mapping"); - UNIMPLEMENTED(); + Math::Vec3<float> perturbation = + texture_color[lighting.config0.bump_selector].xyz().Cast<float>() / 127.5f - + Math::MakeVec(1.0f, 1.0f, 1.0f); + if (lighting.config0.bump_mode == LightingRegs::LightingBumpMode::NormalMap) { + if (!lighting.config0.disable_bump_renorm) { + const float z_square = 1 - perturbation.xy().Length2(); + perturbation.z = std::sqrt(std::max(z_square, 0.0f)); + } + surface_normal = perturbation; + surface_tangent = Math::MakeVec(1.0f, 0.0f, 0.0f); + } else if (lighting.config0.bump_mode == LightingRegs::LightingBumpMode::TangentMap) { + surface_normal = Math::MakeVec(0.0f, 0.0f, 1.0f); + surface_tangent = perturbation; + } else { + LOG_ERROR(HW_GPU, "Unknown bump mode %u", lighting.config0.bump_mode.Value()); + } + } else { + surface_normal = Math::MakeVec(0.0f, 0.0f, 1.0f); + surface_tangent = Math::MakeVec(1.0f, 0.0f, 0.0f); } // Use the normalized the quaternion when performing the rotation auto normal = Math::QuaternionRotate(normquat, surface_normal); + auto tangent = Math::QuaternionRotate(normquat, surface_tangent); Math::Vec4<float> diffuse_sum = {0.0f, 0.0f, 0.0f, 1.0f}; Math::Vec4<float> specular_sum = {0.0f, 0.0f, 0.0f, 1.0f}; @@ -102,6 +121,16 @@ std::tuple<Math::Vec4<u8>, Math::Vec4<u8>> ComputeFragmentsColors( result = Math::Dot(light_vector, spot_dir.Cast<float>() / 2047.0f); break; } + case LightingRegs::LightingLutInput::CP: + if (lighting.config0.config == LightingRegs::LightingConfig::Config7) { + const Math::Vec3<float> norm_half_vector = half_vector.Normalized(); + const Math::Vec3<float> half_vector_proj = + norm_half_vector - normal * Math::Dot(normal, norm_half_vector); + result = Math::Dot(half_vector_proj, tangent); + } else { + result = 0.0f; + } + break; default: LOG_CRITICAL(HW_GPU, "Unknown lighting LUT input %u\n", static_cast<u32>(input)); UNIMPLEMENTED(); diff --git a/src/video_core/swrasterizer/lighting.h b/src/video_core/swrasterizer/lighting.h index 438dca926..d807a3d94 100644 --- a/src/video_core/swrasterizer/lighting.h +++ b/src/video_core/swrasterizer/lighting.h @@ -13,6 +13,7 @@ namespace Pica { std::tuple<Math::Vec4<u8>, Math::Vec4<u8>> ComputeFragmentsColors( const Pica::LightingRegs& lighting, const Pica::State::Lighting& lighting_state, - const Math::Quaternion<float>& normquat, const Math::Vec3<float>& view); + const Math::Quaternion<float>& normquat, const Math::Vec3<float>& view, + const Math::Vec4<u8> (&texture_color)[4]); } // namespace Pica diff --git a/src/video_core/swrasterizer/rasterizer.cpp b/src/video_core/swrasterizer/rasterizer.cpp index fdc1df199..862135614 100644 --- a/src/video_core/swrasterizer/rasterizer.cpp +++ b/src/video_core/swrasterizer/rasterizer.cpp @@ -437,8 +437,8 @@ static void ProcessTriangleInternal(const Vertex& v0, const Vertex& v1, const Ve GetInterpolatedAttribute(v0.view.y, v1.view.y, v2.view.y).ToFloat32(), GetInterpolatedAttribute(v0.view.z, v1.view.z, v2.view.z).ToFloat32(), }; - std::tie(primary_fragment_color, secondary_fragment_color) = - ComputeFragmentsColors(g_state.regs.lighting, g_state.lighting, normquat, view); + std::tie(primary_fragment_color, secondary_fragment_color) = ComputeFragmentsColors( + g_state.regs.lighting, g_state.lighting, normquat, view, texture_color); } for (unsigned tev_stage_index = 0; tev_stage_index < tev_stages.size(); |