Diamond75/Panda3DS
1
0
Fork 0
mirror of https://github.com/wheremyfoodat/Panda3DS.git synced 2025-04-07 14:45:41 +12:00
Code Issues Projects Releases Packages Wiki Activity Actions 9

fix: fog

Browse source
This commit is contained in:
Samuliak 2024-07-24 16:19:59 +02:00
parent 4a9fb9bdc3
commit 37cf0ad0b2
6 changed files with 37 additions and 44 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
CMakeLists.txt
View file

@ -445,7 +445,7 @@ if(ENABLE_METAL AND APPLE)
# TODO: only include sources in debug builds
add_custom_command(
OUTPUT ${SHADER_IR}
COMMAND xcrun -sdk macosx metal -gline-tables-only -frecord-sources -o ${SHADER_IR} -c ${SHADER_SOURCE}
COMMAND xcrun -sdk macosx metal -gline-tables-only -frecord-sources -fno-fast-math -o ${SHADER_IR} -c ${SHADER_SOURCE}
DEPENDS ${SHADER_SOURCE}
VERBATIM)
add_custom_command(

8
include/renderer_mtl/mtl_vertex_buffer_cache.hpp
View file

@ -58,12 +58,14 @@ public:
void reset() {
endFrame();
buffer->release();
create();
if (buffer) {
buffer->release();
create();
}
}
private:
MTL::Buffer* buffer;
MTL::Buffer* buffer = nullptr;
size_t ptr = 0;
std::vector<MTL::Buffer*> additionalAllocations;

4
include/renderer_mtl/pica_to_mtl.hpp
View file

@ -35,8 +35,8 @@ inline MTL::PixelFormat toMTLPixelFormatColor(ColorFmt format) {
switch (format) {
case ColorFmt::RGBA8: return MTL::PixelFormatRGBA8Unorm;
case ColorFmt::RGB8: return MTL::PixelFormatRGBA8Unorm;
case ColorFmt::RGBA5551: return MTL::PixelFormatBGR5A1Unorm;
case ColorFmt::RGB565: return MTL::PixelFormatB5G6R5Unorm;
case ColorFmt::RGBA5551: return MTL::PixelFormatRGBA8Unorm; // TODO: use MTL::PixelFormatBGR5A1Unorm?
case ColorFmt::RGB565: return MTL::PixelFormatRGBA8Unorm; // TODO: use MTL::PixelFormatB5G6R5Unorm?
case ColorFmt::RGBA4: return MTL::PixelFormatABGR4Unorm;
}
}

10
src/core/renderer_mtl/renderer_mtl.cpp
View file

@ -563,7 +563,6 @@ void RendererMTL::drawVertices(PICA::PrimType primType, std::span<const PICA::Ve
renderCommandEncoder->setFragmentBytes(&regs[0x48], (0x200 - 0x48) * sizeof(regs[0]), 0);
renderCommandEncoder->setVertexBytes(&depthUniforms, sizeof(depthUniforms), 2);
renderCommandEncoder->setFragmentBytes(&logicOp, sizeof(logicOp), 2);
renderCommandEncoder->setFragmentBytes(&depthUniforms, sizeof(depthUniforms), 3);
renderCommandEncoder->drawPrimitives(toMTLPrimitiveType(primType), NS::UInteger(0), NS::UInteger(vertices.size()));
}
@ -719,7 +718,7 @@ void RendererMTL::updateLightingLUT(MTL::RenderCommandEncoder* encoder) {
u32 arrayOffset = 0;
renderCommandEncoder->setVertexBytes(&arrayOffset, sizeof(u32), 1);
renderCommandEncoder->drawPrimitives(MTL::PrimitiveTypeTriangle, NS::UInteger(0), GPU::LightingLutSize);
renderCommandEncoder->drawPrimitives(MTL::PrimitiveTypeTriangleStrip, NS::UInteger(0), GPU::LightingLutSize);
}
void RendererMTL::updateFogLUT(MTL::RenderCommandEncoder* encoder) {
@ -740,12 +739,13 @@ void RendererMTL::updateFogLUT(MTL::RenderCommandEncoder* encoder) {
renderCommandEncoder->setRenderPipelineState(copyToLutTexturePipeline);
renderCommandEncoder->setDepthStencilState(defaultDepthStencilState);
renderCommandEncoder->setVertexTexture(lutTexture, 0);
Metal::BufferHandle buffer = vertexBufferCache.get(fogLut.data(), sizeof(fogLut));
renderCommandEncoder->setVertexBuffer(buffer.buffer, buffer.offset, 0);
//Metal::BufferHandle buffer = vertexBufferCache.get(fogLut.data(), sizeof(fogLut));
//renderCommandEncoder->setVertexBuffer(buffer.buffer, buffer.offset, 0);
renderCommandEncoder->setVertexBytes(fogLut.data(), sizeof(fogLut), 0);
u32 arrayOffset = (u32)Lights::LUT_Count;
renderCommandEncoder->setVertexBytes(&arrayOffset, sizeof(u32), 1);
renderCommandEncoder->drawPrimitives(MTL::PrimitiveTypeTriangle, NS::UInteger(0), NS::UInteger(128));
renderCommandEncoder->drawPrimitives(MTL::PrimitiveTypeTriangleStrip, NS::UInteger(0), NS::UInteger(128));
}
void RendererMTL::textureCopyImpl(Metal::ColorRenderTarget& srcFramebuffer, Metal::ColorRenderTarget& destFramebuffer, const Math::Rect<u32>& srcRect, const Math::Rect<u32>& destRect) {

4
src/host_shaders/metal_copy_to_lut_texture.metal
View file

@ -4,6 +4,6 @@ using namespace metal;
constant ushort lutTextureWidth [[function_constant(0)]];
// The copy is done in a vertex shader instead of a compute kernel, since dispatching compute would require ending the render pass
vertex void vertexCopyToLutTexture(uint vid [[vertex_id]], texture2d<float, access::write> out [[texture(0)]], device float2* data [[buffer(0)]], constant uint& arrayOffset [[buffer(1)]]) {
out.write(float4(data[vid].x, 0.0, 0.0, 0.0), uint2(vid % lutTextureWidth, arrayOffset + vid / lutTextureWidth));
vertex void vertexCopyToLutTexture(uint vid [[vertex_id]], texture2d<float, access::write> out [[texture(0)]], constant float2* data [[buffer(0)]], constant uint& arrayOffset [[buffer(1)]]) {
out.write(float4(data[vid], 0.0, 0.0), uint2(vid % lutTextureWidth, arrayOffset + vid / lutTextureWidth));
}

53
src/host_shaders/metal_shaders.metal
View file

@ -102,27 +102,6 @@ struct EnvColor {
}
};
struct DrawVertexOut {
float4 position [[position]];
float depth;
float4 quaternion;
float4 color;
float3 texCoord0;
float2 texCoord1;
float2 texCoord2;
float3 view;
float3 normal;
float3 tangent;
float3 bitangent;
EnvColor textureEnvColor [[flat]];
float4 textureEnvBufferColor [[flat]];
};
struct DrawVertexOutWithClip {
DrawVertexOut out;
float clipDistance [[clip_distance]] [2];
};
float3 rotateFloat3ByQuaternion(float3 v, float4 q) {
float3 u = q.xyz;
float s = q.w;
@ -157,6 +136,26 @@ struct DepthUniforms {
bool depthMapEnable;
};
struct DrawVertexOut {
float4 position [[position]];
float4 quaternion;
float4 color;
float3 texCoord0;
float2 texCoord1;
float2 texCoord2;
float3 view;
float3 normal;
float3 tangent;
float3 bitangent;
EnvColor textureEnvColor [[flat]];
float4 textureEnvBufferColor [[flat]];
};
struct DrawVertexOutWithClip {
DrawVertexOut out;
float clipDistance [[clip_distance]] [2];
};
// TODO: check this
float transformZ(float z, float w, constant DepthUniforms& depthUniforms) {
z = z / w * depthUniforms.depthScale + depthUniforms.depthOffset;
@ -177,7 +176,6 @@ vertex DrawVertexOutWithClip vertexDraw(DrawVertexIn in [[stage_in]], constant P
// Apply depth uniforms
out.position.z = transformZ(out.position.z, out.position.w, depthUniforms);
out.depth = out.position.z;
// Color
out.color = min(abs(in.color), 1.0);
@ -678,7 +676,7 @@ float4 performLogicOp(LogicOp logicOp, float4 s, float4 d) {
return as_type<float4>(performLogicOpU(logicOp, as_type<uint4>(s), as_type<uint4>(d)));
}
fragment float4 fragmentDraw(DrawVertexOut in [[stage_in]], float4 prevColor [[color(0)]], constant PicaRegs& picaRegs [[buffer(0)]], constant FragTEV& tev [[buffer(1)]], constant LogicOp& logicOp [[buffer(2)]], constant DepthUniforms& depthUniforms [[buffer(3)]],
fragment float4 fragmentDraw(DrawVertexOut in [[stage_in]], float4 prevColor [[color(0)]], constant PicaRegs& picaRegs [[buffer(0)]], constant FragTEV& tev [[buffer(1)]], constant LogicOp& logicOp [[buffer(2)]],
texture2d<float> tex0 [[texture(0)]], texture2d<float> tex1 [[texture(1)]], texture2d<float> tex2 [[texture(2)]], texture2d<float> texLut [[texture(3)]],
sampler samplr0 [[sampler(0)]], sampler samplr1 [[sampler(1)]], sampler samplr2 [[sampler(2)]], sampler linearSampler [[sampler(3)]]) {
Globals globals;
@ -727,19 +725,12 @@ fragment float4 fragmentDraw(DrawVertexOut in [[stage_in]], float4 prevColor [[c
float4 color = globals.tevSources[15];
// Depth
float z_over_w = in.position.z;
float depth = z_over_w * depthUniforms.depthScale + depthUniforms.depthOffset;
if (!depthUniforms.depthMapEnable) // Divide z by w if depthmap enable == 0 (ie using W-buffering)
depth /= in.position.w;
// Fog
bool enable_fog = (textureEnvUpdateBuffer & 7u) == 5u;
if (enable_fog) {
bool flip_depth = (textureEnvUpdateBuffer & (1u << 16)) != 0u;
float fog_index = flip_depth ? 1.0 - depth : depth;
float fog_index = flip_depth ? 1.0 - in.position.z : in.position.z;
fog_index *= 128.0;
float clamped_index = clamp(floor(fog_index), 0.0, 127.0);
float delta = fog_index - clamped_index;