zdoom/gzdoom-gles
0
0
Fork 0
mirror of https://github.com/ZDoom/gzdoom-gles.git synced 2024-11-16 01:21:17 +00:00
Code Issues Projects Releases Packages Wiki Activity

- replace softpoly block drawers with span drawers and make them use blending rules directly from render styles

Browse source 
- add dynlights to softpoly pal mode

(cherry picked from commit 4e6226fc2d)
This commit is contained in:
Magnus Norddahl 2018-05-29 23:32:57 +02:00 committed by drfrag666
parent 1055e45566
commit 60929f353f
12 changed files with 626 additions and 1383 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

12
src/polyrenderer/drawers/poly_draw_args.cpp
View file

@ -52,7 +52,7 @@ void PolyDrawArgs::SetTexture(FTexture *texture, FRenderStyle style)
mTexture = texture;
mTextureWidth = texture->GetWidth();
mTextureHeight = texture->GetHeight();
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
mTexturePixels = (const uint8_t *)texture->GetPixelsBgra();
else
mTexturePixels = texture->GetPixels(style);
@ -67,7 +67,7 @@ void PolyDrawArgs::SetTexture(FTexture *texture, uint32_t translationID, FRender
FRemapTable *table = TranslationToTable(translationID);
if (table != nullptr && !table->Inactive)
{
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
mTranslation = (uint8_t*)table->Palette;
else
mTranslation = table->Remap;
@ -121,7 +121,7 @@ void PolyDrawArgs::SetLight(FSWColormap *base_colormap, uint32_t lightlevel, dou
void PolyDrawArgs::SetColor(uint32_t bgra, uint8_t palindex)
{
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
{
mColor = bgra;
}
@ -217,7 +217,7 @@ void RectDrawArgs::SetTexture(FTexture *texture, FRenderStyle style)
mTexture = texture;
mTextureWidth = texture->GetWidth();
mTextureHeight = texture->GetHeight();
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
mTexturePixels = (const uint8_t *)texture->GetPixelsBgra();
else
mTexturePixels = texture->GetPixels(style);
@ -231,7 +231,7 @@ void RectDrawArgs::SetTexture(FTexture *texture, uint32_t translationID, FRender
FRemapTable *table = TranslationToTable(translationID);
if (table != nullptr && !table->Inactive)
{
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
mTranslation = (uint8_t*)table->Palette;
else
mTranslation = table->Remap;
@ -275,7 +275,7 @@ void RectDrawArgs::SetLight(FSWColormap *base_colormap, uint32_t lightlevel)
void RectDrawArgs::SetColor(uint32_t bgra, uint8_t palindex)
{
if (PolyRenderer::Instance()->RenderTarget->IsBgra())
if (PolyTriangleDrawer::IsBgra())
{
mColor = bgra;
}

7
src/polyrenderer/drawers/poly_draw_args.h
View file

@ -75,7 +75,7 @@ public:
void SetWriteColor(bool enable) { mWriteColor = enable; }
void SetWriteStencil(bool enable, uint8_t stencilWriteValue = 0) { mWriteStencil = enable; mStencilWriteValue = stencilWriteValue; }
void SetWriteDepth(bool enable) { mWriteDepth = enable; }
void SetStyle(TriBlendMode blendmode, double alpha = 1.0) { mBlendMode = blendmode; mSrcAlpha = (uint32_t)(alpha * 256.0 + 0.5); }
void SetStyle(TriBlendMode blendmode, double alpha = 1.0) { mBlendMode = blendmode; mAlpha = (uint32_t)(alpha * 256.0 + 0.5); }
void SetStyle(const FRenderStyle &renderstyle, double alpha, uint32_t fillcolor, uint32_t translationID, FTexture *texture, bool fullbright);
void SetColor(uint32_t bgra, uint8_t palindex);
void SetLights(PolyLight *lights, int numLights) { mLights = lights; mNumLights = numLights; }
@ -107,8 +107,7 @@ public:
TriBlendMode BlendMode() const { return mBlendMode; }
uint32_t Color() const { return mColor; }
uint32_t SrcAlpha() const { return mSrcAlpha; }
uint32_t DestAlpha() const { return 256 - mSrcAlpha; }
uint32_t Alpha() const { return mAlpha; }
float GlobVis() const { return mGlobVis; }
uint32_t Light() const { return mLight; }
@ -155,7 +154,7 @@ private:
TriBlendMode mBlendMode = TriBlendMode::Fill;
uint32_t mLight = 0;
uint32_t mColor = 0;
uint32_t mSrcAlpha = 0;
uint32_t mAlpha = 0;
uint16_t mLightAlpha = 0;
uint16_t mLightRed = 0;
uint16_t mLightGreen = 0;

447
src/polyrenderer/drawers/poly_drawer32.h
View file

@ -327,453 +327,6 @@ namespace TriScreenDrawerModes
}
}
template<typename BlendT, typename SamplerT>
class TriScreenDrawer32
{
public:
static void Execute(int x, int y, uint32_t mask0, uint32_t mask1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
bool is_simple_shade = args->uniforms->SimpleShade();
if (SamplerT::Mode == (int)Samplers::Texture)
{
bool is_nearest_filter = args->uniforms->NearestFilter();
if (is_simple_shade)
{
if (is_nearest_filter)
DrawBlock<SimpleShade, NearestFilter>(x, y, mask0, mask1, args);
else
DrawBlock<SimpleShade, LinearFilter>(x, y, mask0, mask1, args);
}
else
{
if (is_nearest_filter)
DrawBlock<AdvancedShade, NearestFilter>(x, y, mask0, mask1, args);
else
DrawBlock<AdvancedShade, LinearFilter>(x, y, mask0, mask1, args);
}
}
else if (SamplerT::Mode == (int)Samplers::Fuzz)
{
DrawBlock<NoShade, NearestFilter>(x, y, mask0, mask1, args);
}
else // no linear filtering for translated, shaded, stencil, fill or skycap
{
if (is_simple_shade)
{
DrawBlock<SimpleShade, NearestFilter>(x, y, mask0, mask1, args);
}
else
{
DrawBlock<AdvancedShade, NearestFilter>(x, y, mask0, mask1, args);
}
}
}
private:
template<typename ShadeModeT, typename FilterModeT>
FORCEINLINE static void DrawBlock(int destX, int destY, uint32_t mask0, uint32_t mask1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
bool is_fixed_light = args->uniforms->FixedLight();
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
uint32_t srcalpha = args->uniforms->SrcAlpha();
uint32_t destalpha = args->uniforms->DestAlpha();
auto lights = args->uniforms->Lights();
auto num_lights = args->uniforms->NumLights();
FVector3 worldnormal = args->uniforms->Normal();
uint32_t dynlightcolor = args->uniforms->DynLightColor();
// Calculate gradients
const ShadedTriVertex &v1 = *args->v1;
ScreenTriangleStepVariables gradientX = args->gradientX;
ScreenTriangleStepVariables gradientY = args->gradientY;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = v1.w + gradientX.W * (destX - v1.x) + gradientY.W * (destY - v1.y);
blockPosY.U = v1.u * v1.w + gradientX.U * (destX - v1.x) + gradientY.U * (destY - v1.y);
blockPosY.V = v1.v * v1.w + gradientX.V * (destX - v1.x) + gradientY.V * (destY - v1.y);
blockPosY.WorldX = v1.worldX * v1.w + gradientX.WorldX * (destX - v1.x) + gradientY.WorldX * (destY - v1.y);
blockPosY.WorldY = v1.worldY * v1.w + gradientX.WorldY * (destX - v1.x) + gradientY.WorldY * (destY - v1.y);
blockPosY.WorldZ = v1.worldZ * v1.w + gradientX.WorldZ * (destX - v1.x) + gradientY.WorldZ * (destY - v1.y);
gradientX.W *= 8.0f;
gradientX.U *= 8.0f;
gradientX.V *= 8.0f;
gradientX.WorldX *= 8.0f;
gradientX.WorldY *= 8.0f;
gradientX.WorldZ *= 8.0f;
// Output
uint32_t * RESTRICT destOrg = (uint32_t*)args->dest;
int pitch = args->pitch;
uint32_t *dest = destOrg + destX + destY * pitch;
// Light
uint32_t light = args->uniforms->Light();
float shade = 2.0f - (light + 12.0f) / 128.0f;
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += (light >> 7); // 255 -> 256
// Sampling stuff
uint32_t color = args->uniforms->Color();
const uint32_t * RESTRICT translation = (const uint32_t *)args->uniforms->Translation();
const uint32_t * RESTRICT texPixels = (const uint32_t *)args->uniforms->TexturePixels();
uint32_t texWidth = args->uniforms->TextureWidth();
uint32_t texHeight = args->uniforms->TextureHeight();
uint32_t oneU, oneV;
if (SamplerT::Mode != (int)Samplers::Fill)
{
oneU = ((0x800000 + texWidth - 1) / texWidth) * 2 + 1;
oneV = ((0x800000 + texHeight - 1) / texHeight) * 2 + 1;
}
else
{
oneU = 0;
oneV = 0;
}
// Shade constants
int inv_desaturate;
BgraColor shade_fade, shade_light;
int desaturate;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
shade_fade.r = args->uniforms->ShadeFadeRed();
shade_fade.g = args->uniforms->ShadeFadeGreen();
shade_fade.b = args->uniforms->ShadeFadeBlue();
shade_light.r = args->uniforms->ShadeLightRed();
shade_light.g = args->uniforms->ShadeLightGreen();
shade_light.b = args->uniforms->ShadeLightBlue();
desaturate = args->uniforms->ShadeDesaturate();
inv_desaturate = 256 - desaturate;
}
else
{
inv_desaturate = 0;
shade_fade.r = 0;
shade_fade.g = 0;
shade_fade.b = 0;
shade_light.r = 0;
shade_light.g = 0;
shade_light.b = 0;
desaturate = 0;
}
if (mask0 == 0xffffffff && mask1 == 0xffffffff)
{
for (int y = 0; y < 8; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
FVector3 worldpos = FVector3(blockPosY.WorldX, blockPosY.WorldY, blockPosY.WorldZ) / blockPosY.W;
BgraColor dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
worldpos = FVector3(blockPosX.WorldX, blockPosX.WorldY, blockPosX.WorldZ) / blockPosX.W;
BgraColor dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
BgraColor dynlightstep;
dynlightstep.r = int32_t(dynlightnext.r - dynlight.r) >> 3;
dynlightstep.g = int32_t(dynlightnext.g - dynlight.g) >> 3;
dynlightstep.b = int32_t(dynlightnext.b - dynlight.b) >> 3;
for (int ix = 0; ix < 8; ix++)
{
// Load bgcolor
BgraColor bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
bgcolor = dest[ix];
else
bgcolor = 0;
// Sample fgcolor
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = dest[ix];
unsigned int ifgcolor = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
unsigned int ifgshade = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + ix, destY + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
BgraColor mlight;
mlight.r = lightpos0;
mlight.g = lightpos0;
mlight.b = lightpos0;
BgraColor shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
uint32_t inv_light = 256 - lightpos0;
shade_fade_lit.r = shade_fade.r * inv_light;
shade_fade_lit.g = shade_fade.g * inv_light;
shade_fade_lit.b = shade_fade.b * inv_light;
}
else
{
shade_fade_lit.r = 0;
shade_fade_lit.g = 0;
shade_fade_lit.b = 0;
}
// Shade and blend
BgraColor fgcolor = Shade32<ShadeModeT>(ifgcolor, mlight, desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
BgraColor outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor, ifgshade, srcalpha, destalpha);
// Store result
dest[ix] = outcolor;
dynlight.r = MAX<int32_t>(dynlight.r + dynlightstep.r, 0);
dynlight.g = MAX<int32_t>(dynlight.g + dynlightstep.g, 0);
dynlight.b = MAX<int32_t>(dynlight.b + dynlightstep.b, 0);
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
}
else
{
// mask0 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
FVector3 worldpos = FVector3(blockPosY.WorldX, blockPosY.WorldY, blockPosY.WorldZ) / blockPosY.W;
BgraColor dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
worldpos = FVector3(blockPosX.WorldX, blockPosX.WorldY, blockPosX.WorldZ) / blockPosX.W;
BgraColor dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
BgraColor dynlightstep;
dynlightstep.r = int32_t(dynlightnext.r - dynlight.r) >> 3;
dynlightstep.g = int32_t(dynlightnext.g - dynlight.g) >> 3;
dynlightstep.b = int32_t(dynlightnext.b - dynlight.b) >> 3;
for (int x = 0; x < 8; x++)
{
// Load bgcolor
BgraColor bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
{
if (mask0 & (1 << 31)) bgcolor = dest[x];
}
else
bgcolor = 0;
// Sample fgcolor
if (SamplerT::Mode == (int)Samplers::FogBoundary && (mask0 & (1 << 31))) color = dest[x];
unsigned int ifgcolor = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
unsigned int ifgshade = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x, destY + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
BgraColor mlight;
mlight.r = lightpos0;
mlight.g = lightpos0;
mlight.b = lightpos0;
BgraColor shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
uint32_t inv_light = 256 - lightpos0;
shade_fade_lit.r = shade_fade.r * inv_light;
shade_fade_lit.g = shade_fade.g * inv_light;
shade_fade_lit.b = shade_fade.b * inv_light;
}
else
{
shade_fade_lit.r = 0;
shade_fade_lit.g = 0;
shade_fade_lit.b = 0;
}
// Shade and blend
BgraColor fgcolor = Shade32<ShadeModeT>(ifgcolor, mlight, desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
BgraColor outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor, ifgshade, srcalpha, destalpha);
// Store result
if (mask0 & (1 << 31)) dest[x] = outcolor;
mask0 <<= 1;
dynlight.r = MAX<int32_t>(dynlight.r + dynlightstep.r, 0);
dynlight.g = MAX<int32_t>(dynlight.g + dynlightstep.g, 0);
dynlight.b = MAX<int32_t>(dynlight.b + dynlightstep.b, 0);
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
// mask1 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
FVector3 worldpos = FVector3(blockPosY.WorldX, blockPosY.WorldY, blockPosY.WorldZ) / blockPosY.W;
BgraColor dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
worldpos = FVector3(blockPosX.WorldX, blockPosX.WorldY, blockPosX.WorldZ) / blockPosX.W;
BgraColor dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
BgraColor dynlightstep;
dynlightstep.r = int32_t(dynlightnext.r - dynlight.r) >> 3;
dynlightstep.g = int32_t(dynlightnext.g - dynlight.g) >> 3;
dynlightstep.b = int32_t(dynlightnext.b - dynlight.b) >> 3;
for (int x = 0; x < 8; x++)
{
// Load bgcolor
BgraColor bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
{
if (mask1 & (1 << 31)) bgcolor = dest[x];
}
else
bgcolor = 0;
// Sample fgcolor
if (SamplerT::Mode == (int)Samplers::FogBoundary && (mask1 & (1 << 31))) color = dest[x];
unsigned int ifgcolor = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
unsigned int ifgshade = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x, destY + 4 + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
BgraColor mlight;
mlight.r = lightpos0;
mlight.g = lightpos0;
mlight.b = lightpos0;
BgraColor shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
uint32_t inv_light = 256 - lightpos0;
shade_fade_lit.r = shade_fade.r * inv_light;
shade_fade_lit.g = shade_fade.g * inv_light;
shade_fade_lit.b = shade_fade.b * inv_light;
}
else
{
shade_fade_lit.r = 0;
shade_fade_lit.g = 0;
shade_fade_lit.b = 0;
}
// Shade and blend
BgraColor fgcolor = Shade32<ShadeModeT>(ifgcolor, mlight, desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
BgraColor outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor, ifgshade, srcalpha, destalpha);
// Store result
if (mask1 & (1 << 31)) dest[x] = outcolor;
mask1 <<= 1;
dynlight.r = MAX<int32_t>(dynlight.r + dynlightstep.r, 0);
dynlight.g = MAX<int32_t>(dynlight.g + dynlightstep.g, 0);
dynlight.b = MAX<int32_t>(dynlight.b + dynlightstep.b, 0);
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
}
}
};
template<typename BlendT, typename SamplerT>
class RectScreenDrawer32
{

455
src/polyrenderer/drawers/poly_drawer32_sse2.h
View file

@ -346,461 +346,6 @@ namespace TriScreenDrawerModes
}
}
template<typename BlendT, typename SamplerT>
class TriScreenDrawer32
{
public:
static void Execute(int x, int y, uint32_t mask0, uint32_t mask1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
bool is_simple_shade = args->uniforms->SimpleShade();
if (SamplerT::Mode == (int)Samplers::Texture)
{
bool is_nearest_filter = args->uniforms->NearestFilter();
if (is_simple_shade)
{
if (is_nearest_filter)
DrawBlock<SimpleShade, NearestFilter>(x, y, mask0, mask1, args);
else
DrawBlock<SimpleShade, LinearFilter>(x, y, mask0, mask1, args);
}
else
{
if (is_nearest_filter)
DrawBlock<AdvancedShade, NearestFilter>(x, y, mask0, mask1, args);
else
DrawBlock<AdvancedShade, LinearFilter>(x, y, mask0, mask1, args);
}
}
else if (SamplerT::Mode == (int)Samplers::Fuzz)
{
DrawBlock<NoShade, NearestFilter>(x, y, mask0, mask1, args);
}
else // no linear filtering for translated, shaded, stencil, fill or skycap
{
if (is_simple_shade)
{
DrawBlock<SimpleShade, NearestFilter>(x, y, mask0, mask1, args);
}
else
{
DrawBlock<AdvancedShade, NearestFilter>(x, y, mask0, mask1, args);
}
}
}
private:
template<typename ShadeModeT, typename FilterModeT>
FORCEINLINE static void VECTORCALL DrawBlock(int destX, int destY, uint32_t mask0, uint32_t mask1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
bool is_fixed_light = args->uniforms->FixedLight();
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
uint32_t srcalpha = args->uniforms->SrcAlpha();
uint32_t destalpha = args->uniforms->DestAlpha();
auto lights = args->uniforms->Lights();
auto num_lights = args->uniforms->NumLights();
__m128 worldnormal = _mm_setr_ps(args->uniforms->Normal().X, args->uniforms->Normal().Y, args->uniforms->Normal().Z, 0.0f);
uint32_t dynlightcolor = args->uniforms->DynLightColor();
// Calculate gradients
const ShadedTriVertex &v1 = *args->v1;
ScreenTriangleStepVariables gradientX = args->gradientX;
ScreenTriangleStepVariables gradientY = args->gradientY;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = v1.w + gradientX.W * (destX - v1.x) + gradientY.W * (destY - v1.y);
blockPosY.U = v1.u * v1.w + gradientX.U * (destX - v1.x) + gradientY.U * (destY - v1.y);
blockPosY.V = v1.v * v1.w + gradientX.V * (destX - v1.x) + gradientY.V * (destY - v1.y);
blockPosY.WorldX = v1.worldX * v1.w + gradientX.WorldX * (destX - v1.x) + gradientY.WorldX * (destY - v1.y);
blockPosY.WorldY = v1.worldY * v1.w + gradientX.WorldY * (destX - v1.x) + gradientY.WorldY * (destY - v1.y);
blockPosY.WorldZ = v1.worldZ * v1.w + gradientX.WorldZ * (destX - v1.x) + gradientY.WorldZ * (destY - v1.y);
gradientX.W *= 8.0f;
gradientX.U *= 8.0f;
gradientX.V *= 8.0f;
gradientX.WorldX *= 8.0f;
gradientX.WorldY *= 8.0f;
gradientX.WorldZ *= 8.0f;
// Output
uint32_t * RESTRICT destOrg = (uint32_t*)args->dest;
int pitch = args->pitch;
uint32_t *dest = destOrg + destX + destY * pitch;
// Light
uint32_t light = args->uniforms->Light();
float shade = 2.0f - (light + 12.0f) / 128.0f;
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += (light >> 7); // 255 -> 256
// Sampling stuff
uint32_t color = args->uniforms->Color();
const uint32_t * RESTRICT translation = (const uint32_t *)args->uniforms->Translation();
const uint32_t * RESTRICT texPixels = (const uint32_t *)args->uniforms->TexturePixels();
uint32_t texWidth = args->uniforms->TextureWidth();
uint32_t texHeight = args->uniforms->TextureHeight();
uint32_t oneU, oneV;
if (SamplerT::Mode != (int)Samplers::Fill)
{
oneU = ((0x800000 + texWidth - 1) / texWidth) * 2 + 1;
oneV = ((0x800000 + texHeight - 1) / texHeight) * 2 + 1;
}
else
{
oneU = 0;
oneV = 0;
}
// Shade constants
__m128i inv_desaturate, shade_fade, shade_light;
int desaturate;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
inv_desaturate = _mm_setr_epi16(256, 256 - args->uniforms->ShadeDesaturate(), 256 - args->uniforms->ShadeDesaturate(), 256 - args->uniforms->ShadeDesaturate(), 256, 256 - args->uniforms->ShadeDesaturate(), 256 - args->uniforms->ShadeDesaturate(), 256 - args->uniforms->ShadeDesaturate());
shade_fade = _mm_set_epi16(args->uniforms->ShadeFadeAlpha(), args->uniforms->ShadeFadeRed(), args->uniforms->ShadeFadeGreen(), args->uniforms->ShadeFadeBlue(), args->uniforms->ShadeFadeAlpha(), args->uniforms->ShadeFadeRed(), args->uniforms->ShadeFadeGreen(), args->uniforms->ShadeFadeBlue());
shade_light = _mm_set_epi16(args->uniforms->ShadeLightAlpha(), args->uniforms->ShadeLightRed(), args->uniforms->ShadeLightGreen(), args->uniforms->ShadeLightBlue(), args->uniforms->ShadeLightAlpha(), args->uniforms->ShadeLightRed(), args->uniforms->ShadeLightGreen(), args->uniforms->ShadeLightBlue());
desaturate = args->uniforms->ShadeDesaturate();
}
else
{
inv_desaturate = _mm_setzero_si128();
shade_fade = _mm_setzero_si128();
shade_fade = _mm_setzero_si128();
shade_light = _mm_setzero_si128();
desaturate = 0;
}
if (mask0 == 0xffffffff && mask1 == 0xffffffff)
{
for (int y = 0; y < 8; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
__m128 mrcpW = _mm_set1_ps(1.0f / blockPosY.W);
__m128 worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosY.WorldX), mrcpW);
__m128i dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
mrcpW = _mm_set1_ps(1.0f / blockPosX.W);
worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosX.WorldX), mrcpW);
__m128i dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
__m128i dynlightstep = _mm_srai_epi16(_mm_sub_epi16(dynlightnext, dynlight), 3);
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, _mm_and_si128(dynlightstep, _mm_set_epi32(0xffff,0xffff,0,0))), _mm_set1_epi16(256)), _mm_setzero_si128());
dynlightstep = _mm_slli_epi16(dynlightstep, 1);
for (int ix = 0; ix < 4; ix++)
{
// Load bgcolor
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)(dest + ix * 2)), _mm_setzero_si128());
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2], ifgshade[2];
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = dest[ix * 2];
ifgcolor[0] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[0] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + ix * 2, destY + y);
posU += stepU;
posV += stepV;
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = dest[ix * 2 + 1];
ifgcolor[1] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[1] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + ix * 2 + 1, destY + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
int lightpos1 = lightpos >> 8;
lightpos += lightstep;
__m128i mlight = _mm_set_epi16(256, lightpos1, lightpos1, lightpos1, 256, lightpos0, lightpos0, lightpos0);
__m128i shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
__m128i inv_light = _mm_sub_epi16(_mm_set_epi16(0, 256, 256, 256, 0, 256, 256, 256), mlight);
shade_fade_lit = _mm_mullo_epi16(shade_fade, inv_light);
}
else
{
shade_fade_lit = _mm_setzero_si128();
}
// Shade and blend
__m128i fgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)ifgcolor), _mm_setzero_si128());
fgcolor = Shade32<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
__m128i outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor[0], ifgcolor[1], ifgshade[0], ifgshade[1], srcalpha, destalpha);
// Store result
_mm_storel_epi64((__m128i*)(dest + ix * 2), outcolor);
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, dynlightstep), _mm_set1_epi16(256)), _mm_setzero_si128());
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
}
else
{
// mask0 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
__m128 mrcpW = _mm_set1_ps(1.0f / blockPosY.W);
__m128 worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosY.WorldX), mrcpW);
__m128i dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
mrcpW = _mm_set1_ps(1.0f / blockPosX.W);
worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosX.WorldX), mrcpW);
__m128i dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
__m128i dynlightstep = _mm_srai_epi16(_mm_sub_epi16(dynlightnext, dynlight), 3);
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, _mm_and_si128(dynlightstep, _mm_set_epi32(0xffff, 0xffff, 0, 0))), _mm_set1_epi16(256)), _mm_setzero_si128());
dynlightstep = _mm_slli_epi16(dynlightstep, 1);
for (int x = 0; x < 4; x++)
{
// Load bgcolor
uint32_t desttmp[2];
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
{
if (mask0 & (1 << 31)) desttmp[0] = dest[x * 2];
if (mask0 & (1 << 30)) desttmp[1] = dest[x * 2 + 1];
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)desttmp), _mm_setzero_si128());
}
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2], ifgshade[2];
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = dest[x * 2];
ifgcolor[0] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[0] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x * 2, destY + y);
posU += stepU;
posV += stepV;
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = dest[x * 2 + 1];
ifgcolor[1] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[1] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x * 2 + 1, destY + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
int lightpos1 = lightpos >> 8;
lightpos += lightstep;
__m128i mlight = _mm_set_epi16(256, lightpos1, lightpos1, lightpos1, 256, lightpos0, lightpos0, lightpos0);
__m128i shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
__m128i inv_light = _mm_sub_epi16(_mm_set_epi16(0, 256, 256, 256, 0, 256, 256, 256), mlight);
shade_fade_lit = _mm_mullo_epi16(shade_fade, inv_light);
}
else
{
shade_fade_lit = _mm_setzero_si128();
}
// Shade and blend
__m128i fgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)ifgcolor), _mm_setzero_si128());
fgcolor = Shade32<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
__m128i outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor[0], ifgcolor[1], ifgshade[0], ifgshade[1], srcalpha, destalpha);
// Store result
_mm_storel_epi64((__m128i*)desttmp, outcolor);
if (mask0 & (1 << 31)) dest[x * 2] = desttmp[0];
if (mask0 & (1 << 30)) dest[x * 2 + 1] = desttmp[1];
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, dynlightstep), _mm_set1_epi16(256)), _mm_setzero_si128());
mask0 <<= 2;
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
// mask1 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
__m128 mrcpW = _mm_set1_ps(1.0f / blockPosY.W);
__m128 worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosY.WorldX), mrcpW);
__m128i dynlight = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
blockPosX.WorldX += gradientX.WorldX;
blockPosX.WorldY += gradientX.WorldY;
blockPosX.WorldZ += gradientX.WorldZ;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
mrcpW = _mm_set1_ps(1.0f / blockPosX.W);
worldpos = _mm_mul_ps(_mm_loadu_ps(&blockPosX.WorldX), mrcpW);
__m128i dynlightnext = CalcDynamicLight(lights, num_lights, worldpos, worldnormal, dynlightcolor);
__m128i dynlightstep = _mm_srai_epi16(_mm_sub_epi16(dynlightnext, dynlight), 3);
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, _mm_and_si128(dynlightstep, _mm_set_epi32(0xffff, 0xffff, 0, 0))), _mm_set1_epi16(256)), _mm_setzero_si128());
dynlightstep = _mm_slli_epi16(dynlightstep, 1);
for (int x = 0; x < 4; x++)
{
// Load bgcolor
uint32_t desttmp[2];
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
{
if (mask1 & (1 << 31)) desttmp[0] = dest[x * 2];
if (mask1 & (1 << 30)) desttmp[1] = dest[x * 2 + 1];
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)desttmp), _mm_setzero_si128());
}
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2], ifgshade[2];
if (SamplerT::Mode == (int)Samplers::FogBoundary && (mask1 & (1 << 31))) color = dest[x * 2];
ifgcolor[0] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[0] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x * 2, destY + 4 + y);
posU += stepU;
posV += stepV;
if (SamplerT::Mode == (int)Samplers::FogBoundary && (mask1 & (1 << 30))) color = dest[x * 2 + 1];
ifgcolor[1] = Sample32<SamplerT, FilterModeT>(posU, posV, texPixels, texWidth, texHeight, oneU, oneV, color, translation);
ifgshade[1] = SampleShade32<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x * 2 + 1, destY + 4 + y);
posU += stepU;
posV += stepV;
// Setup light
int lightpos0 = lightpos >> 8;
lightpos += lightstep;
int lightpos1 = lightpos >> 8;
lightpos += lightstep;
__m128i mlight = _mm_set_epi16(256, lightpos1, lightpos1, lightpos1, 256, lightpos0, lightpos0, lightpos0);
__m128i shade_fade_lit;
if (ShadeModeT::Mode == (int)ShadeMode::Advanced)
{
__m128i inv_light = _mm_sub_epi16(_mm_set_epi16(0, 256, 256, 256, 0, 256, 256, 256), mlight);
shade_fade_lit = _mm_mullo_epi16(shade_fade, inv_light);
}
else
{
shade_fade_lit = _mm_setzero_si128();
}
// Shade and blend
__m128i fgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)ifgcolor), _mm_setzero_si128());
fgcolor = Shade32<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light, dynlight);
__m128i outcolor = Blend32<BlendT>(fgcolor, bgcolor, ifgcolor[0], ifgcolor[1], ifgshade[0], ifgshade[1], srcalpha, destalpha);
// Store result
_mm_storel_epi64((__m128i*)desttmp, outcolor);
if (mask1 & (1 << 31)) dest[x * 2] = desttmp[0];
if (mask1 & (1 << 30)) dest[x * 2 + 1] = desttmp[1];
dynlight = _mm_max_epi16(_mm_min_epi16(_mm_add_epi16(dynlight, dynlightstep), _mm_set1_epi16(256)), _mm_setzero_si128());
mask1 <<= 2;
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
blockPosY.WorldX += gradientY.WorldX;
blockPosY.WorldY += gradientY.WorldY;
blockPosY.WorldZ += gradientY.WorldZ;
dest += pitch;
}
}
}
};
template<typename BlendT, typename SamplerT>
class RectScreenDrawer32
{

200
src/polyrenderer/drawers/poly_drawer8.h
View file

@ -225,206 +225,6 @@ namespace TriScreenDrawerModes
}
}
template<typename BlendT, typename SamplerT>
class TriScreenDrawer8
{
public:
static void Execute(int destX, int destY, uint32_t mask0, uint32_t mask1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
bool is_fixed_light = args->uniforms->FixedLight();
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
auto colormaps = args->uniforms->BaseColormap();
uint32_t srcalpha = args->uniforms->SrcAlpha();
uint32_t destalpha = args->uniforms->DestAlpha();
// Calculate gradients
const ShadedTriVertex &v1 = *args->v1;
ScreenTriangleStepVariables gradientX = args->gradientX;
ScreenTriangleStepVariables gradientY = args->gradientY;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = v1.w + gradientX.W * (destX - v1.x) + gradientY.W * (destY - v1.y);
blockPosY.U = v1.u * v1.w + gradientX.U * (destX - v1.x) + gradientY.U * (destY - v1.y);
blockPosY.V = v1.v * v1.w + gradientX.V * (destX - v1.x) + gradientY.V * (destY - v1.y);
gradientX.W *= 8.0f;
gradientX.U *= 8.0f;
gradientX.V *= 8.0f;
// Output
uint8_t * RESTRICT destOrg = args->dest;
int pitch = args->pitch;
uint8_t *dest = destOrg + destX + destY * pitch;
// Light
uint32_t light = args->uniforms->Light();
float shade = 2.0f - (light + 12.0f) / 128.0f;
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
light += light >> 7; // 255 -> 256
// Sampling stuff
uint32_t color = args->uniforms->Color();
const uint8_t * RESTRICT translation = args->uniforms->Translation();
const uint8_t * RESTRICT texPixels = args->uniforms->TexturePixels();
uint32_t texWidth = args->uniforms->TextureWidth();
uint32_t texHeight = args->uniforms->TextureHeight();
if (mask0 == 0xffffffff && mask1 == 0xffffffff)
{
for (int y = 0; y < 8; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
for (int ix = 0; ix < 8; ix++)
{
int lightshade = lightpos >> 8;
uint8_t bgcolor = dest[ix];
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = bgcolor;
uint8_t fgcolor = Sample8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, color, translation);
uint32_t fgshade = SampleShade8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + ix, destY + y);
if (SamplerT::Mode == (int)Samplers::Fuzz) lightshade = 256;
dest[ix] = ShadeAndBlend8<BlendT>(fgcolor, bgcolor, fgshade, lightshade, colormaps, srcalpha, destalpha);
posU += stepU;
posV += stepV;
lightpos += lightstep;
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
dest += pitch;
}
}
else
{
// mask0 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
for (int x = 0; x < 8; x++)
{
if (mask0 & (1 << 31))
{
int lightshade = lightpos >> 8;
uint8_t bgcolor = dest[x];
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = bgcolor;
uint8_t fgcolor = Sample8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, color, translation);
uint32_t fgshade = SampleShade8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x, destY + y);
if (SamplerT::Mode == (int)Samplers::Fuzz) lightshade = 256;
dest[x] = ShadeAndBlend8<BlendT>(fgcolor, bgcolor, fgshade, lightshade, colormaps, srcalpha, destalpha);
}
posU += stepU;
posV += stepV;
lightpos += lightstep;
mask0 <<= 1;
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
dest += pitch;
}
// mask1 loop:
for (int y = 0; y < 4; y++)
{
float rcpW = 0x01000000 / blockPosY.W;
int32_t posU = (int32_t)(blockPosY.U * rcpW);
int32_t posV = (int32_t)(blockPosY.V * rcpW);
fixed_t lightpos = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
ScreenTriangleStepVariables blockPosX = blockPosY;
blockPosX.W += gradientX.W;
blockPosX.U += gradientX.U;
blockPosX.V += gradientX.V;
rcpW = 0x01000000 / blockPosX.W;
int32_t nextU = (int32_t)(blockPosX.U * rcpW);
int32_t nextV = (int32_t)(blockPosX.V * rcpW);
int32_t stepU = (nextU - posU) / 8;
int32_t stepV = (nextV - posV) / 8;
fixed_t lightnext = FRACUNIT - (fixed_t)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
for (int x = 0; x < 8; x++)
{
if (mask1 & (1 << 31))
{
int lightshade = lightpos >> 8;
uint8_t bgcolor = dest[x];
if (SamplerT::Mode == (int)Samplers::FogBoundary) color = bgcolor;
uint8_t fgcolor = Sample8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, color, translation);
uint32_t fgshade = SampleShade8<SamplerT>(posU, posV, texPixels, texWidth, texHeight, destX + x, destY + 4 + y);
if (SamplerT::Mode == (int)Samplers::Fuzz) lightshade = 256;
dest[x] = ShadeAndBlend8<BlendT>(fgcolor, bgcolor, fgshade, lightshade, colormaps, srcalpha, destalpha);
}
posU += stepU;
posV += stepV;
lightpos += lightstep;
mask1 <<= 1;
}
blockPosY.W += gradientY.W;
blockPosY.U += gradientY.U;
blockPosY.V += gradientY.V;
dest += pitch;
}
}
}
};
template<typename BlendT, typename SamplerT>
class RectScreenDrawer8
{

8
src/polyrenderer/drawers/poly_triangle.cpp
View file

@ -39,12 +39,19 @@
#include "screen_triangle.h"
#include "x86.h"
static bool isBgraRenderTarget = false;
void PolyTriangleDrawer::ClearBuffers(DCanvas *canvas)
{
PolyStencilBuffer::Instance()->Clear(canvas->GetWidth(), canvas->GetHeight(), 0);
PolyZBuffer::Instance()->Resize(canvas->GetPitch(), canvas->GetHeight());
}
bool PolyTriangleDrawer::IsBgra()
{
return isBgraRenderTarget;
}
void PolyTriangleDrawer::SetViewport(const DrawerCommandQueuePtr &queue, int x, int y, int width, int height, DCanvas *canvas, bool span_drawers)
{
uint8_t *dest = (uint8_t*)canvas->GetBuffer();
@ -52,6 +59,7 @@ void PolyTriangleDrawer::SetViewport(const DrawerCommandQueuePtr &queue, int x,
int dest_height = canvas->GetHeight();
int dest_pitch = canvas->GetPitch();
bool dest_bgra = canvas->IsBgra();
isBgraRenderTarget = dest_bgra;
int offsetx = clamp(x, 0, dest_width);
int offsety = clamp(y, 0, dest_height);

2
src/polyrenderer/drawers/poly_triangle.h
View file

@ -38,6 +38,8 @@ public:
static void SetTwoSided(const DrawerCommandQueuePtr &queue, bool twosided);
static void SetWeaponScene(const DrawerCommandQueuePtr &queue, bool enable);
static void SetTransform(const DrawerCommandQueuePtr &queue, const Mat4f *objectToClip);
static bool IsBgra();
};
class PolyTriangleThreadData

795
src/polyrenderer/drawers/screen_triangle.cpp
View file

@ -307,7 +307,7 @@ void TriangleBlock::RenderBlock(int x0, int y0, int x1, int y1)
bool writeDepth = args->uniforms->WriteDepth();
int bmode = (int)args->uniforms->BlendMode();
auto drawFunc = args->destBgra ? ScreenTriangle::TriDrawers32[bmode] : ScreenTriangle::TriDrawers8[bmode];
auto drawFunc = args->destBgra ? ScreenTriangle::SpanDrawers32[bmode] : ScreenTriangle::SpanDrawers8[bmode];
// Loop through blocks
for (int y = start_miny; y < y1; y += q * num_cores)
@ -345,7 +345,66 @@ void TriangleBlock::RenderBlock(int x0, int y0, int x1, int y1)
}
if (writeColor)
drawFunc(X, Y, Mask0, Mask1, args);
{
if (Mask0 == 0xffffffff)
{
drawFunc(Y, X, X + 8, args);
drawFunc(Y + 1, X, X + 8, args);
drawFunc(Y + 2, X, X + 8, args);
drawFunc(Y + 3, X, X + 8, args);
}
else if (Mask0 != 0)
{
uint32_t mask = Mask0;
for (int j = 0; j < 4; j++)
{
int start = 0;
int i;
for (i = 0; i < 8; i++)
{
if (!(mask & 0x80000000))
{
if (i > start)
drawFunc(Y + j, X + start, X + i, args);
start = i + 1;
}
mask <<= 1;
}
if (i > start)
drawFunc(Y + j, X + start, X + i, args);
}
}
if (Mask1 == 0xffffffff)
{
drawFunc(Y + 4, X, X + 8, args);
drawFunc(Y + 5, X, X + 8, args);
drawFunc(Y + 6, X, X + 8, args);
drawFunc(Y + 7, X, X + 8, args);
}
else if (Mask1 != 0)
{
uint32_t mask = Mask1;
for (int j = 4; j < 8; j++)
{
int start = 0;
int i;
for (i = 0; i < 8; i++)
{
if (!(mask & 0x80000000))
{
if (i > start)
drawFunc(Y + j, X + start, X + i, args);
start = i + 1;
}
mask <<= 1;
}
if (i > start)
drawFunc(Y + j, X + start, X + i, args);
}
}
}
if (writeStencil)
StencilWrite();
if (writeDepth)
@ -1249,210 +1308,102 @@ void ScreenTriangle::DrawSWRender(const TriDrawTriangleArgs *args, PolyTriangleT
}
}
template<typename ModeT>
void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
template<typename ModeT, typename OptT>
void DrawSpanOpt32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
float v1X = args->v1->x;
float v1Y = args->v1->y;
float v1W = args->v1->w;
float v1U = args->v1->u * v1W;
float v1V = args->v1->v * v1W;
float stepXW = args->gradientX.W;
float stepXU = args->gradientX.U;
float stepXV = args->gradientX.V;
float startX = x0 + (0.5f - v1X);
float startY = y + (0.5f - v1Y);
float posXW = v1W + stepXW * startX + args->gradientY.W * startY;
float posXU = v1U + stepXU * startX + args->gradientY.U * startY;
float posXV = v1V + stepXV * startX + args->gradientY.V * startY;
float v1X, v1Y, v1W, v1U, v1V, v1WorldX, v1WorldY, v1WorldZ;
float startX, startY;
float stepXW, stepXU, stepXV, stepWorldX, stepWorldY, stepWorldZ;
float posXW, posXU, posXV, posWorldX, posWorldY, posWorldZ;
const uint32_t *texPixels = (const uint32_t*)args->uniforms->TexturePixels();
const uint32_t *translation = (const uint32_t*)args->uniforms->Translation();
int texWidth = args->uniforms->TextureWidth();
int texHeight = args->uniforms->TextureHeight();
PolyLight *lights;
int num_lights;
float worldnormalX, worldnormalY, worldnormalZ;
uint32_t dynlightcolor;
const uint32_t *texPixels, *translation;
int texWidth, texHeight;
uint32_t fillcolor;
int alpha;
uint32_t light;
fixed_t shade, lightpos, lightstep;
uint32_t shade_fade_r, shade_fade_g, shade_fade_b, shade_light_r, shade_light_g, shade_light_b, desaturate, inv_desaturate;
int fillcolor = args->uniforms->Color();
int alpha = args->uniforms->SrcAlpha();
v1X = args->v1->x;
v1Y = args->v1->y;
v1W = args->v1->w;
v1U = args->v1->u * v1W;
v1V = args->v1->v * v1W;
startX = x0 + (0.5f - v1X);
startY = y + (0.5f - v1Y);
stepXW = args->gradientX.W;
stepXU = args->gradientX.U;
stepXV = args->gradientX.V;
posXW = v1W + stepXW * startX + args->gradientY.W * startY;
posXU = v1U + stepXU * startX + args->gradientY.U * startY;
posXV = v1V + stepXV * startX + args->gradientY.V * startY;
bool is_fixed_light = args->uniforms->FixedLight();
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
uint32_t light = args->uniforms->Light();
float shade = 2.0f - (light + 12.0f) / 128.0f;
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
texPixels = (const uint32_t*)args->uniforms->TexturePixels();
translation = (const uint32_t*)args->uniforms->Translation();
texWidth = args->uniforms->TextureWidth();
texHeight = args->uniforms->TextureHeight();
fillcolor = args->uniforms->Color();
alpha = args->uniforms->Alpha();
light = args->uniforms->Light();
if (OptT::Flags & SWOPT_FixedLight)
{
light += light >> 7; // 255 -> 256
}
else
{
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
shade = (fixed_t)((2.0f - (light + 12.0f) / 128.0f) * (float)FRACUNIT);
lightpos = (fixed_t)(globVis * posXW * (float)FRACUNIT);
lightstep = (fixed_t)(globVis * stepXW * (float)FRACUNIT);
}
if (OptT::Flags & SWOPT_DynLights)
{
v1WorldX = args->v1->worldX * v1W;
v1WorldY = args->v1->worldY * v1W;
v1WorldZ = args->v1->worldZ * v1W;
stepWorldX = args->gradientX.WorldX;
stepWorldY = args->gradientX.WorldY;
stepWorldZ = args->gradientX.WorldZ;
posWorldX = v1WorldX + stepWorldX * startX + args->gradientY.WorldX * startY;
posWorldY = v1WorldY + stepWorldY * startX + args->gradientY.WorldY * startY;
posWorldZ = v1WorldZ + stepWorldZ * startX + args->gradientY.WorldZ * startY;
lights = args->uniforms->Lights();
num_lights = args->uniforms->NumLights();
worldnormalX = args->uniforms->Normal().X;
worldnormalY = args->uniforms->Normal().Y;
worldnormalZ = args->uniforms->Normal().Z;
dynlightcolor = args->uniforms->DynLightColor();
}
if (OptT::Flags & SWOPT_ColoredFog)
{
shade_fade_r = args->uniforms->ShadeFadeRed();
shade_fade_g = args->uniforms->ShadeFadeGreen();
shade_fade_b = args->uniforms->ShadeFadeBlue();
shade_light_r = args->uniforms->ShadeLightRed();
shade_light_g = args->uniforms->ShadeLightGreen();
shade_light_b = args->uniforms->ShadeLightBlue();
desaturate = args->uniforms->ShadeDesaturate();
inv_desaturate = 256 - desaturate;
}
uint32_t *dest = (uint32_t*)args->dest;
uint32_t *destLine = dest + args->pitch * y;
int x = x0;
#ifndef NO_SSE
__m128i mfillcolor = _mm_set1_epi32(fillcolor);
__m128i mcapcolor = _mm_unpacklo_epi8(mfillcolor, _mm_setzero_si128());
__m128i malpha = _mm_set1_epi32(alpha);
int sseEnd = x0 + ((x1 - x0) & ~3);
while (x < sseEnd)
{
__m128i fg;
if (ModeT::SWFlags & SWSTYLEF_Fill)
{
fg = mfillcolor;
}
else if (ModeT::SWFlags & SWSTYLEF_FogBoundary)
{
fg = _mm_loadl_epi64((const __m128i*)(destLine + x));
}
else
{
float rcpW0 = 0x01000000 / posXW;
float rcpW1 = 0x01000000 / (posXW + stepXW);
int32_t u0 = (int32_t)(posXU * rcpW0);
int32_t u1 = (int32_t)((posXU + stepXU) * rcpW1);
int32_t v0 = (int32_t)(posXV * rcpW0);
int32_t v1 = (int32_t)((posXV + stepXV) * rcpW1);
uint32_t texelX0 = ((((uint32_t)u0 << 8) >> 16) * texWidth) >> 16;
uint32_t texelX1 = ((((uint32_t)u1 << 8) >> 16) * texWidth) >> 16;
uint32_t texelY0 = ((((uint32_t)v0 << 8) >> 16) * texHeight) >> 16;
uint32_t texelY1 = ((((uint32_t)v1 << 8) >> 16) * texHeight) >> 16;
if (ModeT::SWFlags & SWSTYLEF_Translated)
{
uint32_t fg0 = translation[((const uint8_t*)texPixels)[texelX0 * texHeight + texelY0]];
uint32_t fg1 = translation[((const uint8_t*)texPixels)[texelX1 * texHeight + texelY1]];
fg = _mm_setr_epi32(fg0, fg1, 0, 0);
}
else
{
uint32_t fg0 = texPixels[texelX0 * texHeight + texelY0];
uint32_t fg1 = texPixels[texelX1 * texHeight + texelY1];
fg = _mm_setr_epi32(fg0, fg1, 0, 0);
}
}
if (ModeT::SWFlags & SWSTYLEF_Skycap)
{
float rcpW0 = 0x01000000 / posXW;
float rcpW1 = 0x01000000 / (posXW + stepXW);
int32_t v0 = (int32_t)(posXV * rcpW0);
int32_t v1 = (int32_t)((posXV + stepXV) * rcpW1);
int start_fade = 2; // How fast it should fade out
__m128i v = _mm_setr_epi32(v0, v0, v1, v1);
__m128i alpha_top = _mm_min_epi16(_mm_max_epi16(_mm_srai_epi32(v, 16 - start_fade), _mm_setzero_si128()), _mm_set1_epi16(256));
__m128i alpha_bottom = _mm_min_epi16(_mm_max_epi16(_mm_srai_epi32(_mm_sub_epi32(_mm_set1_epi32(2 << 24), v), 16 - start_fade), _mm_setzero_si128()), _mm_set1_epi16(256));
__m128i a = _mm_min_epi16(alpha_top, alpha_bottom);
a = _mm_shufflelo_epi16(_mm_shufflehi_epi16(a, _MM_SHUFFLE(0, 0, 0, 0)), _MM_SHUFFLE(0, 0, 0, 0));
__m128i inv_a = _mm_sub_epi32(_mm_set1_epi32(256), a);
fg = _mm_unpacklo_epi8(fg, _mm_setzero_si128());
__m128i c = _mm_srli_epi16(_mm_add_epi16(_mm_add_epi16(_mm_mullo_epi16(fg, a), _mm_mullo_epi16(mcapcolor, inv_a)), _mm_set1_epi16(127)), 8);
_mm_storel_epi64((__m128i*)(destLine + x), _mm_packus_epi16(c, c));
}
else
{
if ((ModeT::Flags & STYLEF_ColorIsFixed) && !(ModeT::SWFlags & SWSTYLEF_Fill))
{
__m128i rgbmask = _mm_set1_epi32(0x00ffffff);
if (ModeT::Flags & STYLEF_RedIsAlpha)
fg = _mm_or_si128(_mm_andnot_si128(rgbmask, _mm_slli_epi32(fg, 8)), _mm_and_si128(rgbmask, mfillcolor));
else
fg = _mm_or_si128(_mm_andnot_si128(rgbmask, fg), _mm_and_si128(rgbmask, mfillcolor));
}
if (!(ModeT::Flags & STYLEF_Alpha1))
{
__m128i a = _mm_srli_epi32(fg, 24);
a = _mm_srli_epi32(_mm_mullo_epi16(a, malpha), 8);
fg = _mm_or_si128(_mm_and_si128(fg, _mm_set1_epi32(0x00ffffff)), _mm_slli_epi32(a, 24));
}
fg = _mm_unpacklo_epi8(fg, _mm_setzero_si128());
fixed_t lightpos0 = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * posXW), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
fixed_t lightpos1 = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * (posXW + stepXW)), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos0 = (lightpos0 & lightmask) | ((light << 8) & ~lightmask);
lightpos1 = (lightpos1 & lightmask) | ((light << 8) & ~lightmask);
int lightshade0 = lightpos0 >> 8;
int lightshade1 = lightpos1 >> 8;
__m128i shadedfg = _mm_srli_epi16(_mm_mullo_epi16(fg, _mm_setr_epi16(lightshade0, lightshade0, lightshade0, 256, lightshade1, lightshade1, lightshade1, 256)), 8);
__m128i out;
if (ModeT::BlendSrc == STYLEALPHA_One && ModeT::BlendDest == STYLEALPHA_Zero)
{
out = shadedfg;
}
else if (ModeT::BlendSrc == STYLEALPHA_One && ModeT::BlendDest == STYLEALPHA_One)
{
__m128i dest = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i*)(destLine + x)), _mm_setzero_si128());
if (ModeT::BlendOp == STYLEOP_Add)
{
out = _mm_add_epi16(dest, shadedfg);
}
else if (ModeT::BlendOp == STYLEOP_RevSub)
{
out = _mm_sub_epi16(dest, shadedfg);
}
else //if (ModeT::BlendOp == STYLEOP_Sub)
{
out = _mm_sub_epi16(shadedfg, dest);
}
}
else if (ModeT::SWFlags & SWSTYLEF_SrcColorOneMinusSrcColor)
{
__m128i dest = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i*)(destLine + x)), _mm_setzero_si128());
__m128i sfactor = _mm_add_epi16(shadedfg, _mm_srli_epi16(shadedfg, 7));
__m128i dfactor = _mm_sub_epi16(_mm_set1_epi16(256), sfactor);
out = _mm_srli_epi16(_mm_add_epi16(_mm_add_epi16(_mm_mullo_epi16(dest, dfactor), _mm_mullo_epi16(shadedfg, sfactor)), _mm_set1_epi16(127)), 8);
}
else
{
__m128i dest = _mm_unpacklo_epi8(_mm_loadl_epi64((const __m128i*)(destLine + x)), _mm_setzero_si128());
__m128i sfactor = _mm_shufflehi_epi16(_mm_shufflelo_epi16(shadedfg, _MM_SHUFFLE(3, 3, 3, 3)), _MM_SHUFFLE(3, 3, 3, 3));
sfactor = _mm_add_epi16(sfactor, _mm_srli_epi16(sfactor, 7)); // 255 -> 256
__m128i dfactor = _mm_sub_epi16(_mm_set1_epi16(256), sfactor);
__m128i src = _mm_mullo_epi16(shadedfg, sfactor);
if (ModeT::BlendDest == STYLEALPHA_One)
{
dest = _mm_slli_epi16(dest, 8);
}
else
{
__m128i dfactor = _mm_sub_epi16(_mm_set1_epi16(256), sfactor);
dest = _mm_mullo_epi16(dest, dfactor);
}
if (ModeT::BlendOp == STYLEOP_Add)
{
out = _mm_srli_epi16(_mm_add_epi16(_mm_add_epi16(dest, src), _mm_set1_epi16(127)), 8);
}
else if (ModeT::BlendOp == STYLEOP_RevSub)
{
out = _mm_srli_epi16(_mm_add_epi16(_mm_sub_epi16(dest, src), _mm_set1_epi16(127)), 8);
}
else //if (ModeT::BlendOp == STYLEOP_Sub)
{
out = _mm_srli_epi16(_mm_add_epi16(_mm_sub_epi16(src, dest), _mm_set1_epi16(127)), 8);
}
}
_mm_storel_epi64((__m128i*)(destLine + x), _mm_or_si128(_mm_packus_epi16(out, out), _mm_set1_epi32(0xff000000)));
}
posXW += stepXW + stepXW;
posXU += stepXU + stepXU;
posXV += stepXV + stepXV;
x += 2;
}
#endif
while (x < x1)
{
uint32_t fg;
uint32_t fg = 0;
if (ModeT::SWFlags & SWSTYLEF_Fill)
{
@ -1462,7 +1413,7 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
fg = destLine[x];
}
else
else if (ModeT::BlendOp != STYLEOP_Fuzz)
{
float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW);
@ -1474,13 +1425,48 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
fg = translation[((const uint8_t*)texPixels)[texelX * texHeight + texelY]];
}
else if (ModeT::Flags & STYLEF_RedIsAlpha)
{
fg = ((const uint8_t*)texPixels)[texelX * texHeight + texelY];
}
else
{
fg = texPixels[texelX * texHeight + texelY];
}
}
if (ModeT::SWFlags & SWSTYLEF_Skycap)
if (ModeT::BlendOp == STYLEOP_Fuzz)
{
using namespace swrenderer;
float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW);
int32_t v = (int32_t)(posXV * rcpW);
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
unsigned int sampleshadeout = APART(texPixels[texelX * texHeight + texelY]);
sampleshadeout += sampleshadeout >> 7; // 255 -> 256
fixed_t fuzzscale = (200 << FRACBITS) / viewheight;
int scaled_x = (x * fuzzscale) >> FRACBITS;
int fuzz_x = fuzz_random_x_offset[scaled_x % FUZZ_RANDOM_X_SIZE] + fuzzpos;
fixed_t fuzzcount = FUZZTABLE << FRACBITS;
fixed_t fuzz = ((fuzz_x << FRACBITS) + y * fuzzscale) % fuzzcount;
unsigned int alpha = fuzzoffset[fuzz >> FRACBITS];
sampleshadeout = (sampleshadeout * alpha) >> 5;
uint32_t a = 256 - sampleshadeout;
uint32_t dest = destLine[x];
uint32_t out_r = (RPART(dest) * a) >> 8;
uint32_t out_g = (GPART(dest) * a) >> 8;
uint32_t out_b = (BPART(dest) * a) >> 8;
destLine[x] = MAKEARGB(255, out_r, out_g, out_b);
}
else if (ModeT::SWFlags & SWSTYLEF_Skycap)
{
float rcpW = 0x01000000 / posXW;
int32_t v = (int32_t)(posXV * rcpW);
@ -1516,7 +1502,7 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
if ((ModeT::Flags & STYLEF_ColorIsFixed) && !(ModeT::SWFlags & SWSTYLEF_Fill))
{
if (ModeT::Flags & STYLEF_RedIsAlpha)
fg = ((fg << 8) & 0xff000000) | (fillcolor & 0x00ffffff);
fg = (fg << 24) | (fillcolor & 0x00ffffff);
else
fg = (fg & 0xff000000) | (fillcolor & 0x00ffffff);
}
@ -1528,12 +1514,105 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
fgalpha = (fgalpha * alpha) >> 8;
}
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * posXW), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
int lightshade = lightpos >> 8;
uint32_t shadedfg_r = (RPART(fg) * lightshade) >> 8;
uint32_t shadedfg_g = (GPART(fg) * lightshade) >> 8;
uint32_t shadedfg_b = (BPART(fg) * lightshade) >> 8;
int lightshade;
if (OptT::Flags & SWOPT_FixedLight)
{
lightshade = light;
}
else
{
fixed_t maxvis = 24 * FRACUNIT / 32;
fixed_t maxlight = 31 * FRACUNIT / 32;
lightshade = (FRACUNIT - clamp<fixed_t>(shade - MIN(maxvis, lightpos), 0, maxlight)) >> 8;
}
uint32_t lit_r = 0, lit_g = 0, lit_b = 0;
if (OptT::Flags & SWOPT_DynLights)
{
lit_r = RPART(dynlightcolor);
lit_g = GPART(dynlightcolor);
lit_b = BPART(dynlightcolor);
float rcp_posXW = 1.0f / posXW;
float worldposX = posWorldX * rcp_posXW;
float worldposY = posWorldY * rcp_posXW;
float worldposZ = posWorldZ * rcp_posXW;
for (int i = 0; i < num_lights; i++)
{
float lightposX = lights[i].x;
float lightposY = lights[i].y;
float lightposZ = lights[i].z;
float light_radius = lights[i].radius;
uint32_t light_color = lights[i].color;
bool is_attenuated = light_radius < 0.0f;
if (is_attenuated)
light_radius = -light_radius;
// L = light-pos
// dist = sqrt(dot(L, L))
// distance_attenuation = 1 - MIN(dist * (1/radius), 1)
float Lx = lightposX - worldposX;
float Ly = lightposY - worldposY;
float Lz = lightposZ - worldposZ;
float dist2 = Lx * Lx + Ly * Ly + Lz * Lz;
#ifdef NO_SSE
//float rcp_dist = 1.0f / sqrt(dist2);
float rcp_dist = 1.0f / (dist2 * 0.01f);
#else
float rcp_dist = _mm_cvtss_f32(_mm_rsqrt_ss(_mm_set_ss(dist2)));
#endif
float dist = dist2 * rcp_dist;
float distance_attenuation = 256.0f - MIN(dist * light_radius, 256.0f);
// The simple light type
float simple_attenuation = distance_attenuation;
// The point light type
// diffuse = max(dot(N,normalize(L)),0) * attenuation
Lx *= rcp_dist;
Ly *= rcp_dist;
Lz *= rcp_dist;
float dotNL = worldnormalX * Lx + worldnormalY * Ly + worldnormalZ * Lz;
float point_attenuation = MAX(dotNL, 0.0f) * distance_attenuation;
uint32_t attenuation = (uint32_t)(is_attenuated ? (int32_t)point_attenuation : (int32_t)simple_attenuation);
lit_r += (RPART(light_color) * attenuation) >> 8;
lit_g += (GPART(light_color) * attenuation) >> 8;
lit_b += (BPART(light_color) * attenuation) >> 8;
}
}
uint32_t shadedfg_r, shadedfg_g, shadedfg_b;
if (OptT::Flags & SWOPT_ColoredFog)
{
uint32_t fg_r = RPART(fg);
uint32_t fg_g = GPART(fg);
uint32_t fg_b = BPART(fg);
uint32_t intensity = ((fg_r * 77 + fg_g * 143 + fg_b * 37) >> 8) * desaturate;
shadedfg_r = (((shade_fade_r + ((fg_r * inv_desaturate + intensity) >> 8) * lightshade) >> 8) * shade_light_r) >> 8;
shadedfg_g = (((shade_fade_g + ((fg_g * inv_desaturate + intensity) >> 8) * lightshade) >> 8) * shade_light_g) >> 8;
shadedfg_b = (((shade_fade_b + ((fg_b * inv_desaturate + intensity) >> 8) * lightshade) >> 8) * shade_light_b) >> 8;
lit_r = MIN(lit_r, (uint32_t)256);
lit_g = MIN(lit_g, (uint32_t)256);
lit_b = MIN(lit_b, (uint32_t)256);
shadedfg_r = MIN(shadedfg_r + ((fg_r * lit_r) >> 8), (uint32_t)255);
shadedfg_g = MIN(shadedfg_g + ((fg_g * lit_g) >> 8), (uint32_t)255);
shadedfg_b = MIN(shadedfg_b + ((fg_b * lit_b) >> 8), (uint32_t)255);
}
else
{
lit_r = MIN(lightshade + lit_r, (uint32_t)256);
lit_g = MIN(lightshade + lit_g, (uint32_t)256);
lit_b = MIN(lightshade + lit_b, (uint32_t)256);
shadedfg_r = (RPART(fg) * lit_r) >> 8;
shadedfg_g = (GPART(fg) * lit_g) >> 8;
shadedfg_b = (BPART(fg) * lit_b) >> 8;
}
if (ModeT::BlendSrc == STYLEALPHA_One && ModeT::BlendDest == STYLEALPHA_Zero)
{
@ -1582,16 +1661,15 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
destLine[x] = MAKEARGB(255, out_r, out_g, out_b);
}
else if (fgalpha == 255)
else if (ModeT::BlendSrc == STYLEALPHA_Src && ModeT::BlendDest == STYLEALPHA_InvSrc && fgalpha == 255)
{
destLine[x] = MAKEARGB(255, shadedfg_r, shadedfg_g, shadedfg_b);
}
else if (fgalpha != 0)
else if (ModeT::BlendSrc != STYLEALPHA_Src || ModeT::BlendDest != STYLEALPHA_InvSrc || fgalpha != 0)
{
uint32_t dest = destLine[x];
uint32_t sfactor = fgalpha; sfactor += sfactor >> 7; // 255 -> 256
uint32_t dfactor = 256 - sfactor;
uint32_t src_r = shadedfg_r * sfactor;
uint32_t src_g = shadedfg_g * sfactor;
uint32_t src_b = shadedfg_b * sfactor;
@ -1648,49 +1726,139 @@ void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
posXW += stepXW;
posXU += stepXU;
posXV += stepXV;
if (OptT::Flags & SWOPT_DynLights)
{
posWorldX += stepWorldX;
posWorldY += stepWorldY;
posWorldZ += stepWorldZ;
}
if (!(OptT::Flags & SWOPT_FixedLight))
lightpos += lightstep;
x++;
}
}
template<typename ModeT>
void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
void DrawSpan32(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
float v1X = args->v1->x;
float v1Y = args->v1->y;
float v1W = args->v1->w;
float v1U = args->v1->u * v1W;
float v1V = args->v1->v * v1W;
float stepXW = args->gradientX.W;
float stepXU = args->gradientX.U;
float stepXV = args->gradientX.V;
float startX = x0 + (0.5f - v1X);
float startY = y + (0.5f - v1Y);
float posXW = v1W + stepXW * startX + args->gradientY.W * startY;
float posXU = v1U + stepXU * startX + args->gradientY.U * startY;
float posXV = v1V + stepXV * startX + args->gradientY.V * startY;
if (args->uniforms->NumLights() == 0)
{
if (!args->uniforms->FixedLight())
{
if (args->uniforms->SimpleShade())
DrawSpanOpt32<ModeT, DrawerOpt>(y, x0, x1, args);
else
DrawSpanOpt32<ModeT, DrawerOptC>(y, x0, x1, args);
}
else
{
if (args->uniforms->SimpleShade())
DrawSpanOpt32<ModeT, DrawerOptF>(y, x0, x1, args);
else
DrawSpanOpt32<ModeT, DrawerOptCF>(y, x0, x1, args);
}
}
else
{
if (!args->uniforms->FixedLight())
{
if (args->uniforms->SimpleShade())
DrawSpanOpt32<ModeT, DrawerOptL>(y, x0, x1, args);
else
DrawSpanOpt32<ModeT, DrawerOptLC>(y, x0, x1, args);
}
else
{
if (args->uniforms->SimpleShade())
DrawSpanOpt32<ModeT, DrawerOptLF>(y, x0, x1, args);
else
DrawSpanOpt32<ModeT, DrawerOptLCF>(y, x0, x1, args);
}
}
}
auto colormaps = args->uniforms->BaseColormap();
template<typename ModeT, typename OptT>
void DrawSpanOpt8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
const uint8_t *texPixels = args->uniforms->TexturePixels();
const uint8_t *translation = args->uniforms->Translation();
int texWidth = args->uniforms->TextureWidth();
int texHeight = args->uniforms->TextureHeight();
float v1X, v1Y, v1W, v1U, v1V, v1WorldX, v1WorldY, v1WorldZ;
float startX, startY;
float stepXW, stepXU, stepXV, stepWorldX, stepWorldY, stepWorldZ;
float posXW, posXU, posXV, posWorldX, posWorldY, posWorldZ;
int fillcolor = args->uniforms->Color();
int alpha = args->uniforms->SrcAlpha();
PolyLight *lights;
int num_lights;
float worldnormalX, worldnormalY, worldnormalZ;
uint32_t dynlightcolor;
const uint8_t *colormaps, *texPixels, *translation;
int texWidth, texHeight;
uint32_t fillcolor, capcolor;
int alpha;
uint32_t light;
fixed_t shade, lightpos, lightstep;
v1X = args->v1->x;
v1Y = args->v1->y;
v1W = args->v1->w;
v1U = args->v1->u * v1W;
v1V = args->v1->v * v1W;
startX = x0 + (0.5f - v1X);
startY = y + (0.5f - v1Y);
stepXW = args->gradientX.W;
stepXU = args->gradientX.U;
stepXV = args->gradientX.V;
posXW = v1W + stepXW * startX + args->gradientY.W * startY;
posXU = v1U + stepXU * startX + args->gradientY.U * startY;
posXV = v1V + stepXV * startX + args->gradientY.V * startY;
texPixels = args->uniforms->TexturePixels();
translation = args->uniforms->Translation();
texWidth = args->uniforms->TextureWidth();
texHeight = args->uniforms->TextureHeight();
fillcolor = args->uniforms->Color();
alpha = args->uniforms->Alpha();
colormaps = args->uniforms->BaseColormap();
light = args->uniforms->Light();
uint32_t capcolor = fillcolor;
if (ModeT::SWFlags & SWSTYLEF_Skycap)
capcolor = GPalette.BaseColors[capcolor].d;
capcolor = GPalette.BaseColors[fillcolor].d;
bool is_fixed_light = args->uniforms->FixedLight();
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
uint32_t light = args->uniforms->Light();
float shade = 2.0f - (light + 12.0f) / 128.0f;
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
if (OptT::Flags & SWOPT_FixedLight)
{
light += light >> 7; // 255 -> 256
light = ((256 - light) * NUMCOLORMAPS) & 0xffffff00;
}
else
{
float globVis = args->uniforms->GlobVis() * (1.0f / 32.0f);
shade = (fixed_t)((2.0f - (light + 12.0f) / 128.0f) * (float)FRACUNIT);
lightpos = (fixed_t)(globVis * posXW * (float)FRACUNIT);
lightstep = (fixed_t)(globVis * stepXW * (float)FRACUNIT);
}
if (OptT::Flags & SWOPT_DynLights)
{
v1WorldX = args->v1->worldX * v1W;
v1WorldY = args->v1->worldY * v1W;
v1WorldZ = args->v1->worldZ * v1W;
stepWorldX = args->gradientX.WorldX;
stepWorldY = args->gradientX.WorldY;
stepWorldZ = args->gradientX.WorldZ;
posWorldX = v1WorldX + stepWorldX * startX + args->gradientY.WorldX * startY;
posWorldY = v1WorldY + stepWorldY * startX + args->gradientY.WorldY * startY;
posWorldZ = v1WorldZ + stepWorldZ * startX + args->gradientY.WorldZ * startY;
lights = args->uniforms->Lights();
num_lights = args->uniforms->NumLights();
worldnormalX = args->uniforms->Normal().X;
worldnormalY = args->uniforms->Normal().Y;
worldnormalZ = args->uniforms->Normal().Z;
dynlightcolor = args->uniforms->DynLightColor();
}
uint8_t *dest = (uint8_t*)args->dest;
uint8_t *destLine = dest + args->pitch * y;
@ -1698,7 +1866,7 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
int x = x0;
while (x < x1)
{
int fg;
int fg = 0;
int fgalpha = 255;
if (ModeT::SWFlags & SWSTYLEF_Fill)
@ -1709,7 +1877,7 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
fg = destLine[x];
}
else
else if (ModeT::BlendOp != STYLEOP_Fuzz)
{
float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW);
@ -1724,7 +1892,37 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
fgalpha = (fg != 0) ? 255 : 0;
}
if (ModeT::SWFlags & SWSTYLEF_Skycap)
if (ModeT::BlendOp == STYLEOP_Fuzz)
{
using namespace swrenderer;
float rcpW = 0x01000000 / posXW;
int32_t u = (int32_t)(posXU * rcpW);
int32_t v = (int32_t)(posXV * rcpW);
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
unsigned int sampleshadeout = (texPixels[texelX * texHeight + texelY] != 0) ? 256 : 0;
fixed_t fuzzscale = (200 << FRACBITS) / viewheight;
int scaled_x = (x * fuzzscale) >> FRACBITS;
int fuzz_x = fuzz_random_x_offset[scaled_x % FUZZ_RANDOM_X_SIZE] + fuzzpos;
fixed_t fuzzcount = FUZZTABLE << FRACBITS;
fixed_t fuzz = ((fuzz_x << FRACBITS) + y * fuzzscale) % fuzzcount;
unsigned int alpha = fuzzoffset[fuzz >> FRACBITS];
sampleshadeout = (sampleshadeout * alpha) >> 5;
uint32_t a = 256 - sampleshadeout;
uint32_t dest = GPalette.BaseColors[destLine[x]].d;
uint32_t r = (RPART(dest) * a) >> 8;
uint32_t g = (GPART(dest) * a) >> 8;
uint32_t b = (BPART(dest) * a) >> 8;
destLine[x] = RGB256k.All[((r >> 2) << 12) | ((g >> 2) << 6) | (b >> 2)];
}
else if (ModeT::SWFlags & SWSTYLEF_Skycap)
{
float rcpW = 0x01000000 / posXW;
int32_t v = (int32_t)(posXV * rcpW);
@ -1771,11 +1969,95 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
fgalpha = (fgalpha * alpha) >> 8;
}
fixed_t lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * posXW), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
int lightshade = lightpos >> 8;
lightshade = ((256 - lightshade) * NUMCOLORMAPS) & 0xffffff00;
uint8_t shadedfg = colormaps[lightshade + fg];
uint8_t shadedfg;
if (OptT::Flags & SWOPT_FixedLight)
{
shadedfg = colormaps[light + fg];
}
else
{
fixed_t maxvis = 24 * FRACUNIT / 32;
fixed_t maxlight = 31 * FRACUNIT / 32;
int lightshade = (FRACUNIT - clamp<fixed_t>(shade - MIN(maxvis, lightpos), 0, maxlight)) >> 8;
lightshade = ((256 - lightshade) << 5) & 0xffffff00;
shadedfg = colormaps[lightshade + fg];
}
if (OptT::Flags & SWOPT_DynLights)
{
uint32_t lit_r = RPART(dynlightcolor);
uint32_t lit_g = GPART(dynlightcolor);
uint32_t lit_b = BPART(dynlightcolor);
#ifdef NO_SSE
float rcp_posXW = 1.0f / posXW;
#else
float rcp_posXW = _mm_cvtss_f32(_mm_rcp_ss(_mm_set_ss(posXW)));
#endif
float worldposX = posWorldX * rcp_posXW;
float worldposY = posWorldY * rcp_posXW;
float worldposZ = posWorldZ * rcp_posXW;
for (int i = 0; i < num_lights; i++)
{
float lightposX = lights[i].x;
float lightposY = lights[i].y;
float lightposZ = lights[i].z;
float light_radius = lights[i].radius;
uint32_t light_color = lights[i].color;
bool is_attenuated = light_radius < 0.0f;
if (is_attenuated)
light_radius = -light_radius;
// L = light-pos
// dist = sqrt(dot(L, L))
// distance_attenuation = 1 - MIN(dist * (1/radius), 1)
float Lx = lightposX - worldposX;
float Ly = lightposY - worldposY;
float Lz = lightposZ - worldposZ;
float dist2 = Lx * Lx + Ly * Ly + Lz * Lz;
#ifdef NO_SSE
//float rcp_dist = 1.0f / sqrt(dist2);
float rcp_dist = 1.0f / (dist2 * 0.01f);
#else
float rcp_dist = _mm_cvtss_f32(_mm_rsqrt_ss(_mm_set_ss(dist2)));
#endif
float dist = dist2 * rcp_dist;
float distance_attenuation = 256.0f - MIN(dist * light_radius, 256.0f);
// The simple light type
float simple_attenuation = distance_attenuation;
// The point light type
// diffuse = max(dot(N,normalize(L)),0) * attenuation
Lx *= rcp_dist;
Ly *= rcp_dist;
Lz *= rcp_dist;
float dotNL = worldnormalX * Lx + worldnormalY * Ly + worldnormalZ * Lz;
float point_attenuation = MAX(dotNL, 0.0f) * distance_attenuation;
uint32_t attenuation = (uint32_t)(is_attenuated ? (int32_t)point_attenuation : (int32_t)simple_attenuation);
lit_r += (RPART(light_color) * attenuation) >> 8;
lit_g += (GPART(light_color) * attenuation) >> 8;
lit_b += (BPART(light_color) * attenuation) >> 8;
}
if (lit_r || lit_g || lit_b)
{
lit_r = MIN(lit_r, (uint32_t)256);
lit_g = MIN(lit_g, (uint32_t)256);
lit_b = MIN(lit_b, (uint32_t)256);
uint32_t fgrgb = GPalette.BaseColors[fg];
uint32_t shadedfgrgb = GPalette.BaseColors[shadedfg];
uint32_t out_r = MIN(((RPART(fgrgb) * lit_r) >> 8) + RPART(shadedfgrgb), (uint32_t)255);
uint32_t out_g = MIN(((GPART(fgrgb) * lit_g) >> 8) + GPART(shadedfgrgb), (uint32_t)255);
uint32_t out_b = MIN(((BPART(fgrgb) * lit_b) >> 8) + BPART(shadedfgrgb), (uint32_t)255);
shadedfg = RGB256k.All[((out_r >> 2) << 12) | ((out_g >> 2) << 6) | (out_b >> 2)];
}
}
if (ModeT::BlendSrc == STYLEALPHA_One && ModeT::BlendDest == STYLEALPHA_Zero)
{
@ -1826,11 +2108,11 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
destLine[x] = RGB256k.All[((out_r >> 2) << 12) | ((out_g >> 2) << 6) | (out_b >> 2)];
}
else if (fgalpha == 255)
else if (ModeT::BlendSrc == STYLEALPHA_Src && ModeT::BlendDest == STYLEALPHA_InvSrc && fgalpha == 255)
{
destLine[x] = shadedfg;
}
else if (fgalpha != 0)
else if (ModeT::BlendSrc != STYLEALPHA_Src || ModeT::BlendDest != STYLEALPHA_InvSrc || fgalpha != 0)
{
uint32_t src = GPalette.BaseColors[shadedfg];
uint32_t dest = GPalette.BaseColors[destLine[x]];
@ -1893,10 +2175,39 @@ void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
posXW += stepXW;
posXU += stepXU;
posXV += stepXV;
if (OptT::Flags & SWOPT_DynLights)
{
posWorldX += stepWorldX;
posWorldY += stepWorldY;
posWorldZ += stepWorldZ;
}
if (!(OptT::Flags & SWOPT_FixedLight))
lightpos += lightstep;
x++;
}
}
template<typename ModeT>
void DrawSpan8(int y, int x0, int x1, const TriDrawTriangleArgs *args)
{
using namespace TriScreenDrawerModes;
if (args->uniforms->NumLights() == 0)
{
if (!args->uniforms->FixedLight())
DrawSpanOpt8<ModeT, DrawerOptC>(y, x0, x1, args);
else
DrawSpanOpt8<ModeT, DrawerOptCF>(y, x0, x1, args);
}
else
{
if (!args->uniforms->FixedLight())
DrawSpanOpt8<ModeT, DrawerOptLC>(y, x0, x1, args);
else
DrawSpanOpt8<ModeT, DrawerOptLCF>(y, x0, x1, args);
}
}
void(*ScreenTriangle::SpanDrawers8[])(int, int, int, const TriDrawTriangleArgs *) =
{
&DrawSpan8<TriScreenDrawerModes::StyleOpaque>,

16
src/polyrenderer/drawers/screen_triangle.h
View file

@ -251,6 +251,22 @@ namespace TriScreenDrawerModes
struct FuzzSampler { static const int Mode = (int)Samplers::Fuzz; };
struct FogBoundarySampler { static const int Mode = (int)Samplers::FogBoundary; };
enum SWOptFlags
{
SWOPT_DynLights = 1,
SWOPT_ColoredFog = 2,
SWOPT_FixedLight = 4
};
struct DrawerOpt { static const int Flags = 0; };
struct DrawerOptF { static const int Flags = SWOPT_FixedLight; };
struct DrawerOptC { static const int Flags = SWOPT_ColoredFog; };
struct DrawerOptCF { static const int Flags = SWOPT_ColoredFog | SWOPT_FixedLight; };
struct DrawerOptL { static const int Flags = SWOPT_DynLights; };
struct DrawerOptLC { static const int Flags = SWOPT_DynLights | SWOPT_ColoredFog; };
struct DrawerOptLF { static const int Flags = SWOPT_DynLights | SWOPT_FixedLight; };
struct DrawerOptLCF { static const int Flags = SWOPT_DynLights | SWOPT_ColoredFog | SWOPT_FixedLight; };
static const int fuzzcolormap[FUZZTABLE] =
{
6, 11, 6, 11, 6, 6, 11, 6, 6, 11,

14
src/polyrenderer/scene/poly_model.cpp
View file

@ -56,12 +56,16 @@ void PolyModelRenderer::BeginDrawModel(AActor *actor, FSpriteModelFrame *smf, co
ModelActor = actor;
const_cast<VSMatrix &>(objectToWorldMatrix).copy(ObjectToWorld.Matrix);
SetTransform();
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal] || !!(smf->flags & MDL_DONTCULLBACKFACES))
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, true);
}
void PolyModelRenderer::EndDrawModel(AActor *actor, FSpriteModelFrame *smf)
{
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal] || !!(smf->flags & MDL_DONTCULLBACKFACES))
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, false);
ModelActor = nullptr;
}
@ -100,6 +104,8 @@ void PolyModelRenderer::BeginDrawHUDModel(AActor *actor, const VSMatrix &objectT
const_cast<VSMatrix &>(objectToWorldMatrix).copy(ObjectToWorld.Matrix);
SetTransform();
PolyTriangleDrawer::SetWeaponScene(Thread->DrawQueue, true);
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal])
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, true);
}
@ -107,6 +113,8 @@ void PolyModelRenderer::EndDrawHUDModel(AActor *actor)
{
ModelActor = nullptr;
PolyTriangleDrawer::SetWeaponScene(Thread->DrawQueue, false);
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal])
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, false);
}
@ -146,8 +154,7 @@ void PolyModelRenderer::DrawArrays(int start, int count)
args.SetLight(GetColorTable(sector->Colormap, sector->SpecialColors[sector_t::sprites], true), lightlevel, PolyRenderer::Instance()->Light.SpriteGlobVis(foggy), fullbrightSprite);
args.SetStencilTestValue(StencilValue);
args.SetClipPlane(0, PolyClipPlane());
args.SetStyle(TriBlendMode::Opaque);
args.SetTexture(SkinTexture, DefaultRenderStyle());
args.SetStyle(ModelActor->RenderStyle, ModelActor->Alpha, ModelActor->fillcolor, ModelActor->Translation, SkinTexture, fullbrightSprite);
args.SetDepthTest(true);
args.SetWriteDepth(true);
args.SetWriteStencil(false);
@ -169,8 +176,7 @@ void PolyModelRenderer::DrawElements(int numIndices, size_t offset)
args.SetLight(GetColorTable(sector->Colormap, sector->SpecialColors[sector_t::sprites], true), lightlevel, PolyRenderer::Instance()->Light.SpriteGlobVis(foggy), fullbrightSprite);
args.SetStencilTestValue(StencilValue);
args.SetClipPlane(0, PolyClipPlane());
args.SetStyle(TriBlendMode::Opaque);
args.SetTexture(SkinTexture, DefaultRenderStyle());
args.SetStyle(ModelActor->RenderStyle, ModelActor->Alpha, ModelActor->fillcolor, ModelActor->Translation, SkinTexture, fullbrightSprite);
args.SetDepthTest(true);
args.SetWriteDepth(true);
args.SetWriteStencil(false);

9
src/polyrenderer/scene/poly_wall.cpp
View file

@ -354,7 +354,14 @@ void RenderPolyWall::Render(PolyRenderThread *thread)
}
else
{
args.SetStyle(Additive ? TriBlendMode::Add : TriBlendMode::Normal, MIN(Alpha, 1.0));
double a = MIN(Alpha, 1.0);
if (Additive)
args.SetStyle(TriBlendMode::Add, a);
else if (a < 1.0)
args.SetStyle(TriBlendMode::Translucent, a);
else
args.SetStyle(TriBlendMode::Normal);
args.SetStencilTestValue(StencilValue + 1);
args.SetDepthTest(true);
args.SetWriteDepth(true);

26
src/swrenderer/things/r_model.cpp
View file

@ -120,13 +120,17 @@ namespace swrenderer
}
SetTransform();
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal] || !!(smf->flags & MDL_DONTCULLBACKFACES))
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, true);
}
void SWModelRenderer::EndDrawModel(AActor *actor, FSpriteModelFrame *smf)
{
ModelActor = nullptr;
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal] || !!(smf->flags & MDL_DONTCULLBACKFACES))
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, false);
ModelActor = nullptr;
}
IModelVertexBuffer *SWModelRenderer::CreateVertexBuffer(bool needindex, bool singleframe)
@ -185,6 +189,8 @@ namespace swrenderer
ClipBottom = {};
SetTransform();
PolyTriangleDrawer::SetWeaponScene(Thread->DrawQueue, true);
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal])
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, true);
}
@ -192,6 +198,8 @@ namespace swrenderer
{
ModelActor = nullptr;
PolyTriangleDrawer::SetWeaponScene(Thread->DrawQueue, false);
if (actor->RenderStyle == LegacyRenderStyles[STYLE_Normal])
PolyTriangleDrawer::SetTwoSided(Thread->DrawQueue, false);
}
@ -229,13 +237,7 @@ namespace swrenderer
PolyDrawArgs args;
args.SetLight(GetColorTable(sector->Colormap, sector->SpecialColors[sector_t::sprites], true), lightlevel, Thread->Light->SpriteGlobVis(foggy), fullbrightSprite);
args.SetStyle(TriBlendMode::Opaque);
if (Thread->Viewport->RenderTarget->IsBgra())
args.SetTexture((const uint8_t *)SkinTexture->GetPixelsBgra(), SkinTexture->GetWidth(), SkinTexture->GetHeight());
else
args.SetTexture(SkinTexture->GetPixels(DefaultRenderStyle()), SkinTexture->GetWidth(), SkinTexture->GetHeight());
args.SetStyle(ModelActor->RenderStyle, ModelActor->Alpha, ModelActor->fillcolor, ModelActor->Translation, SkinTexture, fullbrightSprite);
args.SetDepthTest(true);
args.SetWriteDepth(true);
args.SetWriteStencil(false);
@ -259,13 +261,7 @@ namespace swrenderer
PolyDrawArgs args;
args.SetLight(GetColorTable(sector->Colormap, sector->SpecialColors[sector_t::sprites], true), lightlevel, Thread->Light->SpriteGlobVis(foggy), fullbrightSprite);
args.SetStyle(TriBlendMode::Opaque);
if (Thread->Viewport->RenderTarget->IsBgra())
args.SetTexture((const uint8_t *)SkinTexture->GetPixelsBgra(), SkinTexture->GetWidth(), SkinTexture->GetHeight());
else
args.SetTexture(SkinTexture->GetPixels(DefaultRenderStyle()), SkinTexture->GetWidth(), SkinTexture->GetHeight());
args.SetStyle(ModelActor->RenderStyle, ModelActor->Alpha, ModelActor->fillcolor, ModelActor->Translation, SkinTexture, fullbrightSprite);
args.SetDepthTest(true);
args.SetWriteDepth(true);
args.SetWriteStencil(false);