zdoom/gzdoom-gles
0
0
Fork 0
mirror of https://github.com/ZDoom/gzdoom-gles.git synced 2024-11-25 13:41:05 +00:00
Code Issues Projects Releases Packages Wiki Activity

- replaced the softpoly drawers with templated versions

Browse source
This commit is contained in:
Magnus Norddahl 2017-03-21 14:12:19 +01:00
parent 7aa6a6b0b3
commit d437b342db
14 changed files with 1131 additions and 16129 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

1
src/CMakeLists.txt
View file

@ -784,7 +784,6 @@ file( GLOB HEADER_FILES
polyrenderer/*.h
polyrenderer/math/*.h
polyrenderer/drawers/*.h
polyrenderer/drawers/*.php
polyrenderer/scene/*.h
gl/*.h
gl/api/*.h

649
src/polyrenderer/drawers/poly_drawer32_sse2.h Normal file
View file

@ -0,0 +1,649 @@
/*
** Projected triangle drawer
** Copyright (c) 2016 Magnus Norddahl
**
** This software is provided 'as-is', without any express or implied
** warranty. In no event will the authors be held liable for any damages
** arising from the use of this software.
**
** Permission is granted to anyone to use this software for any purpose,
** including commercial applications, and to alter it and redistribute it
** freely, subject to the following restrictions:
**
** 1. The origin of this software must not be misrepresented; you must not
** claim that you wrote the original software. If you use this software
** in a product, an acknowledgment in the product documentation would be
** appreciated but is not required.
** 2. Altered source versions must be plainly marked as such, and must not be
** misrepresented as being the original software.
** 3. This notice may not be removed or altered from any source distribution.
**
*/
#pragma once
#include "screen_triangle.h"
template<typename BlendT, typename SamplerT>
class TriScreenDrawer32
{
public:
static void Execute(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
{
using namespace TriScreenDrawerModes;
auto flags = args->uniforms->flags;
bool is_simple_shade = (flags & TriUniforms::simple_shade) == TriUniforms::simple_shade;
if (SamplerT::Mode == (int)Samplers::Texture)
{
bool is_nearest_filter = (flags & TriUniforms::nearest_filter) == TriUniforms::nearest_filter;
if (is_simple_shade)
{
if (is_nearest_filter)
Loop<SimpleShade, NearestFilter>(args, thread);
else
Loop<SimpleShade, LinearFilter>(args, thread);
}
else
{
if (is_nearest_filter)
Loop<AdvancedShade, NearestFilter>(args, thread);
else
Loop<AdvancedShade, LinearFilter>(args, thread);
}
}
else // no linear filtering for translated, shaded, fill or skycap
{
if (is_simple_shade)
{
Loop<SimpleShade, NearestFilter>(args, thread);
}
else
{
Loop<AdvancedShade, NearestFilter>(args, thread);
}
}
}
template<typename ShadeModeT, typename FilterModeT>
FORCEINLINE static void VECTORCALL Loop(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
{
using namespace TriScreenDrawerModes;
int numSpans = thread->NumFullSpans;
auto fullSpans = thread->FullSpans;
int numBlocks = thread->NumPartialBlocks;
auto partialBlocks = thread->PartialBlocks;
int startX = thread->StartX;
int startY = thread->StartY;
auto flags = args->uniforms->flags;
bool is_fixed_light = (flags & TriUniforms::fixed_light) == TriUniforms::fixed_light;
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
auto colormaps = args->colormaps;
uint32_t srcalpha = args->uniforms->srcalpha;
uint32_t destalpha = args->uniforms->destalpha;
// Calculate gradients
const TriVertex &v1 = *args->v1;
const TriVertex &v2 = *args->v2;
const TriVertex &v3 = *args->v3;
ScreenTriangleStepVariables gradientX;
ScreenTriangleStepVariables gradientY;
ScreenTriangleStepVariables start;
gradientX.W = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
gradientY.W = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
start.W = v1.w + gradientX.W * (startX - v1.x) + gradientY.W * (startY - v1.y);
for (int i = 0; i < TriVertex::NumVarying; i++)
{
gradientX.Varying[i] = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
gradientY.Varying[i] = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
start.Varying[i] = v1.varying[i] * v1.w + gradientX.Varying[i] * (startX - v1.x) + gradientY.Varying[i] * (startY - v1.y);
}
// Output
uint32_t * RESTRICT destOrg = (uint32_t*)args->dest;
int pitch = args->pitch;
// Light
uint32_t light = args->uniforms->light;
float shade = (64.0f - (light * 255 / 256 + 12.0f) * 32.0f / 128.0f) / 32.0f;
float globVis = args->uniforms->globvis * (1.0f / 32.0f);
// Sampling stuff
uint32_t color = args->uniforms->color;
const uint32_t * RESTRICT translation = (const uint32_t *)args->translation;
const uint32_t * RESTRICT texPixels = (const uint32_t *)args->texturePixels;
uint32_t texWidth = args->textureWidth;
uint32_t texHeight = args->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->desaturate, 256 - args->uniforms->desaturate, 256 - args->uniforms->desaturate, 256, 256 - args->uniforms->desaturate, 256 - args->uniforms->desaturate, 256 - args->uniforms->desaturate);
shade_fade = _mm_set_epi16(args->uniforms->fade_alpha, args->uniforms->fade_red, args->uniforms->fade_green, args->uniforms->fade_blue, args->uniforms->fade_alpha, args->uniforms->fade_red, args->uniforms->fade_green, args->uniforms->fade_blue);
shade_light = _mm_set_epi16(args->uniforms->light_alpha, args->uniforms->light_red, args->uniforms->light_green, args->uniforms->light_blue, args->uniforms->light_alpha, args->uniforms->light_red, args->uniforms->light_green, args->uniforms->light_blue);
desaturate = args->uniforms->desaturate;
}
else
{
inv_desaturate = _mm_setzero_si128();
shade_fade = _mm_setzero_si128();
shade_fade = _mm_setzero_si128();
shade_light = _mm_setzero_si128();
desaturate = 0;
}
for (int i = 0; i < numSpans; i++)
{
const auto &span = fullSpans[i];
uint32_t *dest = destOrg + span.X + span.Y * pitch;
int width = span.Length;
int height = 8;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (span.X - startX) + gradientY.W * (span.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (span.X - startX) + gradientY.Varying[j] * (span.Y - startY);
for (int y = 0; y < height; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
for (int x = 0; x < width; x++)
{
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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 < 4; ix++)
{
// Load bgcolor
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)(dest + x * 8 + ix * 2)), _mm_setzero_si128());
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2];
ifgcolor[0] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
ifgcolor[1] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
// 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 = Shade<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light);
__m128i outcolor = Blend(fgcolor, bgcolor, ifgcolor[0], ifgcolor[1], srcalpha, destalpha);
// Store result
_mm_storel_epi64((__m128i*)(dest + x * 8 + ix * 2), outcolor);
}
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
}
for (int i = 0; i < numBlocks; i++)
{
const auto &block = partialBlocks[i];
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (block.X - startX) + gradientY.W * (block.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (block.X - startX) + gradientY.Varying[j] * (block.Y - startY);
uint32_t *dest = destOrg + block.X + block.Y * pitch;
uint32_t mask0 = block.Mask0;
uint32_t mask1 = block.Mask1;
// mask0 loop:
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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 < 4; x++)
{
// Load bgcolor
uint32_t desttmp[2];
if (mask0 & (1 << 31)) desttmp[0] = dest[x * 2];
if (mask0 & (1 << 30)) desttmp[1] = dest[x * 2 + 1];
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)desttmp), _mm_setzero_si128());
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2];
ifgcolor[0] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
ifgcolor[1] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
// 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 = Shade<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light);
__m128i outcolor = Blend(fgcolor, bgcolor, ifgcolor[0], ifgcolor[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];
mask0 <<= 2;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
// mask1 loop:
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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 < 4; x++)
{
// Load bgcolor
uint32_t desttmp[2];
if (mask1 & (1 << 31)) desttmp[0] = dest[x * 2];
if (mask1 & (1 << 30)) desttmp[1] = dest[x * 2 + 1];
__m128i bgcolor;
if (BlendT::Mode != (int)BlendModes::Opaque)
bgcolor = _mm_unpacklo_epi8(_mm_loadl_epi64((__m128i*)desttmp), _mm_setzero_si128());
else
bgcolor = _mm_setzero_si128();
// Sample fgcolor
unsigned int ifgcolor[2];
ifgcolor[0] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
ifgcolor[1] = Sample<FilterModeT>(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, oneU, oneV, color, translation);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
// 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 = Shade<ShadeModeT>(fgcolor, mlight, ifgcolor[0], ifgcolor[1], desaturate, inv_desaturate, shade_fade_lit, shade_light);
__m128i outcolor = Blend(fgcolor, bgcolor, ifgcolor[0], ifgcolor[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];
mask1 <<= 2;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
}
}
private:
template<typename FilterModeT>
FORCEINLINE static unsigned int VECTORCALL Sample(int32_t u, int32_t v, const uint32_t *texPixels, int texWidth, int texHeight, uint32_t oneU, uint32_t oneV, uint32_t color, const uint32_t *translation)
{
using namespace TriScreenDrawerModes;
uint32_t texel;
if (SamplerT::Mode == (int)Samplers::Shaded || SamplerT::Mode == (int)Samplers::Fill)
{
return color;
}
else if (SamplerT::Mode == (int)Samplers::Translated)
{
const uint8_t *texpal = (const uint8_t *)texPixels;
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
return translation[texpal[texelX * texHeight + texelY]];
}
else if (FilterModeT::Mode == (int)FilterModes::Nearest)
{
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
texel = texPixels[texelX * texHeight + texelY];
}
else
{
u -= oneU >> 1;
v -= oneV >> 1;
unsigned int frac_x0 = (((uint32_t)u << 8) >> FRACBITS) * texWidth;
unsigned int frac_x1 = ((((uint32_t)u << 8) + oneU) >> FRACBITS) * texWidth;
unsigned int frac_y0 = (((uint32_t)v << 8) >> FRACBITS) * texHeight;
unsigned int frac_y1 = ((((uint32_t)v << 8) + oneV) >> FRACBITS) * texHeight;
unsigned int x0 = frac_x0 >> FRACBITS;
unsigned int x1 = frac_x1 >> FRACBITS;
unsigned int y0 = frac_y0 >> FRACBITS;
unsigned int y1 = frac_y1 >> FRACBITS;
unsigned int p00 = texPixels[x0 * texHeight + y0];
unsigned int p01 = texPixels[x0 * texHeight + y1];
unsigned int p10 = texPixels[x1 * texHeight + y0];
unsigned int p11 = texPixels[x1 * texHeight + y1];
unsigned int inv_a = (frac_x1 >> (FRACBITS - 4)) & 15;
unsigned int inv_b = (frac_y1 >> (FRACBITS - 4)) & 15;
unsigned int a = 16 - inv_a;
unsigned int b = 16 - inv_b;
unsigned int sred = (RPART(p00) * (a * b) + RPART(p01) * (inv_a * b) + RPART(p10) * (a * inv_b) + RPART(p11) * (inv_a * inv_b) + 127) >> 8;
unsigned int sgreen = (GPART(p00) * (a * b) + GPART(p01) * (inv_a * b) + GPART(p10) * (a * inv_b) + GPART(p11) * (inv_a * inv_b) + 127) >> 8;
unsigned int sblue = (BPART(p00) * (a * b) + BPART(p01) * (inv_a * b) + BPART(p10) * (a * inv_b) + BPART(p11) * (inv_a * inv_b) + 127) >> 8;
unsigned int salpha = (APART(p00) * (a * b) + APART(p01) * (inv_a * b) + APART(p10) * (a * inv_b) + APART(p11) * (inv_a * inv_b) + 127) >> 8;
texel = (salpha << 24) | (sred << 16) | (sgreen << 8) | sblue;
}
if (SamplerT::Mode == (int)Samplers::Skycap)
{
int start_fade = 2; // How fast it should fade out
int alpha_top = clamp(v >> (16 - start_fade), 0, 256);
int alpha_bottom = clamp(((2 << 24) - v) >> (16 - start_fade), 0, 256);
int a = MIN(alpha_top, alpha_bottom);
int inv_a = 256 - a;
uint32_t r = RPART(texel);
uint32_t g = GPART(texel);
uint32_t b = BPART(texel);
uint32_t fg_a = APART(texel);
uint32_t bg_red = RPART(color);
uint32_t bg_green = GPART(color);
uint32_t bg_blue = BPART(color);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
return MAKEARGB(fg_a, r, g, b);
}
else
{
return texel;
}
}
template<typename ShadeModeT>
FORCEINLINE static __m128i VECTORCALL Shade(__m128i fgcolor, __m128i mlight, unsigned int ifgcolor0, unsigned int ifgcolor1, int desaturate, __m128i inv_desaturate, __m128i shade_fade, __m128i shade_light)
{
using namespace TriScreenDrawerModes;
if (ShadeModeT::Mode == (int)ShadeMode::Simple)
{
fgcolor = _mm_srli_epi16(_mm_mullo_epi16(fgcolor, mlight), 8);
}
else
{
int blue0 = BPART(ifgcolor0);
int green0 = GPART(ifgcolor0);
int red0 = RPART(ifgcolor0);
int intensity0 = ((red0 * 77 + green0 * 143 + blue0 * 37) >> 8) * desaturate;
int blue1 = BPART(ifgcolor1);
int green1 = GPART(ifgcolor1);
int red1 = RPART(ifgcolor1);
int intensity1 = ((red1 * 77 + green1 * 143 + blue1 * 37) >> 8) * desaturate;
__m128i intensity = _mm_set_epi16(0, intensity1, intensity1, intensity1, 0, intensity0, intensity0, intensity0);
fgcolor = _mm_srli_epi16(_mm_add_epi16(_mm_mullo_epi16(fgcolor, inv_desaturate), intensity), 8);
fgcolor = _mm_mullo_epi16(fgcolor, mlight);
fgcolor = _mm_srli_epi16(_mm_add_epi16(shade_fade, fgcolor), 8);
fgcolor = _mm_srli_epi16(_mm_mullo_epi16(fgcolor, shade_light), 8);
}
return fgcolor;
}
FORCEINLINE static __m128i VECTORCALL Blend(__m128i fgcolor, __m128i bgcolor, unsigned int ifgcolor0, unsigned int ifgcolor1, uint32_t srcalpha, uint32_t destalpha)
{
using namespace TriScreenDrawerModes;
if (BlendT::Mode == (int)BlendModes::Opaque)
{
__m128i outcolor = fgcolor;
outcolor = _mm_packus_epi16(outcolor, _mm_setzero_si128());
return outcolor;
}
else if (BlendT::Mode == (int)BlendModes::Masked)
{
__m128i mask = _mm_cmpeq_epi32(_mm_packus_epi16(fgcolor, _mm_setzero_si128()), _mm_setzero_si128());
mask = _mm_unpacklo_epi8(mask, _mm_setzero_si128());
__m128i outcolor = _mm_or_si128(_mm_and_si128(mask, bgcolor), _mm_andnot_si128(mask, fgcolor));
outcolor = _mm_packus_epi16(outcolor, _mm_setzero_si128());
outcolor = _mm_or_si128(outcolor, _mm_set1_epi32(0xff000000));
return outcolor;
}
else if (BlendT::Mode == (int)BlendModes::AddSrcColorOneMinusSrcColor)
{
__m128i inv_srccolor = _mm_sub_epi16(_mm_set1_epi16(256), _mm_add_epi16(fgcolor, _mm_srli_epi16(fgcolor, 7)));
__m128i outcolor = _mm_add_epi16(fgcolor, _mm_srli_epi16(_mm_mullo_epi16(bgcolor, inv_srccolor), 8));
outcolor = _mm_packus_epi16(outcolor, _mm_setzero_si128());
return outcolor;
}
else
{
uint32_t alpha0 = APART(ifgcolor0);
uint32_t alpha1 = APART(ifgcolor1);
alpha0 += alpha0 >> 7; // 255->256
alpha1 += alpha1 >> 7; // 255->256
uint32_t inv_alpha0 = 256 - alpha0;
uint32_t inv_alpha1 = 256 - alpha1;
uint32_t bgalpha0 = (destalpha * alpha0 + (inv_alpha0 << 8) + 128) >> 8;
uint32_t bgalpha1 = (destalpha * alpha1 + (inv_alpha1 << 8) + 128) >> 8;
uint32_t fgalpha0 = (srcalpha * alpha0 + 128) >> 8;
uint32_t fgalpha1 = (srcalpha * alpha1 + 128) >> 8;
__m128i bgalpha = _mm_set_epi16(bgalpha1, bgalpha1, bgalpha1, bgalpha1, bgalpha0, bgalpha0, bgalpha0, bgalpha0);
__m128i fgalpha = _mm_set_epi16(fgalpha1, fgalpha1, fgalpha1, fgalpha1, fgalpha0, fgalpha0, fgalpha0, fgalpha0);
fgcolor = _mm_mullo_epi16(fgcolor, fgalpha);
bgcolor = _mm_mullo_epi16(bgcolor, bgalpha);
__m128i fg_lo = _mm_unpacklo_epi16(fgcolor, _mm_setzero_si128());
__m128i bg_lo = _mm_unpacklo_epi16(bgcolor, _mm_setzero_si128());
__m128i fg_hi = _mm_unpackhi_epi16(fgcolor, _mm_setzero_si128());
__m128i bg_hi = _mm_unpackhi_epi16(bgcolor, _mm_setzero_si128());
__m128i out_lo, out_hi;
if (BlendT::Mode == (int)BlendModes::AddClamp)
{
out_lo = _mm_add_epi32(fg_lo, bg_lo);
out_hi = _mm_add_epi32(fg_hi, bg_hi);
}
else if (BlendT::Mode == (int)BlendModes::SubClamp)
{
out_lo = _mm_sub_epi32(fg_lo, bg_lo);
out_hi = _mm_sub_epi32(fg_hi, bg_hi);
}
else if (BlendT::Mode == (int)BlendModes::RevSubClamp)
{
out_lo = _mm_sub_epi32(bg_lo, fg_lo);
out_hi = _mm_sub_epi32(bg_hi, fg_hi);
}
out_lo = _mm_srai_epi32(out_lo, 8);
out_hi = _mm_srai_epi32(out_hi, 8);
__m128i outcolor = _mm_packs_epi32(out_lo, out_hi);
outcolor = _mm_packus_epi16(outcolor, _mm_setzero_si128());
outcolor = _mm_or_si128(outcolor, _mm_set1_epi32(0xff000000));
return outcolor;
}
}
static float FindGradientX(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (y0 - y2) - (c0 - c2) * (y1 - y2);
float bottom = (x1 - x2) * (y0 - y2) - (x0 - x2) * (y1 - y2);
return top / bottom;
}
static float FindGradientY(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (x0 - x2) - (c0 - c2) * (x1 - x2);
float bottom = (x0 - x2) * (y1 - y2) - (x1 - x2) * (y0 - y2);
return top / bottom;
}
};

366
src/polyrenderer/drawers/poly_drawer8.h Normal file
View file

@ -0,0 +1,366 @@
/*
** Projected triangle drawer
** Copyright (c) 2016 Magnus Norddahl
**
** This software is provided 'as-is', without any express or implied
** warranty. In no event will the authors be held liable for any damages
** arising from the use of this software.
**
** Permission is granted to anyone to use this software for any purpose,
** including commercial applications, and to alter it and redistribute it
** freely, subject to the following restrictions:
**
** 1. The origin of this software must not be misrepresented; you must not
** claim that you wrote the original software. If you use this software
** in a product, an acknowledgment in the product documentation would be
** appreciated but is not required.
** 2. Altered source versions must be plainly marked as such, and must not be
** misrepresented as being the original software.
** 3. This notice may not be removed or altered from any source distribution.
**
*/
#pragma once
#include "screen_triangle.h"
template<typename BlendT, typename SamplerT>
class TriScreenDrawer8
{
public:
static void Execute(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
{
using namespace TriScreenDrawerModes;
int numSpans = thread->NumFullSpans;
auto fullSpans = thread->FullSpans;
int numBlocks = thread->NumPartialBlocks;
auto partialBlocks = thread->PartialBlocks;
int startX = thread->StartX;
int startY = thread->StartY;
auto flags = args->uniforms->flags;
bool is_fixed_light = (flags & TriUniforms::fixed_light) == TriUniforms::fixed_light;
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
auto colormaps = args->colormaps;
uint32_t srcalpha = args->uniforms->srcalpha;
uint32_t destalpha = args->uniforms->destalpha;
// Calculate gradients
const TriVertex &v1 = *args->v1;
const TriVertex &v2 = *args->v2;
const TriVertex &v3 = *args->v3;
ScreenTriangleStepVariables gradientX;
ScreenTriangleStepVariables gradientY;
ScreenTriangleStepVariables start;
gradientX.W = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
gradientY.W = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
start.W = v1.w + gradientX.W * (startX - v1.x) + gradientY.W * (startY - v1.y);
for (int i = 0; i < TriVertex::NumVarying; i++)
{
gradientX.Varying[i] = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
gradientY.Varying[i] = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
start.Varying[i] = v1.varying[i] * v1.w + gradientX.Varying[i] * (startX - v1.x) + gradientY.Varying[i] * (startY - v1.y);
}
// Output
uint8_t * RESTRICT destOrg = args->dest;
int pitch = args->pitch;
// Light
uint32_t light = args->uniforms->light;
float shade = (64.0f - (light * 255 / 256 + 12.0f) * 32.0f / 128.0f) / 32.0f;
float globVis = args->uniforms->globvis * (1.0f / 32.0f);
// Sampling stuff
uint32_t color = args->uniforms->color;
const uint8_t * RESTRICT translation = args->translation;
const uint8_t * RESTRICT texPixels = args->texturePixels;
uint32_t texWidth = args->textureWidth;
uint32_t texHeight = args->textureHeight;
for (int i = 0; i < numSpans; i++)
{
const auto &span = fullSpans[i];
uint8_t *dest = destOrg + span.X + span.Y * pitch;
int width = span.Length;
int height = 8;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (span.X - startX) + gradientY.W * (span.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (span.X - startX) + gradientY.Varying[j] * (span.Y - startY);
for (int y = 0; y < height; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
for (int x = 0; x < width; x++)
{
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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[x * 8 + ix];
uint8_t fgcolor = Sample(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, color, translation);
dest[x * 8 + ix] = ShadeAndBlend(fgcolor, bgcolor, lightshade, colormaps, srcalpha, destalpha);
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
}
for (int i = 0; i < numBlocks; i++)
{
const auto &block = partialBlocks[i];
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (block.X - startX) + gradientY.W * (block.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (block.X - startX) + gradientY.Varying[j] * (block.Y - startY);
uint8_t *dest = destOrg + block.X + block.Y * pitch;
uint32_t mask0 = block.Mask0;
uint32_t mask1 = block.Mask1;
// mask0 loop:
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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];
uint8_t fgcolor = Sample(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, color, translation);
dest[x] = ShadeAndBlend(fgcolor, bgcolor, lightshade, colormaps, srcalpha, destalpha);
}
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
mask0 <<= 1;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
// mask1 loop:
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * 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);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 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];
uint8_t fgcolor = Sample(varyingPos[0], varyingPos[1], texPixels, texWidth, texHeight, color, translation);
dest[x] = ShadeAndBlend(fgcolor, bgcolor, lightshade, colormaps, srcalpha, destalpha);
}
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
mask1 <<= 1;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
}
}
private:
FORCEINLINE static unsigned int VECTORCALL Sample(int32_t u, int32_t v, const uint8_t *texPixels, int texWidth, int texHeight, uint32_t color, const uint8_t *translation)
{
using namespace TriScreenDrawerModes;
uint8_t texel;
if (SamplerT::Mode == (int)Samplers::Shaded || SamplerT::Mode == (int)Samplers::Fill)
{
return color;
}
else if (SamplerT::Mode == (int)Samplers::Translated)
{
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
return translation[texPixels[texelX * texHeight + texelY]];
}
else
{
uint32_t texelX = ((((uint32_t)u << 8) >> 16) * texWidth) >> 16;
uint32_t texelY = ((((uint32_t)v << 8) >> 16) * texHeight) >> 16;
texel = texPixels[texelX * texHeight + texelY];
}
if (SamplerT::Mode == (int)Samplers::Skycap)
{
/*
int start_fade = 2; // How fast it should fade out
int alpha_top = clamp(v >> (16 - start_fade), 0, 256);
int alpha_bottom = clamp(((2 << 24) - v) >> (16 - start_fade), 0, 256);
int a = MIN(alpha_top, alpha_bottom);
int inv_a = 256 - a;
uint32_t r = RPART(texel);
uint32_t g = GPART(texel);
uint32_t b = BPART(texel);
uint32_t fg_a = APART(texel);
uint32_t bg_red = RPART(color);
uint32_t bg_green = GPART(color);
uint32_t bg_blue = BPART(color);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
return MAKEARGB(fg_a, r, g, b);
*/
return texel;
}
else
{
return texel;
}
}
FORCEINLINE static uint8_t ShadeAndBlend(uint8_t fgcolor, uint8_t bgcolor, uint32_t lightshade, const uint8_t *colormaps, uint32_t srcalpha, uint32_t destalpha)
{
using namespace TriScreenDrawerModes;
lightshade = ((256 - lightshade) * (NUMCOLORMAPS - 1) + (NUMCOLORMAPS - 1) / 2) / 256;
uint8_t shadedfg = colormaps[lightshade * 256 + fgcolor];
if (BlendT::Mode == (int)BlendModes::Opaque)
{
return shadedfg;
}
else if (BlendT::Mode == (int)BlendModes::Masked)
{
return (fgcolor != 0) ? shadedfg : bgcolor;
}
else if (BlendT::Mode == (int)BlendModes::AddSrcColorOneMinusSrcColor)
{
return (fgcolor != 0) ? shadedfg : bgcolor;
}
else
{
if (BlendT::Mode == (int)BlendModes::AddClamp)
{
}
else if (BlendT::Mode == (int)BlendModes::SubClamp)
{
}
else if (BlendT::Mode == (int)BlendModes::RevSubClamp)
{
}
return (fgcolor != 0) ? shadedfg : bgcolor;
}
}
static float FindGradientX(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (y0 - y2) - (c0 - c2) * (y1 - y2);
float bottom = (x1 - x2) * (y0 - y2) - (x0 - x2) * (y1 - y2);
return top / bottom;
}
static float FindGradientY(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (x0 - x2) - (c0 - c2) * (x1 - x2);
float bottom = (x0 - x2) * (y1 - y2) - (x1 - x2) * (y0 - y2);
return top / bottom;
}
};

15414
src/polyrenderer/drawers/poly_drawers.h
View file

File diff suppressed because it is too large Load diff

449
src/polyrenderer/drawers/poly_drawers.php
View file

@ -1,449 +0,0 @@
/*
** Projected triangle drawer
** Copyright (c) 2016 Magnus Norddahl
**
** This software is provided 'as-is', without any express or implied
** warranty. In no event will the authors be held liable for any damages
** arising from the use of this software.
**
** Permission is granted to anyone to use this software for any purpose,
** including commercial applications, and to alter it and redistribute it
** freely, subject to the following restrictions:
**
** 1. The origin of this software must not be misrepresented; you must not
** claim that you wrote the original software. If you use this software
** in a product, an acknowledgment in the product documentation would be
** appreciated but is not required.
** 2. Altered source versions must be plainly marked as such, and must not be
** misrepresented as being the original software.
** 3. This notice may not be removed or altered from any source distribution.
**
*/
/*
Warning: this C++ source file has been auto-generated. Please modify the original php script that generated it.
*/
#pragma once
#include "screen_triangle.h"
static float FindGradientX(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (y0 - y2) - (c0 - c2) * (y1 - y2);
float bottom = (x1 - x2) * (y0 - y2) - (x0 - x2) * (y1 - y2);
return top / bottom;
}
static float FindGradientY(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
{
float top = (c1 - c2) * (x0 - x2) - (c0 - c2) * (x1 - x2);
float bottom = (x0 - x2) * (y1 - y2) - (x1 - x2) * (y0 - y2);
return top / bottom;
}
<?
OutputDrawers(true, true, false);
OutputDrawers(true, false, false);
OutputDrawers(false, true, false);
OutputDrawers(false, false, false);
OutputDrawers(true, true, true);
OutputDrawers(true, false, true);
OutputDrawers(false, true, true);
OutputDrawers(false, false, true);
function OutputDrawers($isTruecolor, $isColorFill, $isListEntry)
{
$namePrefix = "";
if ($isTruecolor == true && $isColorFill == true)
{
$namePrefix = "TriFill32";
}
else if ($isTruecolor == true)
{
$namePrefix = "TriDraw32";
}
else if ($isColorFill == true)
{
$namePrefix = "TriFill8";
}
else
{
$namePrefix = "TriDraw8";
}
if ($isListEntry)
{ ?>
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::<?=$namePrefix?> =
{
<? }
OutputDrawer($namePrefix."Copy", "opaque", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."AlphaBlend", "masked", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."AddSolid", "translucent", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."Add", "add", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."Sub", "sub", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."RevSub", "revsub", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."Stencil", "stencil", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."Shaded", "shaded", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."TranslateCopy", "opaque", true, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."TranslateAlphaBlend", "masked", true, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."TranslateAdd", "add", true, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."TranslateSub", "sub", true, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."TranslateRevSub", "revsub", true, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."AddSrcColorOneMinusSrcColor", "addsrccolor", false, $isTruecolor, $isColorFill, $isListEntry);
OutputDrawer($namePrefix."Skycap", "skycap", false, $isTruecolor, $isColorFill, $isListEntry);
if ($isListEntry)
{ ?>
};
<? }
}
function OutputDrawer($drawerName, $blendmode, $isTranslated, $isTruecolor, $isColorFill, $isListEntry)
{
if ($isListEntry)
{ ?>
&<?=$drawerName?>,
<? }
else
{
GenerateDrawer($drawerName, $blendmode, $isTranslated, $isTruecolor, $isColorFill);
}
}
function GenerateDrawer($drawerName, $blendmode, $isTranslated, $isTruecolor, $isColorFill)
{
$pixeltype = $isTruecolor ? "uint32_t" : "uint8_t";
?>
static void <?=$drawerName?>(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
{
int numSpans = thread->NumFullSpans;
auto fullSpans = thread->FullSpans;
int numBlocks = thread->NumPartialBlocks;
auto partialBlocks = thread->PartialBlocks;
int startX = thread->StartX;
int startY = thread->StartY;
auto flags = args->uniforms->flags;
bool is_simple_shade = (flags & TriUniforms::simple_shade) == TriUniforms::simple_shade;
bool is_nearest_filter = (flags & TriUniforms::nearest_filter) == TriUniforms::nearest_filter;
bool is_fixed_light = (flags & TriUniforms::fixed_light) == TriUniforms::fixed_light;
uint32_t lightmask = is_fixed_light ? 0 : 0xffffffff;
auto colormaps = args->colormaps;
uint32_t srcalpha = args->uniforms->srcalpha;
uint32_t destalpha = args->uniforms->destalpha;
// Calculate gradients
const TriVertex &v1 = *args->v1;
const TriVertex &v2 = *args->v2;
const TriVertex &v3 = *args->v3;
ScreenTriangleStepVariables gradientX;
ScreenTriangleStepVariables gradientY;
ScreenTriangleStepVariables start;
gradientX.W = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
gradientY.W = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
start.W = v1.w + gradientX.W * (startX - v1.x) + gradientY.W * (startY - v1.y);
for (int i = 0; i < TriVertex::NumVarying; i++)
{
gradientX.Varying[i] = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
gradientY.Varying[i] = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
start.Varying[i] = v1.varying[i] * v1.w + gradientX.Varying[i] * (startX - v1.x) + gradientY.Varying[i] * (startY - v1.y);
}
<? if ($isTranslated || $blendmode == "shaded")
{ ?>
const uint8_t * RESTRICT texPixels = args->texturePixels;
const <?=$pixeltype?> * RESTRICT translation = (const <?=$pixeltype?> *)args->translation;
<? }
else
{ ?>
const <?=$pixeltype?> * RESTRICT texPixels = (const <?=$pixeltype?> *)args->texturePixels;
<? }?>
uint32_t texWidth = args->textureWidth;
uint32_t texHeight = args->textureHeight;
<?=$pixeltype?> * RESTRICT destOrg = (<?=$pixeltype?>*)args->dest;
int pitch = args->pitch;
uint32_t light = args->uniforms->light;
float shade = (64.0f - (light * 255 / 256 + 12.0f) * 32.0f / 128.0f) / 32.0f;
float globVis = args->uniforms->globvis * (1.0f / 32.0f);
<?=$pixeltype?> color = args->uniforms->color;
for (int i = 0; i < numSpans; i++)
{
const auto &span = fullSpans[i];
<?=$pixeltype?> *dest = destOrg + span.X + span.Y * pitch;
int width = span.Length;
int height = 8;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (span.X - startX) + gradientY.W * (span.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (span.X - startX) + gradientY.Varying[j] * (span.Y - startY);
for (int y = 0; y < height; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * rcpW);
int lightpos = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosY.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
lightpos = (lightpos & lightmask) | ((light << 8) & ~lightmask);
for (int x = 0; x < width; x++)
{
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 8;
}
int lightnext = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
int lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
for (int ix = 0; ix < 8; ix++)
{
<?=$pixeltype?> *destptr = dest + x * 8 + ix;
<? ProcessPixel($blendmode, $isTranslated, $isTruecolor, $isColorFill, $pixeltype); ?>
*destptr = fg;
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
}
for (int i = 0; i < numBlocks; i++)
{
const auto &block = partialBlocks[i];
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (block.X - startX) + gradientY.W * (block.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (block.X - startX) + gradientY.Varying[j] * (block.Y - startY);
<?=$pixeltype?> *dest = destOrg + block.X + block.Y * pitch;
uint32_t mask0 = block.Mask0;
uint32_t mask1 = block.Mask1;
<?
for ($i = 0; $i < 2; $i++)
{
$coveragemask = ($i == 0) ? "mask0" : "mask1";
?>
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * rcpW);
int 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);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 8;
}
int lightnext = FRACUNIT - (int)(clamp(shade - MIN(24.0f / 32.0f, globVis * blockPosX.W), 0.0f, 31.0f / 32.0f) * (float)FRACUNIT);
int lightstep = (lightnext - lightpos) / 8;
lightstep = lightstep & lightmask;
for (int x = 0; x < 8; x++)
{
if (<?=$coveragemask?> & (1 << 31))
{
<?=$pixeltype?> *destptr = dest + x;
<? ProcessPixel($blendmode, $isTranslated, $isTruecolor, $isColorFill, $pixeltype); ?>
*destptr = fg;
}
<?=$coveragemask?> <<= 1;
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
}
<? } ?>
}
}
<?
}
function ProcessPixel($blendmode, $isTranslated, $isTruecolor, $isColorFill, $pixeltype)
{
if ($isColorFill || $blendmode == "shaded")
{ ?>
<?=$pixeltype?> fg = color;
<? }
else
{ ?>
int texelX = ((((uint32_t)varyingPos[0] << 8) >> 16) * texWidth) >> 16;
int texelY = ((((uint32_t)varyingPos[1] << 8) >> 16) * texHeight) >> 16;
<?=$pixeltype?> fg = texPixels[texelX * texHeight + texelY];
<? }
if ($isTranslated)
{ ?>
fg = translation[fg];
<? }
if ($isTruecolor)
{ ?>
uint32_t r = RPART(fg);
uint32_t g = GPART(fg);
uint32_t b = BPART(fg);
r = (r * lightpos) >> 16;
g = (g * lightpos) >> 16;
b = (b * lightpos) >> 16;
<? TruecolorBlend($blendmode); ?>
fg = 0xff000000 | (r << 16) | (g << 8) | b;
<?
}
else
{ ?>
int colormapindex = MIN(((256 - (lightpos >> 8)) * 32) >> 8, 31) << 8;
fg = colormaps[colormapindex + fg];
<? }
}
function TruecolorBlend($blendmode)
{
if ($blendmode == "opaque")
{
}
else if ($blendmode == "masked")
{ ?>
uint32_t a = APART(fg);
a += a >> 7;
uint32_t inv_a = 256 - a;
uint32_t bg = *destptr;
uint32_t bg_red = RPART(bg);
uint32_t bg_green = GPART(bg);
uint32_t bg_blue = BPART(bg);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
<? }
else if ($blendmode == "translucent")
{ ?>
<? }
else if ($blendmode == "shaded")
{ ?>
int texelX = ((((uint32_t)varyingPos[0] << 8) >> 16) * texWidth) >> 16;
int texelY = ((((uint32_t)varyingPos[1] << 8) >> 16) * texHeight) >> 16;
int sample = texPixels[texelX * texHeight + texelY];
uint32_t fgalpha = sample;//clamp(sample, 0, 64) * 4;
uint32_t inv_fgalpha = 256 - fgalpha;
int a = (fgalpha * srcalpha + 128) >> 8;
int inv_a = (destalpha * fgalpha + 256 * inv_fgalpha + 128) >> 8;
uint32_t bg = *destptr;
uint32_t bg_red = RPART(bg);
uint32_t bg_green = GPART(bg);
uint32_t bg_blue = BPART(bg);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
<? }
else if ($blendmode == "stencil")
{ ?>
uint32_t fgalpha = APART(fg);
uint32_t inv_fgalpha = 256 - fgalpha;
int a = (fgalpha * srcalpha + 128) >> 8;
int inv_a = (destalpha * fgalpha + 256 * inv_fgalpha + 128) >> 8;
uint32_t bg = *destptr;
uint32_t bg_red = RPART(bg);
uint32_t bg_green = GPART(bg);
uint32_t bg_blue = BPART(bg);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
<? }
else if ($blendmode == "addsrccolor")
{ ?>
uint32_t inv_r = 256 - (r + (r >> 7));
uint32_t inv_g = 256 - (g + (r >> 7));
uint32_t inv_b = 256 - (b + (r >> 7));
uint32_t bg = *destptr;
uint32_t bg_red = RPART(bg);
uint32_t bg_green = GPART(bg);
uint32_t bg_blue = BPART(bg);
r = r + ((bg_red * inv_r + 127) >> 8);
g = g + ((bg_green * inv_g + 127) >> 8);
b = b + ((bg_blue * inv_b + 127) >> 8);
<? }
else if ($blendmode == "skycap")
{ ?>
int start_fade = 2; // How fast it should fade out
int alpha_top = clamp(varyingPos[1] >> (16 - start_fade), 0, 256);
int alpha_bottom = clamp(((2 << 24) - varyingPos[1]) >> (16 - start_fade), 0, 256);
int a = MIN(alpha_top, alpha_bottom);
int inv_a = 256 - a;
uint32_t bg_red = RPART(color);
uint32_t bg_green = GPART(color);
uint32_t bg_blue = BPART(color);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
<? }
else
{ ?>
uint32_t a = APART(fg);
a += a >> 7;
uint32_t inv_a = 256 - a;
uint32_t bg = *destptr;
uint32_t bg_red = RPART(bg);
uint32_t bg_green = GPART(bg);
uint32_t bg_blue = BPART(bg);
r = (r * a + bg_red * inv_a + 127) >> 8;
g = (g * a + bg_green * inv_a + 127) >> 8;
b = (b * a + bg_blue * inv_a + 127) >> 8;
<? }
}
?>

333
src/polyrenderer/drawers/screen_triangle.cpp
View file

@ -36,7 +36,10 @@
#include "poly_triangle.h"
#include "swrenderer/drawers/r_draw_rgba.h"
#include "screen_triangle.h"
#include "poly_drawers.h"
#ifndef NO_SSE
#include "poly_drawer32_sse2.h"
#endif
#include "poly_drawer8.h"
void ScreenTriangle::SetupNormal(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
{
@ -702,265 +705,87 @@ void ScreenTriangle::SubsectorWrite(const TriDrawTriangleArgs *args, WorkerThrea
}
}
#if 0
float ScreenTriangle::FindGradientX(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriDraw8 =
{
float top = (c1 - c2) * (y0 - y2) - (c0 - c2) * (y1 - y2);
float bottom = (x1 - x2) * (y0 - y2) - (x0 - x2) * (y1 - y2);
return top / bottom;
}
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Copy", "opaque", false
&TriScreenDrawer8<TriScreenDrawerModes::MaskedBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AlphaBlend", "masked", false
&TriScreenDrawer8<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AddSolid", "translucent", false
&TriScreenDrawer8<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Add", "add", false
&TriScreenDrawer8<TriScreenDrawerModes::SubClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Sub", "sub", false
&TriScreenDrawer8<TriScreenDrawerModes::RevSubClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "RevSub", "revsub", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Stencil", "stencil", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Shaded", "shaded", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateCopy", "opaque", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAlphaBlend", "masked", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAdd", "add", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateSub", "sub", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateRevSub", "revsub", true
&TriScreenDrawer8<TriScreenDrawerModes::AddSrcColorBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AddSrcColorOneMinusSrcColor", "addsrccolor", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::SkycapSampler>::Execute // "Skycap", "skycap", false
};
float ScreenTriangle::FindGradientY(float x0, float y0, float x1, float y1, float x2, float y2, float c0, float c1, float c2)
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriFill8 =
{
float top = (c1 - c2) * (x0 - x2) - (c0 - c2) * (x1 - x2);
float bottom = (x0 - x2) * (y1 - y2) - (x1 - x2) * (y0 - y2);
return top / bottom;
}
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Copy", "opaque", false
&TriScreenDrawer8<TriScreenDrawerModes::MaskedBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AlphaBlend", "masked", false
&TriScreenDrawer8<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AddSolid", "translucent", false
&TriScreenDrawer8<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Add", "add", false
&TriScreenDrawer8<TriScreenDrawerModes::SubClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Sub", "sub", false
&TriScreenDrawer8<TriScreenDrawerModes::RevSubClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "RevSub", "revsub", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Stencil", "stencil", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Shaded", "shaded", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateCopy", "opaque", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAlphaBlend", "masked", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAdd", "add", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateSub", "sub", true
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateRevSub", "revsub", true
&TriScreenDrawer8<TriScreenDrawerModes::AddSrcColorBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AddSrcColorOneMinusSrcColor", "addsrccolor", false
&TriScreenDrawer8<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::FillSampler>::Execute // "Skycap", "skycap", false
};
void ScreenTriangle::Draw(const TriDrawTriangleArgs *args, WorkerThreadData *thread)
#ifdef NO_SSE
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriDraw32;
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriFill32;
#else
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriDraw32 =
{
int numSpans = thread->NumFullSpans;
auto fullSpans = thread->FullSpans;
int numBlocks = thread->NumPartialBlocks;
auto partialBlocks = thread->PartialBlocks;
int startX = thread->StartX;
int startY = thread->StartY;
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Copy", "opaque", false
&TriScreenDrawer32<TriScreenDrawerModes::MaskedBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AlphaBlend", "masked", false
&TriScreenDrawer32<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AddSolid", "translucent", false
&TriScreenDrawer32<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Add", "add", false
&TriScreenDrawer32<TriScreenDrawerModes::SubClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "Sub", "sub", false
&TriScreenDrawer32<TriScreenDrawerModes::RevSubClampBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "RevSub", "revsub", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Stencil", "stencil", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Shaded", "shaded", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateCopy", "opaque", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAlphaBlend", "masked", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAdd", "add", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateSub", "sub", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateRevSub", "revsub", true
&TriScreenDrawer32<TriScreenDrawerModes::AddSrcColorBlend, TriScreenDrawerModes::TextureSampler>::Execute, // "AddSrcColorOneMinusSrcColor", "addsrccolor", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::SkycapSampler>::Execute // "Skycap", "skycap", false
};
// Calculate gradients
const TriVertex &v1 = *args->v1;
const TriVertex &v2 = *args->v2;
const TriVertex &v3 = *args->v3;
ScreenTriangleStepVariables gradientX;
ScreenTriangleStepVariables gradientY;
ScreenTriangleStepVariables start;
gradientX.W = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
gradientY.W = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.w, v2.w, v3.w);
start.W = v1.w + gradientX.W * (startX - v1.x) + gradientY.W * (startY - v1.y);
for (int i = 0; i < TriVertex::NumVarying; i++)
std::vector<void(*)(const TriDrawTriangleArgs *, WorkerThreadData *)> ScreenTriangle::TriFill32 =
{
gradientX.Varying[i] = FindGradientX(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
gradientY.Varying[i] = FindGradientY(v1.x, v1.y, v2.x, v2.y, v3.x, v3.y, v1.varying[i] * v1.w, v2.varying[i] * v2.w, v3.varying[i] * v3.w);
start.Varying[i] = v1.varying[i] * v1.w + gradientX.Varying[i] * (startX - v1.x) + gradientY.Varying[i] * (startY - v1.y);
}
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Copy", "opaque", false
&TriScreenDrawer32<TriScreenDrawerModes::MaskedBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AlphaBlend", "masked", false
&TriScreenDrawer32<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AddSolid", "translucent", false
&TriScreenDrawer32<TriScreenDrawerModes::AddClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Add", "add", false
&TriScreenDrawer32<TriScreenDrawerModes::SubClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "Sub", "sub", false
&TriScreenDrawer32<TriScreenDrawerModes::RevSubClampBlend, TriScreenDrawerModes::FillSampler>::Execute, // "RevSub", "revsub", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Stencil", "stencil", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::ShadedSampler>::Execute, // "Shaded", "shaded", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateCopy", "opaque", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAlphaBlend", "masked", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateAdd", "add", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateSub", "sub", true
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::TranslatedSampler>::Execute, // "TranslateRevSub", "revsub", true
&TriScreenDrawer32<TriScreenDrawerModes::AddSrcColorBlend, TriScreenDrawerModes::FillSampler>::Execute, // "AddSrcColorOneMinusSrcColor", "addsrccolor", false
&TriScreenDrawer32<TriScreenDrawerModes::OpaqueBlend, TriScreenDrawerModes::FillSampler>::Execute // "Skycap", "skycap", false
};
const uint32_t * RESTRICT texPixels = (const uint32_t *)args->texturePixels;
uint32_t texWidth = args->textureWidth;
uint32_t texHeight = args->textureHeight;
uint32_t * RESTRICT destOrg = (uint32_t*)args->dest;
uint32_t * RESTRICT subsectorGBuffer = (uint32_t*)args->subsectorGBuffer;
int pitch = args->pitch;
uint32_t subsectorDepth = args->uniforms->subsectorDepth;
uint32_t light = args->uniforms->light;
float shade = (64.0f - (light * 255 / 256 + 12.0f) * 32.0f / 128.0f) / 32.0f;
float globVis = 1706.0f;
for (int i = 0; i < numSpans; i++)
{
const auto &span = fullSpans[i];
uint32_t *dest = destOrg + span.X + span.Y * pitch;
uint32_t *subsector = subsectorGBuffer + span.X + span.Y * pitch;
int width = span.Length;
int height = 8;
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (span.X - startX) + gradientY.W * (span.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (span.X - startX) + gradientY.Varying[j] * (span.Y - startY);
for (int y = 0; y < height; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * rcpW);
int lightpos = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
for (int x = 0; x < width; x++)
{
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 8;
}
int lightnext = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
int lightstep = (lightnext - lightpos) / 8;
for (int ix = 0; ix < 8; ix++)
{
int texelX = ((((uint32_t)varyingPos[0] << 8) >> 16) * texWidth) >> 16;
int texelY = ((((uint32_t)varyingPos[1] << 8) >> 16) * texHeight) >> 16;
uint32_t fg = texPixels[texelX * texHeight + texelY];
uint32_t r = RPART(fg);
uint32_t g = GPART(fg);
uint32_t b = BPART(fg);
r = r * lightpos / 256;
g = g * lightpos / 256;
b = b * lightpos / 256;
fg = 0xff000000 | (r << 16) | (g << 8) | b;
dest[x * 8 + ix] = fg;
subsector[x * 8 + ix] = subsectorDepth;
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
subsector += pitch;
}
}
for (int i = 0; i < numBlocks; i++)
{
const auto &block = partialBlocks[i];
ScreenTriangleStepVariables blockPosY;
blockPosY.W = start.W + gradientX.W * (block.X - startX) + gradientY.W * (block.Y - startY);
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] = start.Varying[j] + gradientX.Varying[j] * (block.X - startX) + gradientY.Varying[j] * (block.Y - startY);
uint32_t *dest = destOrg + block.X + block.Y * pitch;
uint32_t *subsector = subsectorGBuffer + block.X + block.Y * pitch;
uint32_t mask0 = block.Mask0;
uint32_t mask1 = block.Mask1;
for (int y = 0; y < 4; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * rcpW);
int lightpos = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 8;
}
int lightnext = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
int lightstep = (lightnext - lightpos) / 8;
for (int x = 0; x < 8; x++)
{
if (mask0 & (1 << 31))
{
int texelX = ((((uint32_t)varyingPos[0] << 8) >> 16) * texWidth) >> 16;
int texelY = ((((uint32_t)varyingPos[1] << 8) >> 16) * texHeight) >> 16;
uint32_t fg = texPixels[texelX * texHeight + texelY];
uint32_t r = RPART(fg);
uint32_t g = GPART(fg);
uint32_t b = BPART(fg);
r = r * lightpos / 256;
g = g * lightpos / 256;
b = b * lightpos / 256;
fg = 0xff000000 | (r << 16) | (g << 8) | b;
dest[x] = fg;
subsector[x] = subsectorDepth;
}
mask0 <<= 1;
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
subsector += pitch;
}
for (int y = 4; y < 8; y++)
{
ScreenTriangleStepVariables blockPosX = blockPosY;
float rcpW = 0x01000000 / blockPosX.W;
int32_t varyingPos[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] = (int32_t)(blockPosX.Varying[j] * rcpW);
int lightpos = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
blockPosX.W += gradientX.W * 8;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosX.Varying[j] += gradientX.Varying[j] * 8;
rcpW = 0x01000000 / blockPosX.W;
int32_t varyingStep[TriVertex::NumVarying];
for (int j = 0; j < TriVertex::NumVarying; j++)
{
int32_t nextPos = (int32_t)(blockPosX.Varying[j] * rcpW);
varyingStep[j] = (nextPos - varyingPos[j]) / 8;
}
int lightnext = 256 - (int)(clamp(shade - MIN(24.0f, globVis * blockPosX.W) / 32.0f, 0.0f, 31.0f / 32.0f) * 256.0f);
int lightstep = (lightnext - lightpos) / 8;
for (int x = 0; x < 8; x++)
{
if (mask1 & (1 << 31))
{
int texelX = ((((uint32_t)varyingPos[0] << 8) >> 16) * texWidth) >> 16;
int texelY = ((((uint32_t)varyingPos[1] << 8) >> 16) * texHeight) >> 16;
uint32_t fg = texPixels[texelX * texHeight + texelY];
uint32_t r = RPART(fg);
uint32_t g = GPART(fg);
uint32_t b = BPART(fg);
r = r * lightpos / 256;
g = g * lightpos / 256;
b = b * lightpos / 256;
fg = 0xff000000 | (r << 16) | (g << 8) | b;
dest[x] = fg;
subsector[x] = subsectorDepth;
}
mask1 <<= 1;
for (int j = 0; j < TriVertex::NumVarying; j++)
varyingPos[j] += varyingStep[j];
lightpos += lightstep;
}
blockPosY.W += gradientY.W;
for (int j = 0; j < TriVertex::NumVarying; j++)
blockPosY.Varying[j] += gradientY.Varying[j];
dest += pitch;
subsector += pitch;
}
}
}
#endif

26
src/polyrenderer/drawers/screen_triangle.h
View file

@ -160,3 +160,29 @@ struct ScreenTriangleStepVariables
float W;
float Varying[TriVertex::NumVarying];
};
namespace TriScreenDrawerModes
{
enum class BlendModes { Opaque, Masked, AddClamp, SubClamp, RevSubClamp, AddSrcColorOneMinusSrcColor };
struct OpaqueBlend { static const int Mode = (int)BlendModes::Opaque; };
struct MaskedBlend { static const int Mode = (int)BlendModes::Masked; };
struct AddClampBlend { static const int Mode = (int)BlendModes::AddClamp; };
struct SubClampBlend { static const int Mode = (int)BlendModes::SubClamp; };
struct RevSubClampBlend { static const int Mode = (int)BlendModes::RevSubClamp; };
struct AddSrcColorBlend { static const int Mode = (int)BlendModes::AddSrcColorOneMinusSrcColor; };
enum class FilterModes { Nearest, Linear };
struct NearestFilter { static const int Mode = (int)FilterModes::Nearest; };
struct LinearFilter { static const int Mode = (int)FilterModes::Linear; };
enum class ShadeMode { Simple, Advanced };
struct SimpleShade { static const int Mode = (int)ShadeMode::Simple; };
struct AdvancedShade { static const int Mode = (int)ShadeMode::Advanced; };
enum class Samplers { Texture, Fill, Shaded, Translated, Skycap };
struct TextureSampler { static const int Mode = (int)Samplers::Texture; };
struct FillSampler { static const int Mode = (int)Samplers::Fill; };
struct ShadedSampler { static const int Mode = (int)Samplers::Shaded; };
struct TranslatedSampler { static const int Mode = (int)Samplers::Translated; };
struct SkycapSampler { static const int Mode = (int)Samplers::Skycap; };
}

2
src/polyrenderer/scene/poly_decal.cpp
View file

@ -136,7 +136,7 @@ void RenderPolyDecal::Render(const TriMatrix &worldToClip, const Vec4f &clipPlan
PolyCameraLight *cameraLight = PolyCameraLight::Instance();
PolyDrawArgs args;
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
args.SetColormap(GetColorTable(front->Colormap));
args.SetTexture(tex, decal->Translation, true);
args.uniforms.globvis = (float)PolyRenderer::Instance()->Light.WallGlobVis(foggy);

4
src/polyrenderer/scene/poly_particle.cpp
View file

@ -81,12 +81,12 @@ void RenderPolyParticle::Render(const TriMatrix &worldToClip, const Vec4f &clipP
if (fullbrightSprite || cameraLight->FixedLightLevel() >= 0 || cameraLight->FixedColormap())
{
args.uniforms.light = 256;
args.uniforms.flags = TriUniforms::fixed_light;
args.uniforms.flags = TriUniforms::fixed_light | TriUniforms::nearest_filter;
}
else
{
args.uniforms.light = (uint32_t)((sub->sector->lightlevel + actualextralight) / 255.0f * 256.0f);
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
}
args.uniforms.subsectorDepth = subsectorDepth;

4
src/polyrenderer/scene/poly_plane.cpp
View file

@ -116,7 +116,7 @@ void RenderPolyPlane::Render3DFloor(const TriMatrix &worldToClip, const Vec4f &c
args.uniforms.light = (uint32_t)(lightlevel / 255.0f * 256.0f);
if (cameraLight->FixedLightLevel() >= 0 || cameraLight->FixedColormap())
args.uniforms.light = 256;
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
args.uniforms.subsectorDepth = subsectorDepth;
TriVertex *vertices = PolyVertexBuffer::GetVertices(sub->numlines);
@ -314,7 +314,7 @@ void RenderPolyPlane::Render(const TriMatrix &worldToClip, const Vec4f &clipPlan
args.uniforms.light = (uint32_t)(frontsector->lightlevel / 255.0f * 256.0f);
if (cameraLight->FixedLightLevel() >= 0 || cameraLight->FixedColormap())
args.uniforms.light = 256;
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
args.uniforms.subsectorDepth = isSky ? RenderPolyScene::SkySubsectorDepth : subsectorDepth;
TriVertex *vertices = PolyVertexBuffer::GetVertices(sub->numlines);

2
src/polyrenderer/scene/poly_sky.cpp
View file

@ -59,7 +59,7 @@ void PolySkyDome::Render(const TriMatrix &worldToClip)
PolyDrawArgs args;
args.uniforms.globvis = (float)PolyRenderer::Instance()->Light.WallGlobVis(foggy);
args.uniforms.light = 256;
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
args.uniforms.subsectorDepth = RenderPolyScene::SkySubsectorDepth;
args.objectToClip = &objectToClip;
args.stenciltestvalue = 255;

2
src/polyrenderer/scene/poly_sprite.cpp
View file

@ -142,7 +142,7 @@ void RenderPolySprite::Render(const TriMatrix &worldToClip, const Vec4f &clipPla
PolyDrawArgs args;
args.uniforms.globvis = (float)PolyRenderer::Instance()->Light.SpriteGlobVis(foggy);
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
if (fullbrightSprite || cameraLight->FixedLightLevel() >= 0 || cameraLight->FixedColormap())
{
args.uniforms.light = 256;

2
src/polyrenderer/scene/poly_wall.cpp
View file

@ -251,7 +251,7 @@ void RenderPolyWall::Render(const TriMatrix &worldToClip, const Vec4f &clipPlane
PolyDrawArgs args;
args.uniforms.globvis = (float)PolyRenderer::Instance()->Light.WallGlobVis(foggy);
args.uniforms.light = (uint32_t)(GetLightLevel() / 255.0f * 256.0f);
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
args.uniforms.subsectorDepth = SubsectorDepth;
args.objectToClip = &worldToClip;
args.vinput = vertices;

4
src/polyrenderer/scene/poly_wallsprite.cpp
View file

@ -106,12 +106,12 @@ void RenderPolyWallSprite::Render(const TriMatrix &worldToClip, const Vec4f &cli
if (fullbrightSprite || cameraLight->FixedLightLevel() >= 0 || cameraLight->FixedColormap())
{
args.uniforms.light = 256;
args.uniforms.flags = TriUniforms::fixed_light;
args.uniforms.flags = TriUniforms::fixed_light | TriUniforms::nearest_filter;
}
else
{
args.uniforms.light = (uint32_t)((thing->Sector->lightlevel + actualextralight) / 255.0f * 256.0f);
args.uniforms.flags = 0;
args.uniforms.flags = TriUniforms::nearest_filter;
}
args.uniforms.subsectorDepth = subsectorDepth;