WIP Metal backend for VISIGen

This commit is contained in:
Luke Street 2021-05-05 19:43:30 -04:00
parent 55438bd2ea
commit 42dba1148d
14 changed files with 1283 additions and 561 deletions

View File

@ -7,7 +7,6 @@ if (NOT MSVC)
set(CMAKE_CXX_STANDARD_REQUIRED ON) set(CMAKE_CXX_STANDARD_REQUIRED ON)
endif() endif()
add_executable(visigen add_executable(visigen
VISIBuilder.cpp VISIBuilder.cpp
VISIBuilder.hpp VISIBuilder.hpp
@ -29,17 +28,21 @@ else()
endif() endif()
if(APPLE) if(APPLE)
target_sources(visigen PRIVATE MainMac.mm) target_sources(visigen PRIVATE MainMac.mm VISIRendererMetal.mm VISIRendererMetal.hh)
set_source_files_properties(MainMac.mm PROPERTIES COMPILE_FLAGS -fobjc-arc) set_source_files_properties(MainMac.mm PROPERTIES COMPILE_FLAGS -fobjc-arc)
set_source_files_properties(VISIRendererMetal.mm PROPERTIES COMPILE_FLAGS -fobjc-arc)
find_library(METAL_LIBRARY Metal REQUIRED)
target_link_libraries(visigen PRIVATE ${METAL_LIBRARY})
elseif(WIN32) elseif(WIN32)
target_sources(visigen PRIVATE MainWin.cpp) target_sources(visigen PRIVATE
MainWin.cpp
VISIRendererOpenGL.cpp
VISIRendererOpenGL.hpp)
else() else()
target_sources(visigen PRIVATE MainXlib.cpp) target_sources(visigen PRIVATE
endif() MainXlib.cpp
VISIRendererOpenGL.cpp
if(APPLE) VISIRendererOpenGL.hpp)
find_library(OPENGL_LIBRARY OpenGL)
target_link_libraries(visigen PRIVATE ${OPENGL_LIBRARY})
endif() endif()
target_link_libraries(visigen PRIVATE target_link_libraries(visigen PRIVATE
@ -50,6 +53,8 @@ target_link_libraries(visigen PRIVATE
lzokay lzokay
xxhash xxhash
zeus zeus
png
${ZLIB_LIBRARIES} ${ZLIB_LIBRARIES}
) )
target_include_directories(visigen PRIVATE ${PNG_INCLUDE_DIR})
endif() endif()

View File

@ -1,119 +1,23 @@
#include "VISIRenderer.hpp" #include "../version.h"
#include <AppKit/AppKit.h> #include "VISIRendererMetal.hh"
#include "athena/Global.hpp" #include "athena/Global.hpp"
#include "logvisor/logvisor.hpp" #include "logvisor/logvisor.hpp"
#include "../version.h" #include <AppKit/AppKit.h>
#include <MetalKit/MetalKit.h>
#include <thread> #include <thread>
#if !__has_feature(objc_arc) #if !__has_feature(objc_arc)
#error ARC Required #error ARC Required
#endif #endif
static std::thread s_task;
static const NSOpenGLPixelFormatAttribute PF_RGBA8_Z24_ATTRS[] =
{
NSOpenGLPFAAccelerated,
NSOpenGLPFAOpenGLProfile, NSOpenGLProfileVersion3_2Core,
//NSOpenGLPFADoubleBuffer,
NSOpenGLPFAColorSize, 24,
NSOpenGLPFAAlphaSize, 8,
NSOpenGLPFADepthSize, 24,
0, 0
};
@interface OpenGLView : NSOpenGLView
{
VISIRenderer* m_renderer;
}
- (id)initWithFrame:(NSRect)frame renderer:(VISIRenderer*)renderer;
@end
static NSWindow* s_Window;
static void UpdatePercent(float percent)
{
dispatch_async(dispatch_get_main_queue(), ^{
s_Window.title = [NSString stringWithFormat:@"VISIGen [%g%%]", percent * 100.f];
});
}
@implementation OpenGLView
- (id)initWithFrame:(NSRect)frame renderer:(VISIRenderer*)renderer;
{
NSOpenGLPixelFormat* pf = [[NSOpenGLPixelFormat alloc] initWithAttributes:PF_RGBA8_Z24_ATTRS];
self = [super initWithFrame:frame pixelFormat:pf];
m_renderer = renderer;
return self;
}
- (void)prepareOpenGL
{
[super prepareOpenGL];
s_task = std::thread([self](){
[[self openGLContext] makeCurrentContext];
m_renderer->Run(UpdatePercent);
[NSApp terminate:nil];
});
}
@end
@interface AppDelegate : NSObject <NSApplicationDelegate>
{
VISIRenderer* m_renderer;
NSWindow* m_window;
NSOpenGLView* m_glView;
int m_instanceIdx;
}
- (id)initWithRenderer:(VISIRenderer*)renderer instIdx:(int)instIdx;
@end
@implementation AppDelegate
- (id)initWithRenderer:(VISIRenderer*)renderer instIdx:(int)instIdx
{
self = [super init];
m_renderer = renderer;
m_instanceIdx = instIdx;
return self;
}
- (void)applicationDidFinishLaunching:(NSNotification*)notification
{
int x = 0;
int y = 0;
if (m_instanceIdx != -1)
{
x = (m_instanceIdx & 1) != 0;
y = (m_instanceIdx & 2) != 0;
}
NSRect cRect = NSMakeRect(x * 768, y * 534, 768, 512);
m_window = [[NSWindow alloc] initWithContentRect:cRect
styleMask:NSWindowStyleMaskTitled | NSWindowStyleMaskMiniaturizable
backing:NSBackingStoreBuffered
defer:NO];
m_window.releasedWhenClosed = NO;
m_window.title = @"VISIGen";
s_Window = m_window;
m_glView = [[OpenGLView alloc] initWithFrame:cRect renderer:m_renderer];
m_window.contentView = m_glView;
[m_window makeKeyAndOrderFront:nil];
}
- (void)applicationWillTerminate:(NSNotification*)notification
{
m_renderer->Terminate();
if (s_task.joinable())
s_task.join();
exit(m_renderer->ReturnVal());
}
@end
static logvisor::Module AthenaLog("Athena"); static logvisor::Module AthenaLog("Athena");
static void AthenaExc(athena::error::Level level, const char * /*file*/, const char *, int /*line*/, static void AthenaExc(athena::error::Level level, const char * /*file*/, const char *, int /*line*/,
fmt::string_view fmt, fmt::format_args args) { fmt::string_view fmt, fmt::format_args args) {
AthenaLog.vreport(logvisor::Level(level), fmt, args); AthenaLog.vreport(logvisor::Level(level), fmt, args);
} }
int main(int argc, const char** argv) int main(int argc, const char **argv) {
{ if (argc > 1 && !strcmp(argv[1], "--dlpackage")) {
if (argc > 1 && !strcmp(argv[1], "--dlpackage"))
{
fmt::print(FMT_STRING("{}\n"), METAFORCE_DLPACKAGE); fmt::print(FMT_STRING("{}\n"), METAFORCE_DLPACKAGE);
return 100; return 100;
} }
@ -121,18 +25,9 @@ int main(int argc, const char** argv)
logvisor::RegisterStandardExceptions(); logvisor::RegisterStandardExceptions();
logvisor::RegisterConsoleLogger(); logvisor::RegisterConsoleLogger();
atSetExceptionHandler(AthenaExc); atSetExceptionHandler(AthenaExc);
VISIRenderer renderer(argc, argv); VISIRendererMetal renderer(argc, argv);
int instIdx = -1; @autoreleasepool {
if (argc > 3) renderer.Run(nullptr);
instIdx = atoi(argv[3]);
@autoreleasepool
{
[[NSApplication sharedApplication] setActivationPolicy:NSApplicationActivationPolicyRegular];
/* Delegate (OS X callbacks) */
AppDelegate* appDelegate = [[AppDelegate alloc] initWithRenderer:&renderer instIdx:instIdx];
[[NSApplication sharedApplication] setDelegate:appDelegate];
[[NSApplication sharedApplication] run];
} }
return renderer.ReturnVal(); return renderer.ReturnVal();
} }

View File

@ -1,4 +1,4 @@
#include "VISIRenderer.hpp" #include "VISIRendererOpenGL.hpp"
#include <Windows.h> #include <Windows.h>
#include <WinUser.h> #include <WinUser.h>
#include <Shlwapi.h> #include <Shlwapi.h>
@ -49,7 +49,7 @@ int wmain(int argc, const hecl::SystemChar** argv) {
logvisor::RegisterStandardExceptions(); logvisor::RegisterStandardExceptions();
logvisor::RegisterConsoleLogger(); logvisor::RegisterConsoleLogger();
atSetExceptionHandler(AthenaExc); atSetExceptionHandler(AthenaExc);
VISIRenderer renderer(argc, argv); VISIRendererOpenGL renderer(argc, argv);
s_Renderer = &renderer; s_Renderer = &renderer;
int instIdx = -1; int instIdx = -1;

View File

@ -1,4 +1,4 @@
#include "VISIRenderer.hpp" #include "VISIRendererOpenGL.hpp"
#include <X11/Xlib.h> #include <X11/Xlib.h>
#include <X11/Xatom.h> #include <X11/Xatom.h>
#include <GL/glx.h> #include <GL/glx.h>
@ -86,7 +86,7 @@ int main(int argc, const char** argv) {
logvisor::RegisterStandardExceptions(); logvisor::RegisterStandardExceptions();
logvisor::RegisterConsoleLogger(); logvisor::RegisterConsoleLogger();
atSetExceptionHandler(AthenaExc); atSetExceptionHandler(AthenaExc);
VISIRenderer renderer(argc, argv); VISIRendererOpenGL renderer(argc, argv);
if (!XInitThreads()) { if (!XInitThreads()) {
Log.report(logvisor::Error, FMT_STRING("X doesn't support multithreading")); Log.report(logvisor::Error, FMT_STRING("X doesn't support multithreading"));

68
visigen/Shader.metal Normal file
View File

@ -0,0 +1,68 @@
#include <metal_stdlib>
#include <simd/simd.h>
#include "ShaderTypes.h"
using namespace metal;
/*static const matrix_float4x4 LookMATs[] = {
{// Forward
{1.f, 0.f, 0.f, 0.f},
{0.f, 0.f, 1.f, 0.f},
{0.f, -1.f, 0.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
{// Backward
{-1.f, 0.f, 0.f, 0.f},
{0.f, 0.f, 1.f, 0.f},
{0.f, 1.f, 0.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
{// Up
{1.f, 0.f, 0.f, 0.f},
{0.f, -1.f, 0.f, 0.f},
{0.f, 0.f, -1.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
{// Down
{1.f, 0.f, 0.f, 0.f},
{0.f, 1.f, 0.f, 0.f},
{0.f, 0.f, 1.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
{// Left
{0.f, 1.f, 0.f, 0.f},
{0.f, 0.f, 1.f, 0.f},
{1.f, 0.f, 0.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
{// Right
{0.f, -1.f, 0.f, 0.f},
{0.f, 0.f, 1.f, 0.f},
{-1.f, 0.f, 0.f, 0.f},
{0.f, 0.f, 0.f, 1.f}},
};*/
typedef struct
{
float4 position [[position]];
float4 color;
} ColorInOut;
typedef struct
{
float3 position [[attribute(VertexAttributePosition)]];
float4 color [[attribute(VertexAttributeColor)]];
} Vertex;
vertex ColorInOut vertexShader(Vertex in [[stage_in]], constant Uniforms& uniforms [[buffer(BufferIndexUniforms)]])
{
ColorInOut out;
float4 position = float4(in.position, 1.0);
position.y *= -1.f;
out.position = uniforms.projectionMatrix * uniforms.modelViewMatrix * position;
out.color = in.color;
return out;
}
fragment float4 fragmentShader(ColorInOut in [[stage_in]])
{
return in.color;
}

32
visigen/ShaderTypes.h Normal file
View File

@ -0,0 +1,32 @@
#ifndef ShaderTypes_h
#define ShaderTypes_h
#ifdef __METAL_VERSION__
#define NS_ENUM(_type, _name) \
enum _name : _type _name; \
enum _name : _type
#define NSInteger metal::int32_t
#else
#import <Foundation/Foundation.h>
#endif
#include <simd/simd.h>
typedef NS_ENUM(NSInteger, BufferIndex) {
BufferIndexVertex = 0,
BufferIndexUniforms = 1,
};
typedef NS_ENUM(NSInteger, VertexAttribute) {
VertexAttributePosition = 0,
VertexAttributeColor = 1,
};
typedef struct Uniforms {
matrix_float4x4 projectionMatrix;
matrix_float4x4 modelViewMatrix;
} Uniforms;
#endif /* ShaderTypes_h */

View File

@ -1,6 +1,9 @@
#include "VISIBuilder.hpp" #include "VISIBuilder.hpp"
#include "logvisor/logvisor.hpp" #include "logvisor/logvisor.hpp"
#include <fstream>
#include <iostream>
#ifndef _WIN32 #ifndef _WIN32
#include <unistd.h> #include <unistd.h>
#include <signal.h> #include <signal.h>
@ -15,16 +18,29 @@ VISIBuilder::PVSRenderCache::PVSRenderCache(VISIRenderer& renderer) : m_renderer
static std::unique_ptr<VISIRenderer::RGBA8[]> RGBABuf(new VISIRenderer::RGBA8[256 * 256 * 6]); static std::unique_ptr<VISIRenderer::RGBA8[]> RGBABuf(new VISIRenderer::RGBA8[256 * 256 * 6]);
size_t VISIBuilder::m_frame = 0;
const VISIBuilder::Leaf& VISIBuilder::PVSRenderCache::GetLeaf(const zeus::CVector3f& vec) { const VISIBuilder::Leaf& VISIBuilder::PVSRenderCache::GetLeaf(const zeus::CVector3f& vec) {
auto search = m_cache.find(vec); auto search = m_cache.find(vec);
if (search != m_cache.cend()) { if (search != m_cache.cend()) {
// Log.report(logvisor::Info, FMT_STRING("Cache hit"));
return *search->second; return *search->second;
} }
// Log.report(logvisor::Info, FMT_STRING("Rendering")); m_renderer.SetupRenderPass(vec);
bool needsTransparent = false; bool needsTransparent = false;
m_renderer.RenderPVSOpaque(RGBABuf.get(), vec, needsTransparent); m_renderer.RenderPVSOpaque(RGBABuf.get(), needsTransparent);
// size_t outsize;
// auto* buf = VISIRenderer::makePNGBuffer(reinterpret_cast<unsigned char*>(RGBABuf.get()), 768, 512, &outsize);
// auto filename = fmt::format(FMT_STRING("outx{}.png"), m_frame++);
// std::cout << "Rendering " << filename << std::endl;
// std::ofstream fout;
// fout.open(filename, std::ios::binary | std::ios::out);
// fout.write(static_cast<const char*>(buf), outsize);
// fout.close();
// free(buf);
std::unique_ptr<Leaf> leafOut = std::make_unique<Leaf>(); std::unique_ptr<Leaf> leafOut = std::make_unique<Leaf>();
for (unsigned i = 0; i < 768 * 512; ++i) { for (unsigned i = 0; i < 768 * 512; ++i) {
const VISIRenderer::RGBA8& pixel = RGBABuf[i]; const VISIRenderer::RGBA8& pixel = RGBABuf[i];
@ -39,18 +55,20 @@ const VISIBuilder::Leaf& VISIBuilder::PVSRenderCache::GetLeaf(const zeus::CVecto
leafOut->setLightEnum(m_lightMetaBit + idx * 2, state); leafOut->setLightEnum(m_lightMetaBit + idx * 2, state);
}; };
if (needsTransparent) if (needsTransparent)
m_renderer.RenderPVSTransparent(setBitLambda, vec); m_renderer.RenderPVSTransparent(setBitLambda);
m_renderer.RenderPVSEntitiesAndLights(setBitLambda, setLightLambda, vec); m_renderer.RenderPVSEntitiesAndLights(setBitLambda, setLightLambda);
return *m_cache.emplace(std::make_pair(vec, std::move(leafOut))).first->second; return *m_cache.emplace(std::make_pair(vec, std::move(leafOut))).first->second;
} }
void VISIBuilder::Progress::report(int divisions) { void VISIBuilder::Progress::report(int divisions) {
m_prog += 1.f / divisions; m_prog += 1.f / divisions;
// printf(" %g%% \r", m_prog * 100.f); if (m_updatePercent != nullptr) {
// fflush(stdout);
if (m_updatePercent)
m_updatePercent(m_prog); m_updatePercent(m_prog);
} else {
printf(" %g%% \r", m_prog * 100.f);
fflush(stdout);
}
} }
void VISIBuilder::Node::buildChildren(int level, int divisions, const zeus::CAABox& curAabb, PVSRenderCache& rc, void VISIBuilder::Node::buildChildren(int level, int divisions, const zeus::CAABox& curAabb, PVSRenderCache& rc,

View File

@ -1,9 +1,9 @@
#pragma once #pragma once
#include "VISIRenderer.hpp" #include "VISIRenderer.hpp"
#include "zeus/CAABox.hpp"
#include "xxhash/xxhash.h"
#include "athena/MemoryWriter.hpp" #include "athena/MemoryWriter.hpp"
#include "xxhash/xxhash.h"
#include "zeus/CAABox.hpp"
#include <unordered_map> #include <unordered_map>
#ifdef _WIN32 #ifdef _WIN32
@ -23,6 +23,8 @@ struct hash<zeus::CVector3f> {
} // namespace std } // namespace std
struct VISIBuilder { struct VISIBuilder {
static size_t m_frame;
struct Leaf { struct Leaf {
std::vector<uint8_t> bits; std::vector<uint8_t> bits;
void setBit(size_t bit) { void setBit(size_t bit) {

View File

@ -1,112 +1,144 @@
#include "VISIRenderer.hpp" #include "VISIRenderer.hpp"
#include "athena/FileReader.hpp"
#include "zeus/CAABox.hpp"
#include "VISIBuilder.hpp" #include "VISIBuilder.hpp"
#include "zeus/CFrustum.hpp" #include "athena/FileReader.hpp"
#include "logvisor/logvisor.hpp" #include "logvisor/logvisor.hpp"
#include <png.h>
static logvisor::Module Log("visigen"); static logvisor::Module Log("visigen");
static const char* VS = /* structure to store PNG image bytes */
"#version 330\n" struct mem_encode
"layout(location=0) in vec4 posIn;\n" {
"layout(location=1) in vec4 colorIn;\n" char *buffer;
"\n" size_t size;
"uniform UniformBlock\n" };
"{\n"
" mat4 xf;\n"
"};\n"
"\n"
"struct VertToFrag\n"
"{\n"
" vec4 color;\n"
"};\n"
"\n"
"out VertToFrag vtf;\n"
"void main()\n"
"{\n"
" vtf.color = colorIn;\n"
" gl_Position = xf * vec4(posIn.xyz, 1.0);\n"
"}\n";
static const char* FS = static void
"#version 330\n" my_png_write_data(png_structp png_ptr, png_bytep data, png_size_t length)
"struct VertToFrag\n" {
"{\n" /* with libpng15 next line causes pointer deference error; use libpng12 */
" vec4 color;\n" struct mem_encode* p=(struct mem_encode*)png_get_io_ptr(png_ptr); /* was png_ptr->io_ptr */
"};\n" size_t nsize = p->size + length;
"\n"
"in VertToFrag vtf;\n"
"layout(location=0) out vec4 colorOut;\n"
"void main()\n"
"{\n"
" colorOut = vtf.color;\n"
"}\n";
static const uint32_t AABBIdxs[20] = {0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 1, 7, 3, 5, 5, 0, 0, 2, 6, 4}; /* allocate or grow buffer */
if(p->buffer)
p->buffer = (char *) realloc(p->buffer, nsize);
else
p->buffer = (char *) malloc(nsize);
bool VISIRenderer::SetupShaders() { if(!p->buffer)
m_vtxShader = glCreateShader(GL_VERTEX_SHADER); png_error(png_ptr, "Write Error");
m_fragShader = glCreateShader(GL_FRAGMENT_SHADER);
m_program = glCreateProgram();
glShaderSource(m_vtxShader, 1, &VS, nullptr); /* copy new bytes to end of buffer */
glCompileShader(m_vtxShader); memcpy(p->buffer + p->size, data, length);
GLint status; p->size += length;
glGetShaderiv(m_vtxShader, GL_COMPILE_STATUS, &status);
if (status != GL_TRUE) {
GLint logLen;
glGetShaderiv(m_vtxShader, GL_INFO_LOG_LENGTH, &logLen);
char* log = (char*)malloc(logLen);
glGetShaderInfoLog(m_vtxShader, logLen, nullptr, log);
Log.report(logvisor::Error, FMT_STRING("unable to compile vert source\n{}\n{}\n"), log, VS);
free(log);
return false;
} }
glShaderSource(m_fragShader, 1, &FS, nullptr); /*
glCompileShader(m_fragShader); write an rgba image to a memory buffer in PNG format, without any fanciness.
glGetShaderiv(m_fragShader, GL_COMPILE_STATUS, &status);
if (status != GL_TRUE) { Params: rgba - the rgba values
GLint logLen; width - image width
glGetShaderiv(m_fragShader, GL_INFO_LOG_LENGTH, &logLen); height - image height
char* log = (char*)malloc(logLen); outsize - return for size of output buffer
glGetShaderInfoLog(m_fragShader, logLen, nullptr, log); Returns: pointer to allocated buffer holding png data
Log.report(logvisor::Error, FMT_STRING("unable to compile frag source\n{}\n{}\n"), log, FS); */
free(log); void *VISIRenderer::makePNGBuffer(unsigned char *rgba, int width, int height, size_t *outsize)
return false; {
int code = 0;
// FILE *fp;
png_structp png_ptr = 0;
png_infop info_ptr =0;
png_bytep row = 0;
struct mem_encode state = {0, 0};
*outsize = 0;
// Initialize write structure
png_ptr = png_create_write_struct(PNG_LIBPNG_VER_STRING, NULL, NULL, NULL);
if (png_ptr == NULL) {
fprintf(stderr, "Could not allocate write struct\n");
code = 1;
goto finalise;
} }
glAttachShader(m_program, m_vtxShader); // Initialize info structure
glAttachShader(m_program, m_fragShader); info_ptr = png_create_info_struct(png_ptr);
if (info_ptr == NULL) {
glLinkProgram(m_program); fprintf(stderr, "Could not allocate info struct\n");
glGetProgramiv(m_program, GL_LINK_STATUS, &status); code = 1;
if (status != GL_TRUE) { goto finalise;
GLint logLen;
glGetProgramiv(m_program, GL_INFO_LOG_LENGTH, &logLen);
char* log = (char*)malloc(logLen);
glGetProgramInfoLog(m_program, logLen, nullptr, log);
Log.report(logvisor::Error, FMT_STRING("unable to link shader program\n{}\n"), log);
free(log);
return false;
} }
glUseProgram(m_program); // Setup Exception handling
m_uniLoc = glGetUniformBlockIndex(m_program, "UniformBlock"); if (setjmp(png_jmpbuf(png_ptr))) {
fprintf(stderr, "Error during png creation\n");
code = 1;
goto finalise;
}
glGenBuffers(1, &m_uniformBufferGL); // png_init_io(png_ptr, fp);
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformBuffer), nullptr, GL_DYNAMIC_DRAW);
glGenBuffers(1, &m_aabbIBO); /* if my_png_flush() is not needed, change the arg to NULL */
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_aabbIBO); png_set_write_fn(png_ptr, &state, my_png_write_data, NULL);
glBufferData(GL_ELEMENT_ARRAY_BUFFER, 20 * 4, AABBIdxs, GL_STATIC_DRAW);
glEnable(GL_PRIMITIVE_RESTART); // Write header (8 bit colour depth)
glPrimitiveRestartIndex(0xffffffff); png_set_IHDR(png_ptr, info_ptr, width, height,
8, PNG_COLOR_TYPE_RGB_ALPHA, PNG_INTERLACE_NONE,
PNG_COMPRESSION_TYPE_BASE, PNG_FILTER_TYPE_BASE);
return true; // Set title
/*
if (title != NULL) {
png_text title_text;
title_text.compression = PNG_TEXT_COMPRESSION_NONE;
title_text.key = "Title";
title_text.text = title;
png_set_text(png_ptr, info_ptr, &title_text, 1);
}
*/
png_write_info(png_ptr, info_ptr);
// Allocate memory for one row (4 bytes per pixel - RGBA)
row = (png_bytep) malloc(4 * width * sizeof(png_byte));
// Write image data
int x, y;
for (y=0 ; y<height ; y++) {
for (x=0 ; x<width ; x++) {
// setRGB(&(row[x*3]), buffer[y*width + x]);
row[x*4] = rgba[(y*width +x)*4];
row[x*4+1] = rgba[(y*width +x)*4+1];
row[x*4+2] = rgba[(y*width +x)*4+2];
row[x*4+3] = rgba[(y*width +x)*4+3];
// row[x*4] = 255;
// row[x*4+1] = 0;
// row[x*4+2] = 255;
// row[x*4+3] = 255;
}
png_write_row(png_ptr, row);
}
// End write
png_write_end(png_ptr, NULL);
finalise:
// if (fp != NULL) fclose(fp);
if (info_ptr != NULL) png_free_data(png_ptr, info_ptr, PNG_FREE_ALL, -1);
if (png_ptr != NULL) png_destroy_write_struct(&png_ptr, (png_infopp)NULL);
if (row != NULL) free(row);
*outsize = state.size;
return state.buffer;
}
zeus::CColor VISIRenderer::ColorForIndex(uint32_t i) {
i += 1;
return zeus::CColor((i & 0xff) / 255.f, ((i >> 8) & 0xff) / 255.f, ((i >> 16) & 0xff) / 255.f, 1.f);
} }
std::vector<VISIRenderer::Model::Vert> VISIRenderer::AABBToVerts(const zeus::CAABox& aabb, const zeus::CColor& color) { std::vector<VISIRenderer::Model::Vert> VISIRenderer::AABBToVerts(const zeus::CAABox& aabb, const zeus::CColor& color) {
@ -128,302 +160,8 @@ std::vector<VISIRenderer::Model::Vert> VISIRenderer::AABBToVerts(const zeus::CAA
return verts; return verts;
} }
static zeus::CColor ColorForIndex(int i) {
i += 1;
return zeus::CColor((i & 0xff) / 255.f, ((i >> 8) & 0xff) / 255.f, ((i >> 16) & 0xff) / 255.f, 1.f);
}
bool VISIRenderer::SetupVertexBuffersAndFormats() {
for (Model& model : m_models) {
glGenVertexArrays(1, &model.vao);
glGenBuffers(1, &model.vbo);
glGenBuffers(1, &model.ibo);
glBindVertexArray(model.vao);
glBindBuffer(GL_ARRAY_BUFFER, model.vbo);
glBufferData(GL_ARRAY_BUFFER, model.verts.size() * sizeof(Model::Vert), model.verts.data(), GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, model.ibo);
glBufferData(GL_ELEMENT_ARRAY_BUFFER, model.idxs.size() * 4, model.idxs.data(), GL_STATIC_DRAW);
}
int idx = m_models.size();
for (Entity& ent : m_entities) {
glGenVertexArrays(1, &ent.vao);
glGenBuffers(1, &ent.vbo);
glBindVertexArray(ent.vao);
auto verts = AABBToVerts(ent.aabb, ColorForIndex(idx++));
glBindBuffer(GL_ARRAY_BUFFER, ent.vbo);
glBufferData(GL_ARRAY_BUFFER, verts.size() * sizeof(Model::Vert), verts.data(), GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_aabbIBO);
}
for (Light& light : m_lights) {
glGenVertexArrays(1, &light.vao);
glGenBuffers(1, &light.vbo);
glBindVertexArray(light.vao);
Model::Vert vert;
vert.pos = light.point;
vert.color = ColorForIndex(idx++);
glBindBuffer(GL_ARRAY_BUFFER, light.vbo);
glBufferData(GL_ARRAY_BUFFER, sizeof(Model::Vert), &vert, GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
}
m_queryCount = m_models.size() + m_entities.size() + m_lights.size();
m_queries.reset(new GLuint[m_queryCount]);
m_queryBools.reset(new bool[m_queryCount]);
glGenQueries(GLsizei(m_queryCount), m_queries.get());
return true;
}
static zeus::CMatrix4f g_Proj;
static void CalculateProjMatrix() {
float znear = 0.2f;
float zfar = 1000.f;
float tfov = std::tan(zeus::degToRad(90.f * 0.5f));
float top = znear * tfov;
float bottom = -top;
float right = znear * tfov;
float left = -right;
float rml = right - left;
float rpl = right + left;
float tmb = top - bottom;
float tpb = top + bottom;
float fpn = zfar + znear;
float fmn = zfar - znear;
g_Proj = zeus::CMatrix4f(2.f * znear / rml, 0.f, rpl / rml, 0.f, 0.f, 2.f * znear / tmb, tpb / tmb, 0.f, 0.f, 0.f,
-fpn / fmn, -2.f * zfar * znear / fmn, 0.f, 0.f, -1.f, 0.f);
}
static const zeus::CMatrix4f LookMATs[] = {
{// Forward
1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Backward
-1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Up
1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Down
1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Left
0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Right
0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
};
void VISIRenderer::RenderPVSOpaque(RGBA8* bufOut, const zeus::CVector3f& pos, bool& needTransparent) {
glViewport(0, 0, 768, 512);
glEnable(GL_CULL_FACE);
glDepthMask(GL_TRUE);
glDepthFunc(GL_LEQUAL);
glEnable(GL_DEPTH_TEST);
glClearColor(0.f, 0.f, 0.f, 1.f);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
zeus::CMatrix4f mv = LookMATs[j] * zeus::CTransform::Translate(-pos).toMatrix4f();
m_uniformBuffer.m_xf = g_Proj * mv;
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformBuffer), &m_uniformBuffer, GL_DYNAMIC_DRAW);
glUniformBlockBinding(m_program, m_uniLoc, 0);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, 0, sizeof(UniformBuffer));
zeus::CFrustum frustum;
frustum.updatePlanes(mv, g_Proj);
// Draw frontfaces
glCullFace(GL_BACK);
glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
for (const Model& model : m_models) {
if (!frustum.aabbFrustumTest(model.aabb))
continue;
glBindVertexArray(model.vao);
for (const Model::Surface& surf : model.surfaces) {
// Non-transparents first
if (!surf.transparent)
glDrawElements(model.topology, surf.count, GL_UNSIGNED_INT,
reinterpret_cast<void*>(uintptr_t(surf.first * 4)));
}
}
}
// m_swapFunc();
glFinish();
glReadPixels(0, 0, 768, 512, GL_RGBA, GL_UNSIGNED_BYTE, (GLvoid*)bufOut);
}
void VISIRenderer::RenderPVSTransparent(const std::function<void(int)>& passFunc, const zeus::CVector3f& pos) {
glDepthMask(GL_FALSE);
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
zeus::CMatrix4f mv = LookMATs[j] * zeus::CTransform::Translate(-pos).toMatrix4f();
m_uniformBuffer.m_xf = g_Proj * mv;
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformBuffer), &m_uniformBuffer, GL_DYNAMIC_DRAW);
glUniformBlockBinding(m_program, m_uniLoc, 0);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, 0, sizeof(UniformBuffer));
zeus::CFrustum frustum;
frustum.updatePlanes(mv, g_Proj);
memset(m_queryBools.get(), 0, m_queryCount);
int idx = 0;
for (const Model& model : m_models) {
if (!frustum.aabbFrustumTest(model.aabb)) {
++idx;
continue;
}
glBindVertexArray(model.vao);
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
m_queryBools[idx] = true;
for (const Model::Surface& surf : model.surfaces) {
// transparents
if (surf.transparent)
glDrawElements(model.topology, surf.count, GL_UNSIGNED_INT,
reinterpret_cast<void*>(uintptr_t(surf.first * 4)));
}
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
++idx;
}
for (int i = 0; i < idx; ++i) {
if (m_queryBools[i]) {
GLint res;
glGetQueryObjectiv(m_queries[i], GL_QUERY_RESULT, &res);
if (res)
passFunc(i);
}
}
}
}
void VISIRenderer::RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc,
const std::function<void(int, EPVSVisSetState)>& lightPassFunc,
const zeus::CVector3f& pos) {
glDepthMask(GL_FALSE);
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
zeus::CMatrix4f mv = LookMATs[j] * zeus::CTransform::Translate(-pos).toMatrix4f();
m_uniformBuffer.m_xf = g_Proj * mv;
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
glBufferData(GL_UNIFORM_BUFFER, sizeof(UniformBuffer), &m_uniformBuffer, GL_DYNAMIC_DRAW);
glUniformBlockBinding(m_program, m_uniLoc, 0);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, 0, sizeof(UniformBuffer));
zeus::CFrustum frustum;
frustum.updatePlanes(mv, g_Proj);
memset(m_queryBools.get(), 0, m_queryCount);
int idx = m_models.size();
for (const Entity& ent : m_entities) {
if (!frustum.aabbFrustumTest(ent.aabb)) {
++idx;
continue;
}
glBindVertexArray(ent.vao);
m_queryBools[idx] = true;
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
glDrawElements(GL_TRIANGLE_STRIP, 20, GL_UNSIGNED_INT, 0);
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
++idx;
}
for (const Light& light : m_lights) {
if (!frustum.pointFrustumTest(light.point)) {
++idx;
continue;
}
glBindVertexArray(light.vao);
m_queryBools[idx] = true;
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
glDrawArrays(GL_POINTS, 0, 1);
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
++idx;
}
idx = m_models.size();
for (const Entity& ent : m_entities) {
(void)ent;
if (m_queryBools[idx]) {
GLint res;
glGetQueryObjectiv(m_queries[idx], GL_QUERY_RESULT, &res);
if (res)
passFunc(idx);
}
++idx;
}
int lightIdx = 0;
for (const Light& light : m_lights) {
if (m_queryBools[idx]) {
GLint res;
glGetQueryObjectiv(m_queries[idx], GL_QUERY_RESULT, &res);
EPVSVisSetState state =
m_totalAABB.pointInside(light.point) ? EPVSVisSetState::EndOfTree : EPVSVisSetState::OutOfBounds;
if (res && state == EPVSVisSetState::EndOfTree)
state = EPVSVisSetState::NodeFound;
lightPassFunc(lightIdx, state);
}
++lightIdx;
++idx;
}
}
}
void VISIRenderer::Run(FPercent updatePercent) { void VISIRenderer::Run(FPercent updatePercent) {
m_updatePercent = updatePercent; m_updatePercent = updatePercent;
CalculateProjMatrix();
if (glewInit() != GLEW_OK) {
Log.report(logvisor::Error, FMT_STRING("unable to init glew"));
m_return = 1;
return;
}
if (!GLEW_ARB_occlusion_query2) {
Log.report(logvisor::Error, FMT_STRING("GL_ARB_occlusion_query2 extension not present"));
m_return = 1;
return;
}
if (!SetupShaders()) { if (!SetupShaders()) {
m_return = 1; m_return = 1;
@ -456,7 +194,7 @@ void VISIRenderer::Run(FPercent updatePercent) {
zeus::CColor color = ColorForIndex(i); zeus::CColor color = ColorForIndex(i);
Model& model = m_models[i]; Model& model = m_models[i];
uint32_t topology = r.readUint32Big(); uint32_t topology = r.readUint32Big();
model.topology = topology ? GL_TRIANGLE_STRIP : GL_TRIANGLES; model.topology = static_cast<hecl::HMDLTopology>(topology);
uint32_t vertCount = r.readUint32Big(); uint32_t vertCount = r.readUint32Big();
model.verts.reserve(vertCount); model.verts.reserve(vertCount);

View File

@ -1,7 +1,7 @@
#pragma once #pragma once
#include "boo/graphicsdev/glew.h"
#include "hecl/SystemChar.hpp" #include "hecl/SystemChar.hpp"
#include "hecl/HMDLMeta.hpp"
#include "zeus/CColor.hpp" #include "zeus/CColor.hpp"
#include "zeus/CMatrix4f.hpp" #include "zeus/CMatrix4f.hpp"
#include "zeus/CAABox.hpp" #include "zeus/CAABox.hpp"
@ -13,18 +13,9 @@ enum class EPVSVisSetState { EndOfTree, NodeFound, OutOfBounds };
class VISIRenderer { class VISIRenderer {
friend struct VISIBuilder; friend struct VISIBuilder;
int m_argc; public:
const hecl::SystemChar** m_argv;
int m_return = 0;
zeus::CAABox m_totalAABB;
struct UniformBuffer {
zeus::CMatrix4f m_xf;
} m_uniformBuffer;
struct Model { struct Model {
GLenum topology; hecl::HMDLTopology topology;
zeus::CAABox aabb; zeus::CAABox aabb;
struct Vert { struct Vert {
@ -34,7 +25,6 @@ class VISIRenderer {
std::vector<Vert> verts; std::vector<Vert> verts;
std::vector<uint32_t> idxs; std::vector<uint32_t> idxs;
GLuint vbo, ibo, vao;
struct Surface { struct Surface {
uint32_t first; uint32_t first;
@ -47,32 +37,29 @@ class VISIRenderer {
struct Entity { struct Entity {
uint32_t entityId; uint32_t entityId;
zeus::CAABox aabb; zeus::CAABox aabb;
GLuint vbo, vao;
}; };
struct Light { struct Light {
zeus::CVector3f point; zeus::CVector3f point;
GLuint vbo, vao;
}; };
GLuint m_vtxShader, m_fragShader, m_program, m_uniLoc; protected:
GLuint m_uniformBufferGL; int m_argc;
GLuint m_aabbIBO; const hecl::SystemChar** m_argv;
bool SetupShaders(); int m_return = 0;
zeus::CAABox m_totalAABB;
virtual bool SetupShaders() = 0;
std::vector<Model> m_models; std::vector<Model> m_models;
std::vector<Entity> m_entities; std::vector<Entity> m_entities;
std::vector<Light> m_lights; std::vector<Light> m_lights;
bool SetupVertexBuffersAndFormats(); virtual bool SetupVertexBuffersAndFormats() = 0;
virtual void SetupRenderPass(const zeus::CVector3f& pos) = 0;
size_t m_queryCount;
std::unique_ptr<GLuint[]> m_queries;
std::unique_ptr<bool[]> m_queryBools;
FPercent m_updatePercent; FPercent m_updatePercent;
static std::vector<Model::Vert> AABBToVerts(const zeus::CAABox& aabb, const zeus::CColor& color);
public: public:
bool m_terminate = false; bool m_terminate = false;
struct RGBA8 { struct RGBA8 {
@ -85,10 +72,13 @@ public:
VISIRenderer(int argc, const hecl::SystemChar** argv) : m_argc(argc), m_argv(argv) {} VISIRenderer(int argc, const hecl::SystemChar** argv) : m_argc(argc), m_argv(argv) {}
void Run(FPercent updatePercent); void Run(FPercent updatePercent);
void Terminate(); void Terminate();
void RenderPVSOpaque(RGBA8* bufOut, const zeus::CVector3f& pos, bool& needTransparent); virtual void RenderPVSOpaque(RGBA8* bufOut, bool& needTransparent) = 0;
void RenderPVSTransparent(const std::function<void(int)>& passFunc, const zeus::CVector3f& pos); virtual void RenderPVSTransparent(const std::function<void(int)>& passFunc) = 0;
void RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc, virtual void RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc,
const std::function<void(int, EPVSVisSetState)>& lightPassFunc, const std::function<void(int, EPVSVisSetState)>& lightPassFunc) = 0;
const zeus::CVector3f& pos);
int ReturnVal() const { return m_return; } int ReturnVal() const { return m_return; }
static std::vector<Model::Vert> AABBToVerts(const zeus::CAABox& aabb, const zeus::CColor& color);
static zeus::CColor ColorForIndex(uint32_t i);
static void* makePNGBuffer(unsigned char* rgba, int width, int height, size_t* outsize);
}; };

View File

@ -0,0 +1,26 @@
#pragma once
#include "VISIRenderer.hpp"
#import <AppKit/AppKit.h>
#import <Metal/Metal.h>
#import <MetalKit/MetalKit.h>
@interface MetalRenderer : NSObject
@end
class VISIRendererMetal : public VISIRenderer {
MetalRenderer* view;
bool SetupShaders() override;
bool SetupVertexBuffersAndFormats() override;
void SetupRenderPass(const zeus::CVector3f& pos) override;
public:
VISIRendererMetal(int argc, const hecl::SystemChar** argv) : VISIRenderer(argc, argv) {
view = [[MetalRenderer alloc] init];
}
void RenderPVSOpaque(RGBA8* out, bool& needTransparent) override;
void RenderPVSTransparent(const std::function<void(int)>& passFunc) override;
void RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc,
const std::function<void(int, EPVSVisSetState)>& lightPassFunc) override;
};

View File

@ -0,0 +1,480 @@
#include "VISIRendererMetal.hh"
#include "ShaderTypes.h"
#include <zeus/CFrustum.hpp>
static zeus::CMatrix4f g_Proj;
constexpr zeus::CMatrix4f VulkanCorrect(1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.5f, 0.5f + FLT_EPSILON,
0.f, 0.f, 0.f, 1.f);
static void CalculateProjMatrix() {
float znear = 0.2f;
float zfar = 1000.f;
float tfov = std::tan(zeus::degToRad(90.f * 0.5f));
float top = znear * tfov;
float bottom = -top;
float right = znear * tfov;
float left = -right;
float rml = right - left;
float rpl = right + left;
float tmb = top - bottom;
float tpb = top + bottom;
float fpn = zfar + znear;
float fmn = zfar - znear;
zeus::CMatrix4f mat2{
2.f * znear / rml, 0.f, rpl / rml, 0.f, 0.f, 2.f * znear / tmb, tpb / tmb, 0.f, 0.f, 0.f, -fpn / fmn,
-2.f * zfar * znear / fmn, 0.f, 0.f, -1.f, 0.f};
g_Proj = VulkanCorrect * mat2;
}
static constexpr std::array<uint16_t, 20> AABBIdxs{0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 1, 7, 3, 5, 5, 0, 0, 2, 6, 4};
static const zeus::CMatrix4f LookMATs[] = {
{// Forward
1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Backward
-1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Up
1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Down
1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Left
0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Right
0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
};
using Vertex = VISIRenderer::Model::Vert;
@implementation MetalRenderer {
MTLPixelFormat _pixelFormat;
MTLPixelFormat _depthPixelFormat;
id<MTLDevice> _device;
id<MTLLibrary> _library;
id<MTLCommandQueue> _commandQueue;
id<MTLRenderPipelineState> _pipelineState;
id<MTLDepthStencilState> _depthState;
id<MTLDepthStencilState> _depthStateNoWrite;
dispatch_semaphore_t _semaphore;
id<MTLBuffer> _uniformBuffer;
id<MTLBuffer> _vertexBuffer;
id<MTLBuffer> _indexBuffer;
id<MTLBuffer> _modelQueryBuffer;
id<MTLBuffer> _entityLightQueryBuffer;
id<MTLBuffer> _aabbIndexBuffer;
id<MTLTexture> _renderTexture;
id<MTLTexture> _depthTexture;
std::array<zeus::CFrustum, 6> _frustums;
NSUInteger _entityVertStart;
}
- (bool)setup {
CalculateProjMatrix();
// Create device.
_device = MTLCreateSystemDefaultDevice();
// Set view settings.
_pixelFormat = MTLPixelFormatRGBA8Unorm;
_depthPixelFormat = MTLPixelFormatDepth32Float_Stencil8;
// Load shaders.
NSError *error = nil;
_library = [_device newLibraryWithFile:@"Shader.metallib" error:&error];
if (_library == nullptr) {
NSLog(@"Failed to load library. error %@", error);
return false;
}
id<MTLFunction> vertFunc = [_library newFunctionWithName:@"vertexShader"];
id<MTLFunction> fragFunc = [_library newFunctionWithName:@"fragmentShader"];
// Create render texture.
MTLTextureDescriptor *renderTexDesc = [MTLTextureDescriptor texture2DDescriptorWithPixelFormat:_pixelFormat
width:768
height:512
mipmapped:NO];
renderTexDesc.usage = MTLTextureUsageRenderTarget | MTLTextureUsageShaderRead;
_renderTexture = [_device newTextureWithDescriptor:renderTexDesc];
// Create depth texture.
MTLTextureDescriptor *depthTexDesc = [MTLTextureDescriptor texture2DDescriptorWithPixelFormat:_depthPixelFormat
width:768
height:512
mipmapped:NO];
depthTexDesc.usage = MTLTextureUsageRenderTarget | MTLTextureUsageShaderRead;
_depthTexture = [_device newTextureWithDescriptor:depthTexDesc];
// Create depth state.
MTLDepthStencilDescriptor *depthDesc = [[MTLDepthStencilDescriptor alloc] init];
depthDesc.depthCompareFunction = MTLCompareFunctionLessEqual;
depthDesc.depthWriteEnabled = YES;
_depthState = [_device newDepthStencilStateWithDescriptor:depthDesc];
// Create depth state (no write).
MTLDepthStencilDescriptor *depthDescNoWrite = [[MTLDepthStencilDescriptor alloc] init];
depthDescNoWrite.depthCompareFunction = MTLCompareFunctionLessEqual;
depthDescNoWrite.depthWriteEnabled = NO;
_depthStateNoWrite = [_device newDepthStencilStateWithDescriptor:depthDescNoWrite];
// Create vertex descriptor.
MTLVertexDescriptor *vertDesc = [[MTLVertexDescriptor alloc] init];
vertDesc.attributes[VertexAttributePosition].format = MTLVertexFormatFloat3;
vertDesc.attributes[VertexAttributePosition].offset = offsetof(Vertex, pos);
vertDesc.attributes[VertexAttributePosition].bufferIndex = BufferIndexVertex;
vertDesc.attributes[VertexAttributeColor].format = MTLVertexFormatFloat4;
vertDesc.attributes[VertexAttributeColor].offset = offsetof(Vertex, color);
vertDesc.attributes[VertexAttributeColor].bufferIndex = BufferIndexVertex;
vertDesc.layouts[BufferIndexVertex].stride = sizeof(Vertex);
vertDesc.layouts[BufferIndexVertex].stepRate = 1;
vertDesc.layouts[BufferIndexVertex].stepFunction = MTLVertexStepFunctionPerVertex;
// Create pipeline state.
MTLRenderPipelineDescriptor *pipelineDesc = [[MTLRenderPipelineDescriptor alloc] init];
pipelineDesc.rasterSampleCount = 1;
pipelineDesc.vertexFunction = vertFunc;
pipelineDesc.fragmentFunction = fragFunc;
pipelineDesc.vertexDescriptor = vertDesc;
pipelineDesc.colorAttachments[0].pixelFormat = _pixelFormat;
pipelineDesc.depthAttachmentPixelFormat = _depthPixelFormat;
_pipelineState = [_device newRenderPipelineStateWithDescriptor:pipelineDesc error:&error];
if (_pipelineState == nullptr) {
NSLog(@"Failed to create pipeline state, error %@", error);
return false;
}
return true;
}
- (bool)setupModels:(std::vector<VISIRenderer::Model> &)models
entities:(std::vector<VISIRenderer::Entity> &)entities
lights:(std::vector<VISIRenderer::Light> &)lights {
NSUInteger vertCount = 0;
NSUInteger indexCount = 0;
for (const auto &model : models) {
vertCount += model.verts.size();
indexCount += model.idxs.size();
}
_entityVertStart = vertCount;
vertCount += 8 * entities.size();
vertCount += lights.size();
_vertexBuffer = [_device newBufferWithLength:vertCount * sizeof(Vertex) options:MTLResourceStorageModeManaged];
_indexBuffer = [_device newBufferWithLength:indexCount * sizeof(uint32_t) options:MTLResourceStorageModeManaged];
_modelQueryBuffer = [_device newBufferWithLength:models.size() * 6 * sizeof(uint64_t)
options:MTLResourceStorageModeManaged];
_entityLightQueryBuffer = [_device newBufferWithLength:(entities.size() + lights.size()) * 6 * sizeof(uint64_t)
options:MTLResourceStorageModeManaged];
auto *buffer = static_cast<Vertex *>([_vertexBuffer contents]);
auto *indexBuffer = static_cast<uint32_t *>([_indexBuffer contents]);
for (const auto &model : models) {
memcpy(buffer, model.verts.data(), model.verts.size() * sizeof(Vertex));
memcpy(indexBuffer, model.idxs.data(), model.idxs.size() * sizeof(uint32_t));
buffer += model.verts.size();
indexBuffer += model.idxs.size();
}
auto idx = static_cast<uint32_t>(models.size());
for (const auto &ent : entities) {
auto verts = VISIRenderer::AABBToVerts(ent.aabb, VISIRenderer::ColorForIndex(idx++));
memcpy(buffer, verts.data(), verts.size() * sizeof(Vertex));
buffer += verts.size();
}
for (const auto &light : lights) {
auto *vert = buffer++;
vert->pos = light.point;
vert->color = VISIRenderer::ColorForIndex(idx++);
}
[_vertexBuffer didModifyRange:NSMakeRange(0, [_vertexBuffer length])];
[_indexBuffer didModifyRange:NSMakeRange(0, [_indexBuffer length])];
_uniformBuffer = [_device newBufferWithLength:sizeof(Uniforms) * 6 options:MTLResourceStorageModeManaged];
_aabbIndexBuffer = [_device newBufferWithBytes:AABBIdxs.data()
length:AABBIdxs.size() * sizeof(uint16_t)
options:MTLResourceStorageModeManaged];
_semaphore = dispatch_semaphore_create(0);
_commandQueue = [_device newCommandQueue];
return true;
}
- (void)setupRenderPass:(const zeus::CVector3f &)pos {
auto posMat = zeus::CTransform::Translate(-pos).toMatrix4f();
auto *buffer = static_cast<Uniforms *>([_uniformBuffer contents]);
for (uint16_t j = 0; j < 6; ++j) {
static_assert(sizeof(zeus::CMatrix4f) == sizeof(matrix_float4x4));
zeus::CMatrix4f modelView = LookMATs[j] * posMat;
_frustums[j].updatePlanes(modelView, g_Proj);
memcpy(&buffer->projectionMatrix, &g_Proj, sizeof(matrix_float4x4));
memcpy(&buffer->modelViewMatrix, &modelView, sizeof(matrix_float4x4));
buffer++;
}
[_uniformBuffer didModifyRange:NSMakeRange(0, [_uniformBuffer length])];
}
- (void)renderPVSOpaque:(std::vector<VISIRenderer::Model> &)models
out:(VISIRenderer::RGBA8 *)out
needTransparent:(bool &)needTransparent {
id<MTLCommandBuffer> commandBuffer = [_commandQueue commandBuffer];
__block dispatch_semaphore_t semaphore = _semaphore;
[commandBuffer addCompletedHandler:^(id<MTLCommandBuffer> buffer) {
dispatch_semaphore_signal(semaphore);
}];
MTLRenderPassDescriptor *passDescriptor = [MTLRenderPassDescriptor renderPassDescriptor];
passDescriptor.colorAttachments[0].texture = _renderTexture;
passDescriptor.colorAttachments[0].loadAction = MTLLoadActionClear;
passDescriptor.depthAttachment.texture = _depthTexture;
passDescriptor.depthAttachment.storeAction = MTLStoreActionStore; // stored for following render passes
id<MTLRenderCommandEncoder> encoder = [commandBuffer renderCommandEncoderWithDescriptor:passDescriptor];
[encoder setDepthStencilState:_depthState];
[encoder setRenderPipelineState:_pipelineState];
[encoder setCullMode:MTLCullModeBack];
[encoder setVertexBuffer:_vertexBuffer offset:0 atIndex:BufferIndexVertex];
[encoder setVertexBuffer:_uniformBuffer offset:0 atIndex:BufferIndexUniforms];
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
[encoder setViewport:{x, y, 256, 256, 0, 1}];
if (j > 0) {
[encoder setVertexBufferOffset:j * sizeof(Uniforms) atIndex:BufferIndexUniforms];
}
NSUInteger vertexBufferOffset = 0;
NSUInteger indexBufferOffset = 0;
for (const auto &model : models) {
if (_frustums[j].aabbFrustumTest(model.aabb)) {
[encoder setVertexBufferOffset:vertexBufferOffset atIndex:BufferIndexVertex];
for (const auto &surf : model.surfaces) {
// Non-transparents first
if (surf.transparent) {
needTransparent = true;
} else {
MTLPrimitiveType type = model.topology == hecl::HMDLTopology::TriStrips ? MTLPrimitiveTypeTriangleStrip
: MTLPrimitiveTypeTriangle;
[encoder drawIndexedPrimitives:type
indexCount:surf.count
indexType:MTLIndexTypeUInt32
indexBuffer:_indexBuffer
indexBufferOffset:indexBufferOffset + surf.first * sizeof(uint32_t)];
}
}
}
vertexBufferOffset += model.verts.size() * sizeof(Vertex);
indexBufferOffset += model.idxs.size() * sizeof(uint32_t);
}
}
[encoder endEncoding];
[commandBuffer commit];
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
[_renderTexture getBytes:out
bytesPerRow:sizeof(VISIRenderer::RGBA8) * _renderTexture.width
fromRegion:MTLRegionMake2D(0, 0, _renderTexture.width, _renderTexture.height)
mipmapLevel:0];
}
- (void)renderPVSTransparent:(std::vector<VISIRenderer::Model> &)models
passFunc:(const std::function<void(int)> &)passFunc {
// Zero out query buffer
memset([_modelQueryBuffer contents], 0, [_modelQueryBuffer length]);
[_modelQueryBuffer didModifyRange:NSMakeRange(0, [_modelQueryBuffer length])];
id<MTLCommandBuffer> commandBuffer = [_commandQueue commandBuffer];
__block dispatch_semaphore_t semaphore = _semaphore;
[commandBuffer addCompletedHandler:^(id<MTLCommandBuffer> buffer) {
dispatch_semaphore_signal(semaphore);
}];
MTLRenderPassDescriptor *passDescriptor = [MTLRenderPassDescriptor renderPassDescriptor];
passDescriptor.colorAttachments[0].texture = _renderTexture;
passDescriptor.colorAttachments[0].storeAction = MTLStoreActionDontCare; // no longer care about the render texture
passDescriptor.depthAttachment.texture = _depthTexture;
passDescriptor.depthAttachment.loadAction = MTLLoadActionLoad;
passDescriptor.depthAttachment.storeAction = MTLStoreActionStore; // still stored for following render passes
passDescriptor.visibilityResultBuffer = _modelQueryBuffer;
id<MTLRenderCommandEncoder> encoder = [commandBuffer renderCommandEncoderWithDescriptor:passDescriptor];
[encoder setDepthStencilState:_depthStateNoWrite];
[encoder setRenderPipelineState:_pipelineState];
[encoder setCullMode:MTLCullModeBack];
[encoder setVertexBuffer:_vertexBuffer offset:0 atIndex:BufferIndexVertex];
[encoder setVertexBuffer:_uniformBuffer offset:0 atIndex:BufferIndexUniforms];
NSUInteger queryCount = 0;
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
[encoder setViewport:{x, y, 256, 256, 0, 1}];
if (j > 0) {
[encoder setVertexBufferOffset:j * sizeof(Uniforms) atIndex:BufferIndexUniforms];
}
NSUInteger vertexBufferOffset = 0;
NSUInteger indexBufferOffset = 0;
for (const auto &model : models) {
if (_frustums[j].aabbFrustumTest(model.aabb)) {
[encoder setVertexBufferOffset:vertexBufferOffset atIndex:BufferIndexVertex];
[encoder setVisibilityResultMode:MTLVisibilityResultModeBoolean offset:queryCount * sizeof(uint64_t)];
for (const auto &surf : model.surfaces) {
// Only transparent surfaces
if (surf.transparent) {
MTLPrimitiveType type = model.topology == hecl::HMDLTopology::TriStrips ? MTLPrimitiveTypeTriangleStrip
: MTLPrimitiveTypeTriangle;
[encoder drawIndexedPrimitives:type
indexCount:surf.count
indexType:MTLIndexTypeUInt32
indexBuffer:_indexBuffer
indexBufferOffset:indexBufferOffset + surf.first * sizeof(uint32_t)];
}
}
}
vertexBufferOffset += model.verts.size() * sizeof(Vertex);
indexBufferOffset += model.idxs.size() * sizeof(uint32_t);
++queryCount;
}
}
[encoder endEncoding];
[commandBuffer commit];
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
auto *queries = static_cast<uint64_t *>([_modelQueryBuffer contents]);
for (int i = 0; i < models.size(); ++i) {
for (int j = 0; j < 6; ++j) {
if (queries[i + j * models.size()] != 0u) {
passFunc(i);
break;
}
}
}
}
- (void)renderPVSEntities:(std::vector<VISIRenderer::Entity> &)entities
entityPassFunc:(const std::function<void(int)> &)entityPassFunc
lights:(std::vector<VISIRenderer::Light> &)lights
lightPassFunc:(const std::function<void(int, EPVSVisSetState)> &)lightPassFunc
totalAABB:(const zeus::CAABox &)totalAABB {
// Zero out query buffer
memset([_entityLightQueryBuffer contents], 0, [_entityLightQueryBuffer length]);
[_entityLightQueryBuffer didModifyRange:NSMakeRange(0, [_entityLightQueryBuffer length])];
id<MTLCommandBuffer> commandBuffer = [_commandQueue commandBuffer];
__block dispatch_semaphore_t semaphore = _semaphore;
[commandBuffer addCompletedHandler:^(id<MTLCommandBuffer> buffer) {
dispatch_semaphore_signal(semaphore);
}];
MTLRenderPassDescriptor *passDescriptor = [MTLRenderPassDescriptor renderPassDescriptor];
passDescriptor.colorAttachments[0].texture = _renderTexture;
passDescriptor.colorAttachments[0].storeAction = MTLStoreActionDontCare; // no longer care about the render texture
passDescriptor.depthAttachment.texture = _depthTexture;
passDescriptor.depthAttachment.loadAction = MTLLoadActionLoad;
passDescriptor.visibilityResultBuffer = _entityLightQueryBuffer;
id<MTLRenderCommandEncoder> encoder = [commandBuffer renderCommandEncoderWithDescriptor:passDescriptor];
[encoder setDepthStencilState:_depthStateNoWrite];
[encoder setRenderPipelineState:_pipelineState];
[encoder setCullMode:MTLCullModeBack];
[encoder setVertexBuffer:_vertexBuffer offset:0 atIndex:BufferIndexVertex];
[encoder setVertexBuffer:_uniformBuffer offset:0 atIndex:BufferIndexUniforms];
NSUInteger queryCount = 0;
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
[encoder setViewport:{x, y, 256, 256, 0, 1}];
if (j > 0) {
[encoder setVertexBufferOffset:j * sizeof(Uniforms) atIndex:BufferIndexUniforms];
}
NSUInteger vertexBufferOffset = _entityVertStart * sizeof(Vertex);
for (const auto &ent : entities) {
if (_frustums[j].aabbFrustumTest(ent.aabb)) {
[encoder setVertexBufferOffset:vertexBufferOffset atIndex:BufferIndexVertex];
[encoder setVisibilityResultMode:MTLVisibilityResultModeBoolean offset:queryCount * sizeof(uint64_t)];
[encoder drawIndexedPrimitives:MTLPrimitiveTypeTriangleStrip
indexCount:20
indexType:MTLIndexTypeUInt32
indexBuffer:_aabbIndexBuffer
indexBufferOffset:0];
}
vertexBufferOffset += 20 * sizeof(Vertex);
++queryCount;
}
for (const auto &light : lights) {
if (_frustums[j].pointFrustumTest(light.point)) {
[encoder setVertexBufferOffset:vertexBufferOffset atIndex:BufferIndexVertex];
[encoder setVisibilityResultMode:MTLVisibilityResultModeBoolean offset:queryCount * sizeof(uint64_t)];
[encoder drawPrimitives:MTLPrimitiveTypePoint vertexStart:0 vertexCount:1];
}
vertexBufferOffset += sizeof(Vertex);
++queryCount;
}
}
[encoder endEncoding];
[commandBuffer commit];
dispatch_semaphore_wait(_semaphore, DISPATCH_TIME_FOREVER);
NSUInteger total = entities.size() + lights.size();
auto *queries = static_cast<uint64_t *>([_entityLightQueryBuffer contents]);
for (int i = 0; i < entities.size(); ++i) {
for (int j = 0; j < 6; ++j) {
if (queries[i + j * total] != 0u) {
entityPassFunc(i);
break;
}
}
}
for (int i = 0; i < lights.size(); ++i) {
bool pointInside = totalAABB.pointInside(lights[i].point);
EPVSVisSetState state = pointInside ? EPVSVisSetState::EndOfTree : EPVSVisSetState::OutOfBounds;
if (pointInside) {
for (int j = 0; j < 6; ++j) {
if (queries[entities.size() + i + j * total] != 0u) {
state = EPVSVisSetState::NodeFound;
break;
}
}
}
lightPassFunc(i, state);
}
}
@end
bool VISIRendererMetal::SetupShaders() { return [view setup]; }
bool VISIRendererMetal::SetupVertexBuffersAndFormats() {
return [view setupModels:m_models entities:m_entities lights:m_lights];
}
void VISIRendererMetal::SetupRenderPass(const zeus::CVector3f &pos) { [view setupRenderPass:pos]; }
void VISIRendererMetal::RenderPVSOpaque(RGBA8 *bufOut, bool &needTransparent) {
[view renderPVSOpaque:m_models out:bufOut needTransparent:needTransparent];
}
void VISIRendererMetal::RenderPVSTransparent(const std::function<void(int)> &passFunc) {
[view renderPVSTransparent:m_models passFunc:passFunc];
}
void VISIRendererMetal::RenderPVSEntitiesAndLights(const std::function<void(int)> &passFunc,
const std::function<void(int, EPVSVisSetState)> &lightPassFunc) {
[view renderPVSEntities:m_entities
entityPassFunc:passFunc
lights:m_lights
lightPassFunc:lightPassFunc
totalAABB:m_totalAABB];
}

View File

@ -0,0 +1,426 @@
#include "VISIRendererOpenGL.hpp"
#include "logvisor/logvisor.hpp"
#include "zeus/CFrustum.hpp"
static logvisor::Module Log("visigen");
static const char* VS =
"#version 330\n"
"layout(location=0) in vec4 posIn;\n"
"layout(location=1) in vec4 colorIn;\n"
"\n"
"uniform UniformBlock\n"
"{\n"
" mat4 xf;\n"
"};\n"
"\n"
"struct VertToFrag\n"
"{\n"
" vec4 color;\n"
"};\n"
"\n"
"out VertToFrag vtf;\n"
"void main()\n"
"{\n"
" vtf.color = colorIn;\n"
" gl_Position = xf * vec4(posIn.xyz, 1.0);\n"
"}\n";
static const char* FS =
"#version 330\n"
"struct VertToFrag\n"
"{\n"
" vec4 color;\n"
"};\n"
"\n"
"in VertToFrag vtf;\n"
"layout(location=0) out vec4 colorOut;\n"
"void main()\n"
"{\n"
" colorOut = vtf.color;\n"
"}\n";
static const uint32_t AABBIdxs[20] = {0, 1, 2, 3, 4, 5, 6, 7, 0, 1, 1, 7, 3, 5, 5, 0, 0, 2, 6, 4};
static zeus::CMatrix4f g_Proj;
static void CalculateProjMatrix() {
float znear = 0.2f;
float zfar = 1000.f;
float tfov = std::tan(zeus::degToRad(90.f * 0.5f));
float top = znear * tfov;
float bottom = -top;
float right = znear * tfov;
float left = -right;
float rml = right - left;
float rpl = right + left;
float tmb = top - bottom;
float tpb = top + bottom;
float fpn = zfar + znear;
float fmn = zfar - znear;
g_Proj = zeus::CMatrix4f(2.f * znear / rml, 0.f, rpl / rml, 0.f, 0.f, 2.f * znear / tmb, tpb / tmb, 0.f, 0.f, 0.f,
-fpn / fmn, -2.f * zfar * znear / fmn, 0.f, 0.f, -1.f, 0.f);
}
static const zeus::CMatrix4f LookMATs[] = {
{// Forward
1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Backward
-1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Up
1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Down
1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Left
0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, 1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
{// Right
0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 1.f, 0.f, -1.f, 0.f, 0.f, 0.f, 0.f, 0.f, 0.f, 1.f},
};
bool VISIRendererOpenGL::SetupShaders() {
CalculateProjMatrix();
if (glewInit() != GLEW_OK) {
Log.report(logvisor::Error, FMT_STRING("unable to init glew"));
return false;
}
if (!GLEW_ARB_occlusion_query2) {
Log.report(logvisor::Error, FMT_STRING("GL_ARB_occlusion_query2 extension not present"));
return false;
}
m_vtxShader = glCreateShader(GL_VERTEX_SHADER);
m_fragShader = glCreateShader(GL_FRAGMENT_SHADER);
m_program = glCreateProgram();
glShaderSource(m_vtxShader, 1, &VS, nullptr);
glCompileShader(m_vtxShader);
GLint status;
glGetShaderiv(m_vtxShader, GL_COMPILE_STATUS, &status);
if (status != GL_TRUE) {
GLint logLen;
glGetShaderiv(m_vtxShader, GL_INFO_LOG_LENGTH, &logLen);
char* log = (char*)malloc(logLen);
glGetShaderInfoLog(m_vtxShader, logLen, nullptr, log);
Log.report(logvisor::Error, FMT_STRING("unable to compile vert source\n{}\n{}\n"), log, VS);
free(log);
return false;
}
glShaderSource(m_fragShader, 1, &FS, nullptr);
glCompileShader(m_fragShader);
glGetShaderiv(m_fragShader, GL_COMPILE_STATUS, &status);
if (status != GL_TRUE) {
GLint logLen;
glGetShaderiv(m_fragShader, GL_INFO_LOG_LENGTH, &logLen);
char* log = (char*)malloc(logLen);
glGetShaderInfoLog(m_fragShader, logLen, nullptr, log);
Log.report(logvisor::Error, FMT_STRING("unable to compile frag source\n{}\n{}\n"), log, FS);
free(log);
return false;
}
glAttachShader(m_program, m_vtxShader);
glAttachShader(m_program, m_fragShader);
glLinkProgram(m_program);
glGetProgramiv(m_program, GL_LINK_STATUS, &status);
if (status != GL_TRUE) {
GLint logLen;
glGetProgramiv(m_program, GL_INFO_LOG_LENGTH, &logLen);
char* log = (char*)malloc(logLen);
glGetProgramInfoLog(m_program, logLen, nullptr, log);
Log.report(logvisor::Error, FMT_STRING("unable to link shader program\n{}\n"), log);
free(log);
return false;
}
glUseProgram(m_program);
m_uniLoc = glGetUniformBlockIndex(m_program, "UniformBlock");
glGenBuffers(1, &m_uniformBufferGL);
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
glBufferData(GL_UNIFORM_BUFFER, sizeof(zeus::CMatrix4f) * 6, nullptr, GL_DYNAMIC_DRAW);
glGenBuffers(1, &m_aabbIBO);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_aabbIBO);
glBufferData(GL_ELEMENT_ARRAY_BUFFER, 20 * 4, AABBIdxs, GL_STATIC_DRAW);
glEnable(GL_PRIMITIVE_RESTART);
glPrimitiveRestartIndex(0xffffffff);
return true;
}
bool VISIRendererOpenGL::SetupVertexBuffersAndFormats() {
m_modelBindings.resize(m_models.size());
m_entityBindings.resize(m_entities.size());
m_lightBindings.resize(m_lights.size());
{
auto model = m_models.begin();
auto modelBinding = m_modelBindings.begin();
while (model != m_models.end()) {
glGenVertexArrays(1, &modelBinding->vao);
glGenBuffers(1, &modelBinding->vbo);
glGenBuffers(1, &modelBinding->ibo);
glBindVertexArray(modelBinding->vao);
glBindBuffer(GL_ARRAY_BUFFER, modelBinding->vbo);
glBufferData(GL_ARRAY_BUFFER, model->verts.size() * sizeof(Model::Vert), model->verts.data(), GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, modelBinding->ibo);
glBufferData(GL_ELEMENT_ARRAY_BUFFER, model->idxs.size() * 4, model->idxs.data(), GL_STATIC_DRAW);
++model;
++modelBinding;
}
}
uint32_t idx = m_models.size();
{
auto ent = m_entities.begin();
auto entBinding = m_entityBindings.begin();
while (ent != m_entities.end()) {
glGenVertexArrays(1, &entBinding->vao);
glGenBuffers(1, &entBinding->vbo);
glBindVertexArray(entBinding->vao);
auto verts = AABBToVerts(ent->aabb, ColorForIndex(idx++));
glBindBuffer(GL_ARRAY_BUFFER, entBinding->vbo);
glBufferData(GL_ARRAY_BUFFER, verts.size() * sizeof(Model::Vert), verts.data(), GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
glBindBuffer(GL_ELEMENT_ARRAY_BUFFER, m_aabbIBO);
++ent;
++entBinding;
}
}
{
auto light = m_lights.begin();
auto lightBinding = m_lightBindings.begin();
while (light != m_lights.end()) {
glGenVertexArrays(1, &lightBinding->vao);
glGenBuffers(1, &lightBinding->vbo);
glBindVertexArray(lightBinding->vao);
Model::Vert vert;
vert.pos = light->point;
vert.color = ColorForIndex(idx++);
glBindBuffer(GL_ARRAY_BUFFER, lightBinding->vbo);
glBufferData(GL_ARRAY_BUFFER, sizeof(Model::Vert), &vert, GL_STATIC_DRAW);
glEnableVertexAttribArray(0);
glEnableVertexAttribArray(1);
glVertexAttribPointer(0, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), 0);
glVertexAttribPointer(1, 4, GL_FLOAT, GL_FALSE, sizeof(Model::Vert), (void*)16);
++light;
++lightBinding;
}
}
m_queryCount = m_models.size() + m_entities.size() + m_lights.size();
m_queries.reset(new GLuint[m_queryCount]);
m_queryBools.reset(new bool[m_queryCount]);
glGenQueries(GLsizei(m_queryCount), m_queries.get());
return true;
}
void VISIRendererOpenGL::SetupRenderPass(const zeus::CVector3f& pos) {
glViewport(0, 0, 768, 512);
glEnable(GL_CULL_FACE);
glDepthMask(GL_TRUE);
glDepthFunc(GL_LEQUAL);
glEnable(GL_DEPTH_TEST);
glClearColor(0.f, 0.f, 0.f, 1.f);
glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT);
std::array<std::aligned_union_t<256, zeus::CMatrix4f>, 6> m_uniformBuffers{};
static_assert(sizeof(m_uniformBuffers) == 256 * 6);
glBindBuffer(GL_UNIFORM_BUFFER, m_uniformBufferGL);
for (int j = 0; j < 6; ++j) {
zeus::CMatrix4f mv = LookMATs[j] * zeus::CTransform::Translate(-pos).toMatrix4f();
*static_cast<zeus::CMatrix4f*>(static_cast<void*>(&m_uniformBuffers[j])) = g_Proj * mv;
m_frustums[j].updatePlanes(mv, g_Proj);
}
glBufferData(GL_UNIFORM_BUFFER, sizeof(m_uniformBuffers), m_uniformBuffers.data(), GL_DYNAMIC_DRAW);
glUniformBlockBinding(m_program, m_uniLoc, 0);
}
void VISIRendererOpenGL::RenderPVSOpaque(RGBA8* bufOut, bool& needTransparent) {
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, static_cast<GLintptr>(256 * j),
sizeof(zeus::CMatrix4f));
// Draw frontfaces
glCullFace(GL_BACK);
glColorMask(GL_TRUE, GL_TRUE, GL_TRUE, GL_TRUE);
auto model = m_models.begin();
auto modelBinding = m_modelBindings.begin();
while (model != m_models.end()) {
if (m_frustums[j].aabbFrustumTest(model->aabb)) {
glBindVertexArray(modelBinding->vao);
for (const Model::Surface& surf : model->surfaces) {
// Non-transparents first
if (surf.transparent) {
needTransparent = true;
} else {
GLenum topology = model->topology == hecl::HMDLTopology::TriStrips ? GL_TRIANGLE_STRIP : GL_TRIANGLES;
glDrawElements(topology, surf.count, GL_UNSIGNED_INT, reinterpret_cast<void*>(uintptr_t(surf.first * 4)));
}
}
}
++model;
++modelBinding;
}
}
// m_swapFunc();
glFinish();
glReadPixels(0, 0, 768, 512, GL_RGBA, GL_UNSIGNED_BYTE, (GLvoid*)bufOut);
}
void VISIRendererOpenGL::RenderPVSTransparent(const std::function<void(int)>& passFunc) {
glDepthMask(GL_FALSE);
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, static_cast<GLintptr>(256 * j),
sizeof(zeus::CMatrix4f));
memset(m_queryBools.get(), 0, m_queryCount);
int idx = 0;
auto model = m_models.begin();
auto modelBinding = m_modelBindings.begin();
while (model != m_models.end()) {
if (m_frustums[j].aabbFrustumTest(model->aabb)) {
glBindVertexArray(modelBinding->vao);
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
m_queryBools[idx] = true;
for (const Model::Surface& surf : model->surfaces) {
// transparents
if (surf.transparent) {
GLenum topology = model->topology == hecl::HMDLTopology::TriStrips ? GL_TRIANGLE_STRIP : GL_TRIANGLES;
glDrawElements(topology, surf.count, GL_UNSIGNED_INT, reinterpret_cast<void*>(uintptr_t(surf.first * 4)));
}
}
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
}
++idx;
++model;
++modelBinding;
}
for (int i = 0; i < idx; ++i) {
if (m_queryBools[i]) {
GLint res;
glGetQueryObjectiv(m_queries[i], GL_QUERY_RESULT, &res);
if (res)
passFunc(i);
}
}
}
}
void VISIRendererOpenGL::RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc,
const std::function<void(int, EPVSVisSetState)>& lightPassFunc) {
glDepthMask(GL_FALSE);
for (int j = 0; j < 6; ++j) {
GLint x = (j % 3) * 256;
GLint y = (j / 3) * 256;
glViewport(x, y, 256, 256);
glBindBufferRange(GL_UNIFORM_BUFFER, 0, m_uniformBufferGL, static_cast<GLintptr>(256 * j),
sizeof(zeus::CMatrix4f));
memset(m_queryBools.get(), 0, m_queryCount);
uint32_t idx = m_models.size();
{
auto ent = m_entities.begin();
auto entBinding = m_entityBindings.begin();
while (ent != m_entities.end()) {
if (m_frustums[j].aabbFrustumTest(ent->aabb)) {
glBindVertexArray(entBinding->vao);
m_queryBools[idx] = true;
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
glDrawElements(GL_TRIANGLE_STRIP, 20, GL_UNSIGNED_INT, 0);
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
}
++idx;
++ent;
++entBinding;
}
}
{
auto light = m_lights.begin();
auto lightBinding = m_lightBindings.begin();
while (light != m_lights.end()) {
if (m_frustums[j].pointFrustumTest(light->point)) {
glBindVertexArray(lightBinding->vao);
m_queryBools[idx] = true;
glBeginQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE, m_queries[idx]);
glDrawArrays(GL_POINTS, 0, 1);
glEndQuery(GL_ANY_SAMPLES_PASSED_CONSERVATIVE);
}
++idx;
++light;
++lightBinding;
}
}
idx = m_models.size();
for (const Entity& ent : m_entities) {
(void)ent;
if (m_queryBools[idx]) {
GLint res;
glGetQueryObjectiv(m_queries[idx], GL_QUERY_RESULT, &res);
if (res)
passFunc(idx);
}
++idx;
}
int lightIdx = 0;
for (const Light& light : m_lights) {
if (m_queryBools[idx]) {
GLint res;
glGetQueryObjectiv(m_queries[idx], GL_QUERY_RESULT, &res);
EPVSVisSetState state =
m_totalAABB.pointInside(light.point) ? EPVSVisSetState::EndOfTree : EPVSVisSetState::OutOfBounds;
if (res && state == EPVSVisSetState::EndOfTree)
state = EPVSVisSetState::NodeFound;
lightPassFunc(lightIdx, state);
}
++lightIdx;
++idx;
}
}
}

View File

@ -0,0 +1,42 @@
#pragma once
#include "VISIRenderer.hpp"
#include "boo/graphicsdev/glew.h"
#include <zeus/CFrustum.hpp>
class VISIRendererOpenGL : public VISIRenderer {
GLuint m_vtxShader, m_fragShader, m_program, m_uniLoc;
GLuint m_uniformBufferGL;
GLuint m_aabbIBO;
std::array<zeus::CFrustum, 6> m_frustums;
struct ModelBinding {
GLuint vbo, ibo, vao;
};
struct EntityBinding {
GLuint vbo, vao;
};
struct LightBinding {
GLuint vbo, vao;
};
std::vector<ModelBinding> m_modelBindings;
std::vector<EntityBinding> m_entityBindings;
std::vector<LightBinding> m_lightBindings;
size_t m_queryCount;
std::unique_ptr<GLuint[]> m_queries;
std::unique_ptr<bool[]> m_queryBools;
bool SetupShaders() override;
bool SetupVertexBuffersAndFormats() override;
void SetupRenderPass(const zeus::CVector3f& pos) override;
public:
VISIRendererOpenGL(int argc, const hecl::SystemChar** argv) : VISIRenderer(argc, argv) {}
void RenderPVSOpaque(RGBA8* bufOut, bool& needTransparent) override;
void RenderPVSTransparent(const std::function<void(int)>& passFunc) override;
void RenderPVSEntitiesAndLights(const std::function<void(int)>& passFunc,
const std::function<void(int, EPVSVisSetState)>& lightPassFunc) override;
};