├── .gitignore ├── Doc └── sample.png ├── Externals └── stb_image │ ├── stb_image.h │ └── stb_image_write.h ├── Luma.sln ├── Luma.vcxproj ├── Luma.vcxproj.filters ├── README.md └── Source ├── Camera.h ├── Image.h ├── Ray.h ├── Scene.h ├── Sphere.h ├── Utils.h ├── Vec3.h ├── main.cpp ├── pch.cpp └── pch.h /.gitignore: -------------------------------------------------------------------------------- 1 | ## Ignore Visual Studio temporary files, build results, and 2 | ## files generated by popular Visual Studio add-ons. 3 | ## 4 | ## Get latest from https://github.com/github/gitignore/blob/master/VisualStudio.gitignore 5 | 6 | # User-specific files 7 | *.rsuser 8 | *.suo 9 | *.user 10 | *.userosscache 11 | *.sln.docstates 12 | 13 | # User-specific files (MonoDevelop/Xamarin Studio) 14 | *.userprefs 15 | 16 | # Build results 17 | Build/ 18 | [Dd]ebug/ 19 | [Dd]ebugPublic/ 20 | [Rr]elease/ 21 | [Rr]eleases/ 22 | x64/ 23 | x86/ 24 | [Aa][Rr][Mm]/ 25 | [Aa][Rr][Mm]64/ 26 | bld/ 27 | [Bb]in/ 28 | [Oo]bj/ 29 | [Ll]og/ 30 | 31 | # Visual Studio 2015/2017 cache/options directory 32 | .vs/ 33 | # Uncomment if you have tasks that create the project's static files in wwwroot 34 | #wwwroot/ 35 | 36 | # Visual Studio 2017 auto generated files 37 | Generated\ Files/ 38 | 39 | # MSTest test Results 40 | [Tt]est[Rr]esult*/ 41 | [Bb]uild[Ll]og.* 42 | 43 | # NUNIT 44 | *.VisualState.xml 45 | TestResult.xml 46 | 47 | # Build Results of an ATL Project 48 | [Dd]ebugPS/ 49 | [Rr]eleasePS/ 50 | dlldata.c 51 | 52 | # Benchmark Results 53 | BenchmarkDotNet.Artifacts/ 54 | 55 | # .NET Core 56 | project.lock.json 57 | project.fragment.lock.json 58 | artifacts/ 59 | 60 | # StyleCop 61 | StyleCopReport.xml 62 | 63 | # Files built by Visual Studio 64 | *_i.c 65 | *_p.c 66 | *_h.h 67 | *.ilk 68 | *.meta 69 | *.obj 70 | *.iobj 71 | *.pch 72 | *.pdb 73 | *.ipdb 74 | *.pgc 75 | *.pgd 76 | *.rsp 77 | *.sbr 78 | *.tlb 79 | *.tli 80 | *.tlh 81 | *.tmp 82 | *.tmp_proj 83 | *_wpftmp.csproj 84 | *.log 85 | *.vspscc 86 | *.vssscc 87 | .builds 88 | *.pidb 89 | *.svclog 90 | *.scc 91 | 92 | # Chutzpah Test files 93 | _Chutzpah* 94 | 95 | # Visual C++ cache files 96 | ipch/ 97 | *.aps 98 | *.ncb 99 | *.opendb 100 | *.opensdf 101 | *.sdf 102 | *.cachefile 103 | *.VC.db 104 | *.VC.VC.opendb 105 | 106 | # Visual Studio profiler 107 | *.psess 108 | *.vsp 109 | *.vspx 110 | *.sap 111 | 112 | # Visual Studio Trace Files 113 | *.e2e 114 | 115 | # TFS 2012 Local Workspace 116 | $tf/ 117 | 118 | # Guidance Automation Toolkit 119 | *.gpState 120 | 121 | # ReSharper is a .NET coding add-in 122 | _ReSharper*/ 123 | *.[Rr]e[Ss]harper 124 | *.DotSettings.user 125 | 126 | # JustCode is a .NET coding add-in 127 | .JustCode 128 | 129 | # TeamCity is a build add-in 130 | _TeamCity* 131 | 132 | # DotCover is a Code Coverage Tool 133 | *.dotCover 134 | 135 | # AxoCover is a Code Coverage Tool 136 | .axoCover/* 137 | !.axoCover/settings.json 138 | 139 | # Visual Studio code coverage results 140 | *.coverage 141 | *.coveragexml 142 | 143 | # NCrunch 144 | _NCrunch_* 145 | .*crunch*.local.xml 146 | nCrunchTemp_* 147 | 148 | # MightyMoose 149 | *.mm.* 150 | AutoTest.Net/ 151 | 152 | # Web workbench (sass) 153 | .sass-cache/ 154 | 155 | # Installshield output folder 156 | [Ee]xpress/ 157 | 158 | # DocProject is a documentation generator add-in 159 | DocProject/buildhelp/ 160 | DocProject/Help/*.HxT 161 | DocProject/Help/*.HxC 162 | DocProject/Help/*.hhc 163 | DocProject/Help/*.hhk 164 | DocProject/Help/*.hhp 165 | DocProject/Help/Html2 166 | DocProject/Help/html 167 | 168 | # Click-Once directory 169 | publish/ 170 | 171 | # Publish Web Output 172 | *.[Pp]ublish.xml 173 | *.azurePubxml 174 | # Note: Comment the next line if you want to checkin your web deploy settings, 175 | # but database connection strings (with potential passwords) will be unencrypted 176 | *.pubxml 177 | *.publishproj 178 | 179 | # Microsoft Azure Web App publish settings. Comment the next line if you want to 180 | # checkin your Azure Web App publish settings, but sensitive information contained 181 | # in these scripts will be unencrypted 182 | PublishScripts/ 183 | 184 | # NuGet Packages 185 | *.nupkg 186 | # The packages folder can be ignored because of Package Restore 187 | **/[Pp]ackages/* 188 | # except build/, which is used as an MSBuild target. 189 | !**/[Pp]ackages/build/ 190 | # Uncomment if necessary however generally it will be regenerated when needed 191 | #!**/[Pp]ackages/repositories.config 192 | # NuGet v3's project.json files produces more ignorable files 193 | *.nuget.props 194 | *.nuget.targets 195 | 196 | # Microsoft Azure Build Output 197 | csx/ 198 | *.build.csdef 199 | 200 | # Microsoft Azure Emulator 201 | ecf/ 202 | rcf/ 203 | 204 | # Windows Store app package directories and files 205 | AppPackages/ 206 | BundleArtifacts/ 207 | Package.StoreAssociation.xml 208 | _pkginfo.txt 209 | *.appx 210 | 211 | # Visual Studio cache files 212 | # files ending in .cache can be ignored 213 | *.[Cc]ache 214 | # but keep track of directories ending in .cache 215 | !?*.[Cc]ache/ 216 | 217 | # Others 218 | ClientBin/ 219 | ~$* 220 | *~ 221 | *.dbmdl 222 | *.dbproj.schemaview 223 | *.jfm 224 | *.pfx 225 | *.publishsettings 226 | orleans.codegen.cs 227 | 228 | # Including strong name files can present a security risk 229 | # (https://github.com/github/gitignore/pull/2483#issue-259490424) 230 | #*.snk 231 | 232 | # Since there are multiple workflows, uncomment next line to ignore bower_components 233 | # (https://github.com/github/gitignore/pull/1529#issuecomment-104372622) 234 | #bower_components/ 235 | 236 | # RIA/Silverlight projects 237 | Generated_Code/ 238 | 239 | # Backup & report files from converting an old project file 240 | # to a newer Visual Studio version. Backup files are not needed, 241 | # because we have git ;-) 242 | _UpgradeReport_Files/ 243 | Backup*/ 244 | UpgradeLog*.XML 245 | UpgradeLog*.htm 246 | ServiceFabricBackup/ 247 | *.rptproj.bak 248 | 249 | # SQL Server files 250 | *.mdf 251 | *.ldf 252 | *.ndf 253 | 254 | # Business Intelligence projects 255 | *.rdl.data 256 | *.bim.layout 257 | *.bim_*.settings 258 | *.rptproj.rsuser 259 | *- Backup*.rdl 260 | 261 | # Microsoft Fakes 262 | FakesAssemblies/ 263 | 264 | # GhostDoc plugin setting file 265 | *.GhostDoc.xml 266 | 267 | # Node.js Tools for Visual Studio 268 | .ntvs_analysis.dat 269 | node_modules/ 270 | 271 | # Visual Studio 6 build log 272 | *.plg 273 | 274 | # Visual Studio 6 workspace options file 275 | *.opt 276 | 277 | # Visual Studio 6 auto-generated workspace file (contains which files were open etc.) 278 | *.vbw 279 | 280 | # Visual Studio LightSwitch build output 281 | **/*.HTMLClient/GeneratedArtifacts 282 | **/*.DesktopClient/GeneratedArtifacts 283 | **/*.DesktopClient/ModelManifest.xml 284 | **/*.Server/GeneratedArtifacts 285 | **/*.Server/ModelManifest.xml 286 | _Pvt_Extensions 287 | 288 | # Paket dependency manager 289 | .paket/paket.exe 290 | paket-files/ 291 | 292 | # FAKE - F# Make 293 | .fake/ 294 | 295 | # JetBrains Rider 296 | .idea/ 297 | *.sln.iml 298 | 299 | # CodeRush personal settings 300 | .cr/personal 301 | 302 | # Python Tools for Visual Studio (PTVS) 303 | __pycache__/ 304 | *.pyc 305 | 306 | # Cake - Uncomment if you are using it 307 | # tools/** 308 | # !tools/packages.config 309 | 310 | # Tabs Studio 311 | *.tss 312 | 313 | # Telerik's JustMock configuration file 314 | *.jmconfig 315 | 316 | # BizTalk build output 317 | *.btp.cs 318 | *.btm.cs 319 | *.odx.cs 320 | *.xsd.cs 321 | 322 | # OpenCover UI analysis results 323 | OpenCover/ 324 | 325 | # Azure Stream Analytics local run output 326 | ASALocalRun/ 327 | 328 | # MSBuild Binary and Structured Log 329 | *.binlog 330 | 331 | # NVidia Nsight GPU debugger configuration file 332 | *.nvuser 333 | 334 | # MFractors (Xamarin productivity tool) working folder 335 | .mfractor/ 336 | 337 | # Local History for Visual Studio 338 | .localhistory/ 339 | 340 | # BeatPulse healthcheck temp database 341 | healthchecksdb -------------------------------------------------------------------------------- /Doc/sample.png: -------------------------------------------------------------------------------- https://raw.githubusercontent.com/pixnblox/Luma/c8d45ae3e82f0b3e3ac8cda623c92b3a01b7a594/Doc/sample.png -------------------------------------------------------------------------------- /Externals/stb_image/stb_image_write.h: -------------------------------------------------------------------------------- 1 | /* stb_image_write - v1.14 - public domain - http://nothings.org/stb 2 | writes out PNG/BMP/TGA/JPEG/HDR images to C stdio - Sean Barrett 2010-2015 3 | no warranty implied; use at your own risk 4 | 5 | Before #including, 6 | 7 | #define STB_IMAGE_WRITE_IMPLEMENTATION 8 | 9 | in the file that you want to have the implementation. 10 | 11 | Will probably not work correctly with strict-aliasing optimizations. 12 | 13 | ABOUT: 14 | 15 | This header file is a library for writing images to C stdio or a callback. 16 | 17 | The PNG output is not optimal; it is 20-50% larger than the file 18 | written by a decent optimizing implementation; though providing a custom 19 | zlib compress function (see STBIW_ZLIB_COMPRESS) can mitigate that. 20 | This library is designed for source code compactness and simplicity, 21 | not optimal image file size or run-time performance. 22 | 23 | BUILDING: 24 | 25 | You can #define STBIW_ASSERT(x) before the #include to avoid using assert.h. 26 | You can #define STBIW_MALLOC(), STBIW_REALLOC(), and STBIW_FREE() to replace 27 | malloc,realloc,free. 28 | You can #define STBIW_MEMMOVE() to replace memmove() 29 | You can #define STBIW_ZLIB_COMPRESS to use a custom zlib-style compress function 30 | for PNG compression (instead of the builtin one), it must have the following signature: 31 | unsigned char * my_compress(unsigned char *data, int data_len, int *out_len, int quality); 32 | The returned data will be freed with STBIW_FREE() (free() by default), 33 | so it must be heap allocated with STBIW_MALLOC() (malloc() by default), 34 | 35 | UNICODE: 36 | 37 | If compiling for Windows and you wish to use Unicode filenames, compile 38 | with 39 | #define STBIW_WINDOWS_UTF8 40 | and pass utf8-encoded filenames. Call stbiw_convert_wchar_to_utf8 to convert 41 | Windows wchar_t filenames to utf8. 42 | 43 | USAGE: 44 | 45 | There are five functions, one for each image file format: 46 | 47 | int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes); 48 | int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data); 49 | int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data); 50 | int stbi_write_jpg(char const *filename, int w, int h, int comp, const void *data, int quality); 51 | int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data); 52 | 53 | void stbi_flip_vertically_on_write(int flag); // flag is non-zero to flip data vertically 54 | 55 | There are also five equivalent functions that use an arbitrary write function. You are 56 | expected to open/close your file-equivalent before and after calling these: 57 | 58 | int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes); 59 | int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data); 60 | int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data); 61 | int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data); 62 | int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality); 63 | 64 | where the callback is: 65 | void stbi_write_func(void *context, void *data, int size); 66 | 67 | You can configure it with these global variables: 68 | int stbi_write_tga_with_rle; // defaults to true; set to 0 to disable RLE 69 | int stbi_write_png_compression_level; // defaults to 8; set to higher for more compression 70 | int stbi_write_force_png_filter; // defaults to -1; set to 0..5 to force a filter mode 71 | 72 | 73 | You can define STBI_WRITE_NO_STDIO to disable the file variant of these 74 | functions, so the library will not use stdio.h at all. However, this will 75 | also disable HDR writing, because it requires stdio for formatted output. 76 | 77 | Each function returns 0 on failure and non-0 on success. 78 | 79 | The functions create an image file defined by the parameters. The image 80 | is a rectangle of pixels stored from left-to-right, top-to-bottom. 81 | Each pixel contains 'comp' channels of data stored interleaved with 8-bits 82 | per channel, in the following order: 1=Y, 2=YA, 3=RGB, 4=RGBA. (Y is 83 | monochrome color.) The rectangle is 'w' pixels wide and 'h' pixels tall. 84 | The *data pointer points to the first byte of the top-left-most pixel. 85 | For PNG, "stride_in_bytes" is the distance in bytes from the first byte of 86 | a row of pixels to the first byte of the next row of pixels. 87 | 88 | PNG creates output files with the same number of components as the input. 89 | The BMP format expands Y to RGB in the file format and does not 90 | output alpha. 91 | 92 | PNG supports writing rectangles of data even when the bytes storing rows of 93 | data are not consecutive in memory (e.g. sub-rectangles of a larger image), 94 | by supplying the stride between the beginning of adjacent rows. The other 95 | formats do not. (Thus you cannot write a native-format BMP through the BMP 96 | writer, both because it is in BGR order and because it may have padding 97 | at the end of the line.) 98 | 99 | PNG allows you to set the deflate compression level by setting the global 100 | variable 'stbi_write_png_compression_level' (it defaults to 8). 101 | 102 | HDR expects linear float data. Since the format is always 32-bit rgb(e) 103 | data, alpha (if provided) is discarded, and for monochrome data it is 104 | replicated across all three channels. 105 | 106 | TGA supports RLE or non-RLE compressed data. To use non-RLE-compressed 107 | data, set the global variable 'stbi_write_tga_with_rle' to 0. 108 | 109 | JPEG does ignore alpha channels in input data; quality is between 1 and 100. 110 | Higher quality looks better but results in a bigger image. 111 | JPEG baseline (no JPEG progressive). 112 | 113 | CREDITS: 114 | 115 | 116 | Sean Barrett - PNG/BMP/TGA 117 | Baldur Karlsson - HDR 118 | Jean-Sebastien Guay - TGA monochrome 119 | Tim Kelsey - misc enhancements 120 | Alan Hickman - TGA RLE 121 | Emmanuel Julien - initial file IO callback implementation 122 | Jon Olick - original jo_jpeg.cpp code 123 | Daniel Gibson - integrate JPEG, allow external zlib 124 | Aarni Koskela - allow choosing PNG filter 125 | 126 | bugfixes: 127 | github:Chribba 128 | Guillaume Chereau 129 | github:jry2 130 | github:romigrou 131 | Sergio Gonzalez 132 | Jonas Karlsson 133 | Filip Wasil 134 | Thatcher Ulrich 135 | github:poppolopoppo 136 | Patrick Boettcher 137 | github:xeekworx 138 | Cap Petschulat 139 | Simon Rodriguez 140 | Ivan Tikhonov 141 | github:ignotion 142 | Adam Schackart 143 | 144 | LICENSE 145 | 146 | See end of file for license information. 147 | 148 | */ 149 | 150 | #ifndef INCLUDE_STB_IMAGE_WRITE_H 151 | #define INCLUDE_STB_IMAGE_WRITE_H 152 | 153 | #include 154 | 155 | // if STB_IMAGE_WRITE_STATIC causes problems, try defining STBIWDEF to 'inline' or 'static inline' 156 | #ifndef STBIWDEF 157 | #ifdef STB_IMAGE_WRITE_STATIC 158 | #define STBIWDEF static 159 | #else 160 | #ifdef __cplusplus 161 | #define STBIWDEF extern "C" 162 | #else 163 | #define STBIWDEF extern 164 | #endif 165 | #endif 166 | #endif 167 | 168 | #ifndef STB_IMAGE_WRITE_STATIC // C++ forbids static forward declarations 169 | extern int stbi_write_tga_with_rle; 170 | extern int stbi_write_png_compression_level; 171 | extern int stbi_write_force_png_filter; 172 | #endif 173 | 174 | #ifndef STBI_WRITE_NO_STDIO 175 | STBIWDEF int stbi_write_png(char const *filename, int w, int h, int comp, const void *data, int stride_in_bytes); 176 | STBIWDEF int stbi_write_bmp(char const *filename, int w, int h, int comp, const void *data); 177 | STBIWDEF int stbi_write_tga(char const *filename, int w, int h, int comp, const void *data); 178 | STBIWDEF int stbi_write_hdr(char const *filename, int w, int h, int comp, const float *data); 179 | STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality); 180 | 181 | #ifdef STBI_WINDOWS_UTF8 182 | STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input); 183 | #endif 184 | #endif 185 | 186 | typedef void stbi_write_func(void *context, void *data, int size); 187 | 188 | STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data, int stride_in_bytes); 189 | STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data); 190 | STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const void *data); 191 | STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int w, int h, int comp, const float *data); 192 | STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality); 193 | 194 | STBIWDEF void stbi_flip_vertically_on_write(int flip_boolean); 195 | 196 | #endif//INCLUDE_STB_IMAGE_WRITE_H 197 | 198 | #ifdef STB_IMAGE_WRITE_IMPLEMENTATION 199 | 200 | #ifdef _WIN32 201 | #ifndef _CRT_SECURE_NO_WARNINGS 202 | #define _CRT_SECURE_NO_WARNINGS 203 | #endif 204 | #ifndef _CRT_NONSTDC_NO_DEPRECATE 205 | #define _CRT_NONSTDC_NO_DEPRECATE 206 | #endif 207 | #endif 208 | 209 | #ifndef STBI_WRITE_NO_STDIO 210 | #include 211 | #endif // STBI_WRITE_NO_STDIO 212 | 213 | #include 214 | #include 215 | #include 216 | #include 217 | 218 | #if defined(STBIW_MALLOC) && defined(STBIW_FREE) && (defined(STBIW_REALLOC) || defined(STBIW_REALLOC_SIZED)) 219 | // ok 220 | #elif !defined(STBIW_MALLOC) && !defined(STBIW_FREE) && !defined(STBIW_REALLOC) && !defined(STBIW_REALLOC_SIZED) 221 | // ok 222 | #else 223 | #error "Must define all or none of STBIW_MALLOC, STBIW_FREE, and STBIW_REALLOC (or STBIW_REALLOC_SIZED)." 224 | #endif 225 | 226 | #ifndef STBIW_MALLOC 227 | #define STBIW_MALLOC(sz) malloc(sz) 228 | #define STBIW_REALLOC(p,newsz) realloc(p,newsz) 229 | #define STBIW_FREE(p) free(p) 230 | #endif 231 | 232 | #ifndef STBIW_REALLOC_SIZED 233 | #define STBIW_REALLOC_SIZED(p,oldsz,newsz) STBIW_REALLOC(p,newsz) 234 | #endif 235 | 236 | 237 | #ifndef STBIW_MEMMOVE 238 | #define STBIW_MEMMOVE(a,b,sz) memmove(a,b,sz) 239 | #endif 240 | 241 | 242 | #ifndef STBIW_ASSERT 243 | #include 244 | #define STBIW_ASSERT(x) assert(x) 245 | #endif 246 | 247 | #define STBIW_UCHAR(x) (unsigned char) ((x) & 0xff) 248 | 249 | #ifdef STB_IMAGE_WRITE_STATIC 250 | static int stbi_write_png_compression_level = 8; 251 | static int stbi_write_tga_with_rle = 1; 252 | static int stbi_write_force_png_filter = -1; 253 | #else 254 | int stbi_write_png_compression_level = 8; 255 | int stbi_write_tga_with_rle = 1; 256 | int stbi_write_force_png_filter = -1; 257 | #endif 258 | 259 | static int stbi__flip_vertically_on_write = 0; 260 | 261 | STBIWDEF void stbi_flip_vertically_on_write(int flag) 262 | { 263 | stbi__flip_vertically_on_write = flag; 264 | } 265 | 266 | typedef struct 267 | { 268 | stbi_write_func *func; 269 | void *context; 270 | } stbi__write_context; 271 | 272 | // initialize a callback-based context 273 | static void stbi__start_write_callbacks(stbi__write_context *s, stbi_write_func *c, void *context) 274 | { 275 | s->func = c; 276 | s->context = context; 277 | } 278 | 279 | #ifndef STBI_WRITE_NO_STDIO 280 | 281 | static void stbi__stdio_write(void *context, void *data, int size) 282 | { 283 | fwrite(data,1,size,(FILE*) context); 284 | } 285 | 286 | #if defined(_MSC_VER) && defined(STBI_WINDOWS_UTF8) 287 | #ifdef __cplusplus 288 | #define STBIW_EXTERN extern "C" 289 | #else 290 | #define STBIW_EXTERN extern 291 | #endif 292 | STBIW_EXTERN __declspec(dllimport) int __stdcall MultiByteToWideChar(unsigned int cp, unsigned long flags, const char *str, int cbmb, wchar_t *widestr, int cchwide); 293 | STBIW_EXTERN __declspec(dllimport) int __stdcall WideCharToMultiByte(unsigned int cp, unsigned long flags, const wchar_t *widestr, int cchwide, char *str, int cbmb, const char *defchar, int *used_default); 294 | 295 | STBIWDEF int stbiw_convert_wchar_to_utf8(char *buffer, size_t bufferlen, const wchar_t* input) 296 | { 297 | return WideCharToMultiByte(65001 /* UTF8 */, 0, input, -1, buffer, (int) bufferlen, NULL, NULL); 298 | } 299 | #endif 300 | 301 | static FILE *stbiw__fopen(char const *filename, char const *mode) 302 | { 303 | FILE *f; 304 | #if defined(_MSC_VER) && defined(STBI_WINDOWS_UTF8) 305 | wchar_t wMode[64]; 306 | wchar_t wFilename[1024]; 307 | if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, filename, -1, wFilename, sizeof(wFilename))) 308 | return 0; 309 | 310 | if (0 == MultiByteToWideChar(65001 /* UTF8 */, 0, mode, -1, wMode, sizeof(wMode))) 311 | return 0; 312 | 313 | #if _MSC_VER >= 1400 314 | if (0 != _wfopen_s(&f, wFilename, wMode)) 315 | f = 0; 316 | #else 317 | f = _wfopen(wFilename, wMode); 318 | #endif 319 | 320 | #elif defined(_MSC_VER) && _MSC_VER >= 1400 321 | if (0 != fopen_s(&f, filename, mode)) 322 | f=0; 323 | #else 324 | f = fopen(filename, mode); 325 | #endif 326 | return f; 327 | } 328 | 329 | static int stbi__start_write_file(stbi__write_context *s, const char *filename) 330 | { 331 | FILE *f = stbiw__fopen(filename, "wb"); 332 | stbi__start_write_callbacks(s, stbi__stdio_write, (void *) f); 333 | return f != NULL; 334 | } 335 | 336 | static void stbi__end_write_file(stbi__write_context *s) 337 | { 338 | fclose((FILE *)s->context); 339 | } 340 | 341 | #endif // !STBI_WRITE_NO_STDIO 342 | 343 | typedef unsigned int stbiw_uint32; 344 | typedef int stb_image_write_test[sizeof(stbiw_uint32)==4 ? 1 : -1]; 345 | 346 | static void stbiw__writefv(stbi__write_context *s, const char *fmt, va_list v) 347 | { 348 | while (*fmt) { 349 | switch (*fmt++) { 350 | case ' ': break; 351 | case '1': { unsigned char x = STBIW_UCHAR(va_arg(v, int)); 352 | s->func(s->context,&x,1); 353 | break; } 354 | case '2': { int x = va_arg(v,int); 355 | unsigned char b[2]; 356 | b[0] = STBIW_UCHAR(x); 357 | b[1] = STBIW_UCHAR(x>>8); 358 | s->func(s->context,b,2); 359 | break; } 360 | case '4': { stbiw_uint32 x = va_arg(v,int); 361 | unsigned char b[4]; 362 | b[0]=STBIW_UCHAR(x); 363 | b[1]=STBIW_UCHAR(x>>8); 364 | b[2]=STBIW_UCHAR(x>>16); 365 | b[3]=STBIW_UCHAR(x>>24); 366 | s->func(s->context,b,4); 367 | break; } 368 | default: 369 | STBIW_ASSERT(0); 370 | return; 371 | } 372 | } 373 | } 374 | 375 | static void stbiw__writef(stbi__write_context *s, const char *fmt, ...) 376 | { 377 | va_list v; 378 | va_start(v, fmt); 379 | stbiw__writefv(s, fmt, v); 380 | va_end(v); 381 | } 382 | 383 | static void stbiw__putc(stbi__write_context *s, unsigned char c) 384 | { 385 | s->func(s->context, &c, 1); 386 | } 387 | 388 | static void stbiw__write3(stbi__write_context *s, unsigned char a, unsigned char b, unsigned char c) 389 | { 390 | unsigned char arr[3]; 391 | arr[0] = a; arr[1] = b; arr[2] = c; 392 | s->func(s->context, arr, 3); 393 | } 394 | 395 | static void stbiw__write_pixel(stbi__write_context *s, int rgb_dir, int comp, int write_alpha, int expand_mono, unsigned char *d) 396 | { 397 | unsigned char bg[3] = { 255, 0, 255}, px[3]; 398 | int k; 399 | 400 | if (write_alpha < 0) 401 | s->func(s->context, &d[comp - 1], 1); 402 | 403 | switch (comp) { 404 | case 2: // 2 pixels = mono + alpha, alpha is written separately, so same as 1-channel case 405 | case 1: 406 | if (expand_mono) 407 | stbiw__write3(s, d[0], d[0], d[0]); // monochrome bmp 408 | else 409 | s->func(s->context, d, 1); // monochrome TGA 410 | break; 411 | case 4: 412 | if (!write_alpha) { 413 | // composite against pink background 414 | for (k = 0; k < 3; ++k) 415 | px[k] = bg[k] + ((d[k] - bg[k]) * d[3]) / 255; 416 | stbiw__write3(s, px[1 - rgb_dir], px[1], px[1 + rgb_dir]); 417 | break; 418 | } 419 | /* FALLTHROUGH */ 420 | case 3: 421 | stbiw__write3(s, d[1 - rgb_dir], d[1], d[1 + rgb_dir]); 422 | break; 423 | } 424 | if (write_alpha > 0) 425 | s->func(s->context, &d[comp - 1], 1); 426 | } 427 | 428 | static void stbiw__write_pixels(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, void *data, int write_alpha, int scanline_pad, int expand_mono) 429 | { 430 | stbiw_uint32 zero = 0; 431 | int i,j, j_end; 432 | 433 | if (y <= 0) 434 | return; 435 | 436 | if (stbi__flip_vertically_on_write) 437 | vdir *= -1; 438 | 439 | if (vdir < 0) { 440 | j_end = -1; j = y-1; 441 | } else { 442 | j_end = y; j = 0; 443 | } 444 | 445 | for (; j != j_end; j += vdir) { 446 | for (i=0; i < x; ++i) { 447 | unsigned char *d = (unsigned char *) data + (j*x+i)*comp; 448 | stbiw__write_pixel(s, rgb_dir, comp, write_alpha, expand_mono, d); 449 | } 450 | s->func(s->context, &zero, scanline_pad); 451 | } 452 | } 453 | 454 | static int stbiw__outfile(stbi__write_context *s, int rgb_dir, int vdir, int x, int y, int comp, int expand_mono, void *data, int alpha, int pad, const char *fmt, ...) 455 | { 456 | if (y < 0 || x < 0) { 457 | return 0; 458 | } else { 459 | va_list v; 460 | va_start(v, fmt); 461 | stbiw__writefv(s, fmt, v); 462 | va_end(v); 463 | stbiw__write_pixels(s,rgb_dir,vdir,x,y,comp,data,alpha,pad, expand_mono); 464 | return 1; 465 | } 466 | } 467 | 468 | static int stbi_write_bmp_core(stbi__write_context *s, int x, int y, int comp, const void *data) 469 | { 470 | int pad = (-x*3) & 3; 471 | return stbiw__outfile(s,-1,-1,x,y,comp,1,(void *) data,0,pad, 472 | "11 4 22 4" "4 44 22 444444", 473 | 'B', 'M', 14+40+(x*3+pad)*y, 0,0, 14+40, // file header 474 | 40, x,y, 1,24, 0,0,0,0,0,0); // bitmap header 475 | } 476 | 477 | STBIWDEF int stbi_write_bmp_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data) 478 | { 479 | stbi__write_context s; 480 | stbi__start_write_callbacks(&s, func, context); 481 | return stbi_write_bmp_core(&s, x, y, comp, data); 482 | } 483 | 484 | #ifndef STBI_WRITE_NO_STDIO 485 | STBIWDEF int stbi_write_bmp(char const *filename, int x, int y, int comp, const void *data) 486 | { 487 | stbi__write_context s; 488 | if (stbi__start_write_file(&s,filename)) { 489 | int r = stbi_write_bmp_core(&s, x, y, comp, data); 490 | stbi__end_write_file(&s); 491 | return r; 492 | } else 493 | return 0; 494 | } 495 | #endif //!STBI_WRITE_NO_STDIO 496 | 497 | static int stbi_write_tga_core(stbi__write_context *s, int x, int y, int comp, void *data) 498 | { 499 | int has_alpha = (comp == 2 || comp == 4); 500 | int colorbytes = has_alpha ? comp-1 : comp; 501 | int format = colorbytes < 2 ? 3 : 2; // 3 color channels (RGB/RGBA) = 2, 1 color channel (Y/YA) = 3 502 | 503 | if (y < 0 || x < 0) 504 | return 0; 505 | 506 | if (!stbi_write_tga_with_rle) { 507 | return stbiw__outfile(s, -1, -1, x, y, comp, 0, (void *) data, has_alpha, 0, 508 | "111 221 2222 11", 0, 0, format, 0, 0, 0, 0, 0, x, y, (colorbytes + has_alpha) * 8, has_alpha * 8); 509 | } else { 510 | int i,j,k; 511 | int jend, jdir; 512 | 513 | stbiw__writef(s, "111 221 2222 11", 0,0,format+8, 0,0,0, 0,0,x,y, (colorbytes + has_alpha) * 8, has_alpha * 8); 514 | 515 | if (stbi__flip_vertically_on_write) { 516 | j = 0; 517 | jend = y; 518 | jdir = 1; 519 | } else { 520 | j = y-1; 521 | jend = -1; 522 | jdir = -1; 523 | } 524 | for (; j != jend; j += jdir) { 525 | unsigned char *row = (unsigned char *) data + j * x * comp; 526 | int len; 527 | 528 | for (i = 0; i < x; i += len) { 529 | unsigned char *begin = row + i * comp; 530 | int diff = 1; 531 | len = 1; 532 | 533 | if (i < x - 1) { 534 | ++len; 535 | diff = memcmp(begin, row + (i + 1) * comp, comp); 536 | if (diff) { 537 | const unsigned char *prev = begin; 538 | for (k = i + 2; k < x && len < 128; ++k) { 539 | if (memcmp(prev, row + k * comp, comp)) { 540 | prev += comp; 541 | ++len; 542 | } else { 543 | --len; 544 | break; 545 | } 546 | } 547 | } else { 548 | for (k = i + 2; k < x && len < 128; ++k) { 549 | if (!memcmp(begin, row + k * comp, comp)) { 550 | ++len; 551 | } else { 552 | break; 553 | } 554 | } 555 | } 556 | } 557 | 558 | if (diff) { 559 | unsigned char header = STBIW_UCHAR(len - 1); 560 | s->func(s->context, &header, 1); 561 | for (k = 0; k < len; ++k) { 562 | stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin + k * comp); 563 | } 564 | } else { 565 | unsigned char header = STBIW_UCHAR(len - 129); 566 | s->func(s->context, &header, 1); 567 | stbiw__write_pixel(s, -1, comp, has_alpha, 0, begin); 568 | } 569 | } 570 | } 571 | } 572 | return 1; 573 | } 574 | 575 | STBIWDEF int stbi_write_tga_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data) 576 | { 577 | stbi__write_context s; 578 | stbi__start_write_callbacks(&s, func, context); 579 | return stbi_write_tga_core(&s, x, y, comp, (void *) data); 580 | } 581 | 582 | #ifndef STBI_WRITE_NO_STDIO 583 | STBIWDEF int stbi_write_tga(char const *filename, int x, int y, int comp, const void *data) 584 | { 585 | stbi__write_context s; 586 | if (stbi__start_write_file(&s,filename)) { 587 | int r = stbi_write_tga_core(&s, x, y, comp, (void *) data); 588 | stbi__end_write_file(&s); 589 | return r; 590 | } else 591 | return 0; 592 | } 593 | #endif 594 | 595 | // ************************************************************************************************* 596 | // Radiance RGBE HDR writer 597 | // by Baldur Karlsson 598 | 599 | #define stbiw__max(a, b) ((a) > (b) ? (a) : (b)) 600 | 601 | static void stbiw__linear_to_rgbe(unsigned char *rgbe, float *linear) 602 | { 603 | int exponent; 604 | float maxcomp = stbiw__max(linear[0], stbiw__max(linear[1], linear[2])); 605 | 606 | if (maxcomp < 1e-32f) { 607 | rgbe[0] = rgbe[1] = rgbe[2] = rgbe[3] = 0; 608 | } else { 609 | float normalize = (float) frexp(maxcomp, &exponent) * 256.0f/maxcomp; 610 | 611 | rgbe[0] = (unsigned char)(linear[0] * normalize); 612 | rgbe[1] = (unsigned char)(linear[1] * normalize); 613 | rgbe[2] = (unsigned char)(linear[2] * normalize); 614 | rgbe[3] = (unsigned char)(exponent + 128); 615 | } 616 | } 617 | 618 | static void stbiw__write_run_data(stbi__write_context *s, int length, unsigned char databyte) 619 | { 620 | unsigned char lengthbyte = STBIW_UCHAR(length+128); 621 | STBIW_ASSERT(length+128 <= 255); 622 | s->func(s->context, &lengthbyte, 1); 623 | s->func(s->context, &databyte, 1); 624 | } 625 | 626 | static void stbiw__write_dump_data(stbi__write_context *s, int length, unsigned char *data) 627 | { 628 | unsigned char lengthbyte = STBIW_UCHAR(length); 629 | STBIW_ASSERT(length <= 128); // inconsistent with spec but consistent with official code 630 | s->func(s->context, &lengthbyte, 1); 631 | s->func(s->context, data, length); 632 | } 633 | 634 | static void stbiw__write_hdr_scanline(stbi__write_context *s, int width, int ncomp, unsigned char *scratch, float *scanline) 635 | { 636 | unsigned char scanlineheader[4] = { 2, 2, 0, 0 }; 637 | unsigned char rgbe[4]; 638 | float linear[3]; 639 | int x; 640 | 641 | scanlineheader[2] = (width&0xff00)>>8; 642 | scanlineheader[3] = (width&0x00ff); 643 | 644 | /* skip RLE for images too small or large */ 645 | if (width < 8 || width >= 32768) { 646 | for (x=0; x < width; x++) { 647 | switch (ncomp) { 648 | case 4: /* fallthrough */ 649 | case 3: linear[2] = scanline[x*ncomp + 2]; 650 | linear[1] = scanline[x*ncomp + 1]; 651 | linear[0] = scanline[x*ncomp + 0]; 652 | break; 653 | default: 654 | linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0]; 655 | break; 656 | } 657 | stbiw__linear_to_rgbe(rgbe, linear); 658 | s->func(s->context, rgbe, 4); 659 | } 660 | } else { 661 | int c,r; 662 | /* encode into scratch buffer */ 663 | for (x=0; x < width; x++) { 664 | switch(ncomp) { 665 | case 4: /* fallthrough */ 666 | case 3: linear[2] = scanline[x*ncomp + 2]; 667 | linear[1] = scanline[x*ncomp + 1]; 668 | linear[0] = scanline[x*ncomp + 0]; 669 | break; 670 | default: 671 | linear[0] = linear[1] = linear[2] = scanline[x*ncomp + 0]; 672 | break; 673 | } 674 | stbiw__linear_to_rgbe(rgbe, linear); 675 | scratch[x + width*0] = rgbe[0]; 676 | scratch[x + width*1] = rgbe[1]; 677 | scratch[x + width*2] = rgbe[2]; 678 | scratch[x + width*3] = rgbe[3]; 679 | } 680 | 681 | s->func(s->context, scanlineheader, 4); 682 | 683 | /* RLE each component separately */ 684 | for (c=0; c < 4; c++) { 685 | unsigned char *comp = &scratch[width*c]; 686 | 687 | x = 0; 688 | while (x < width) { 689 | // find first run 690 | r = x; 691 | while (r+2 < width) { 692 | if (comp[r] == comp[r+1] && comp[r] == comp[r+2]) 693 | break; 694 | ++r; 695 | } 696 | if (r+2 >= width) 697 | r = width; 698 | // dump up to first run 699 | while (x < r) { 700 | int len = r-x; 701 | if (len > 128) len = 128; 702 | stbiw__write_dump_data(s, len, &comp[x]); 703 | x += len; 704 | } 705 | // if there's a run, output it 706 | if (r+2 < width) { // same test as what we break out of in search loop, so only true if we break'd 707 | // find next byte after run 708 | while (r < width && comp[r] == comp[x]) 709 | ++r; 710 | // output run up to r 711 | while (x < r) { 712 | int len = r-x; 713 | if (len > 127) len = 127; 714 | stbiw__write_run_data(s, len, comp[x]); 715 | x += len; 716 | } 717 | } 718 | } 719 | } 720 | } 721 | } 722 | 723 | static int stbi_write_hdr_core(stbi__write_context *s, int x, int y, int comp, float *data) 724 | { 725 | if (y <= 0 || x <= 0 || data == NULL) 726 | return 0; 727 | else { 728 | // Each component is stored separately. Allocate scratch space for full output scanline. 729 | unsigned char *scratch = (unsigned char *) STBIW_MALLOC(x*4); 730 | int i, len; 731 | char buffer[128]; 732 | char header[] = "#?RADIANCE\n# Written by stb_image_write.h\nFORMAT=32-bit_rle_rgbe\n"; 733 | s->func(s->context, header, sizeof(header)-1); 734 | 735 | #ifdef __STDC_WANT_SECURE_LIB__ 736 | len = sprintf_s(buffer, sizeof(buffer), "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x); 737 | #else 738 | len = sprintf(buffer, "EXPOSURE= 1.0000000000000\n\n-Y %d +X %d\n", y, x); 739 | #endif 740 | s->func(s->context, buffer, len); 741 | 742 | for(i=0; i < y; i++) 743 | stbiw__write_hdr_scanline(s, x, comp, scratch, data + comp*x*(stbi__flip_vertically_on_write ? y-1-i : i)); 744 | STBIW_FREE(scratch); 745 | return 1; 746 | } 747 | } 748 | 749 | STBIWDEF int stbi_write_hdr_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const float *data) 750 | { 751 | stbi__write_context s; 752 | stbi__start_write_callbacks(&s, func, context); 753 | return stbi_write_hdr_core(&s, x, y, comp, (float *) data); 754 | } 755 | 756 | #ifndef STBI_WRITE_NO_STDIO 757 | STBIWDEF int stbi_write_hdr(char const *filename, int x, int y, int comp, const float *data) 758 | { 759 | stbi__write_context s; 760 | if (stbi__start_write_file(&s,filename)) { 761 | int r = stbi_write_hdr_core(&s, x, y, comp, (float *) data); 762 | stbi__end_write_file(&s); 763 | return r; 764 | } else 765 | return 0; 766 | } 767 | #endif // STBI_WRITE_NO_STDIO 768 | 769 | 770 | ////////////////////////////////////////////////////////////////////////////// 771 | // 772 | // PNG writer 773 | // 774 | 775 | #ifndef STBIW_ZLIB_COMPRESS 776 | // stretchy buffer; stbiw__sbpush() == vector<>::push_back() -- stbiw__sbcount() == vector<>::size() 777 | #define stbiw__sbraw(a) ((int *) (void *) (a) - 2) 778 | #define stbiw__sbm(a) stbiw__sbraw(a)[0] 779 | #define stbiw__sbn(a) stbiw__sbraw(a)[1] 780 | 781 | #define stbiw__sbneedgrow(a,n) ((a)==0 || stbiw__sbn(a)+n >= stbiw__sbm(a)) 782 | #define stbiw__sbmaybegrow(a,n) (stbiw__sbneedgrow(a,(n)) ? stbiw__sbgrow(a,n) : 0) 783 | #define stbiw__sbgrow(a,n) stbiw__sbgrowf((void **) &(a), (n), sizeof(*(a))) 784 | 785 | #define stbiw__sbpush(a, v) (stbiw__sbmaybegrow(a,1), (a)[stbiw__sbn(a)++] = (v)) 786 | #define stbiw__sbcount(a) ((a) ? stbiw__sbn(a) : 0) 787 | #define stbiw__sbfree(a) ((a) ? STBIW_FREE(stbiw__sbraw(a)),0 : 0) 788 | 789 | static void *stbiw__sbgrowf(void **arr, int increment, int itemsize) 790 | { 791 | int m = *arr ? 2*stbiw__sbm(*arr)+increment : increment+1; 792 | void *p = STBIW_REALLOC_SIZED(*arr ? stbiw__sbraw(*arr) : 0, *arr ? (stbiw__sbm(*arr)*itemsize + sizeof(int)*2) : 0, itemsize * m + sizeof(int)*2); 793 | STBIW_ASSERT(p); 794 | if (p) { 795 | if (!*arr) ((int *) p)[1] = 0; 796 | *arr = (void *) ((int *) p + 2); 797 | stbiw__sbm(*arr) = m; 798 | } 799 | return *arr; 800 | } 801 | 802 | static unsigned char *stbiw__zlib_flushf(unsigned char *data, unsigned int *bitbuffer, int *bitcount) 803 | { 804 | while (*bitcount >= 8) { 805 | stbiw__sbpush(data, STBIW_UCHAR(*bitbuffer)); 806 | *bitbuffer >>= 8; 807 | *bitcount -= 8; 808 | } 809 | return data; 810 | } 811 | 812 | static int stbiw__zlib_bitrev(int code, int codebits) 813 | { 814 | int res=0; 815 | while (codebits--) { 816 | res = (res << 1) | (code & 1); 817 | code >>= 1; 818 | } 819 | return res; 820 | } 821 | 822 | static unsigned int stbiw__zlib_countm(unsigned char *a, unsigned char *b, int limit) 823 | { 824 | int i; 825 | for (i=0; i < limit && i < 258; ++i) 826 | if (a[i] != b[i]) break; 827 | return i; 828 | } 829 | 830 | static unsigned int stbiw__zhash(unsigned char *data) 831 | { 832 | stbiw_uint32 hash = data[0] + (data[1] << 8) + (data[2] << 16); 833 | hash ^= hash << 3; 834 | hash += hash >> 5; 835 | hash ^= hash << 4; 836 | hash += hash >> 17; 837 | hash ^= hash << 25; 838 | hash += hash >> 6; 839 | return hash; 840 | } 841 | 842 | #define stbiw__zlib_flush() (out = stbiw__zlib_flushf(out, &bitbuf, &bitcount)) 843 | #define stbiw__zlib_add(code,codebits) \ 844 | (bitbuf |= (code) << bitcount, bitcount += (codebits), stbiw__zlib_flush()) 845 | #define stbiw__zlib_huffa(b,c) stbiw__zlib_add(stbiw__zlib_bitrev(b,c),c) 846 | // default huffman tables 847 | #define stbiw__zlib_huff1(n) stbiw__zlib_huffa(0x30 + (n), 8) 848 | #define stbiw__zlib_huff2(n) stbiw__zlib_huffa(0x190 + (n)-144, 9) 849 | #define stbiw__zlib_huff3(n) stbiw__zlib_huffa(0 + (n)-256,7) 850 | #define stbiw__zlib_huff4(n) stbiw__zlib_huffa(0xc0 + (n)-280,8) 851 | #define stbiw__zlib_huff(n) ((n) <= 143 ? stbiw__zlib_huff1(n) : (n) <= 255 ? stbiw__zlib_huff2(n) : (n) <= 279 ? stbiw__zlib_huff3(n) : stbiw__zlib_huff4(n)) 852 | #define stbiw__zlib_huffb(n) ((n) <= 143 ? stbiw__zlib_huff1(n) : stbiw__zlib_huff2(n)) 853 | 854 | #define stbiw__ZHASH 16384 855 | 856 | #endif // STBIW_ZLIB_COMPRESS 857 | 858 | STBIWDEF unsigned char * stbi_zlib_compress(unsigned char *data, int data_len, int *out_len, int quality) 859 | { 860 | #ifdef STBIW_ZLIB_COMPRESS 861 | // user provided a zlib compress implementation, use that 862 | return STBIW_ZLIB_COMPRESS(data, data_len, out_len, quality); 863 | #else // use builtin 864 | static unsigned short lengthc[] = { 3,4,5,6,7,8,9,10,11,13,15,17,19,23,27,31,35,43,51,59,67,83,99,115,131,163,195,227,258, 259 }; 865 | static unsigned char lengtheb[]= { 0,0,0,0,0,0,0, 0, 1, 1, 1, 1, 2, 2, 2, 2, 3, 3, 3, 3, 4, 4, 4, 4, 5, 5, 5, 5, 0 }; 866 | static unsigned short distc[] = { 1,2,3,4,5,7,9,13,17,25,33,49,65,97,129,193,257,385,513,769,1025,1537,2049,3073,4097,6145,8193,12289,16385,24577, 32768 }; 867 | static unsigned char disteb[] = { 0,0,0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,8,8,9,9,10,10,11,11,12,12,13,13 }; 868 | unsigned int bitbuf=0; 869 | int i,j, bitcount=0; 870 | unsigned char *out = NULL; 871 | unsigned char ***hash_table = (unsigned char***) STBIW_MALLOC(stbiw__ZHASH * sizeof(unsigned char**)); 872 | if (hash_table == NULL) 873 | return NULL; 874 | if (quality < 5) quality = 5; 875 | 876 | stbiw__sbpush(out, 0x78); // DEFLATE 32K window 877 | stbiw__sbpush(out, 0x5e); // FLEVEL = 1 878 | stbiw__zlib_add(1,1); // BFINAL = 1 879 | stbiw__zlib_add(1,2); // BTYPE = 1 -- fixed huffman 880 | 881 | for (i=0; i < stbiw__ZHASH; ++i) 882 | hash_table[i] = NULL; 883 | 884 | i=0; 885 | while (i < data_len-3) { 886 | // hash next 3 bytes of data to be compressed 887 | int h = stbiw__zhash(data+i)&(stbiw__ZHASH-1), best=3; 888 | unsigned char *bestloc = 0; 889 | unsigned char **hlist = hash_table[h]; 890 | int n = stbiw__sbcount(hlist); 891 | for (j=0; j < n; ++j) { 892 | if (hlist[j]-data > i-32768) { // if entry lies within window 893 | int d = stbiw__zlib_countm(hlist[j], data+i, data_len-i); 894 | if (d >= best) { best=d; bestloc=hlist[j]; } 895 | } 896 | } 897 | // when hash table entry is too long, delete half the entries 898 | if (hash_table[h] && stbiw__sbn(hash_table[h]) == 2*quality) { 899 | STBIW_MEMMOVE(hash_table[h], hash_table[h]+quality, sizeof(hash_table[h][0])*quality); 900 | stbiw__sbn(hash_table[h]) = quality; 901 | } 902 | stbiw__sbpush(hash_table[h],data+i); 903 | 904 | if (bestloc) { 905 | // "lazy matching" - check match at *next* byte, and if it's better, do cur byte as literal 906 | h = stbiw__zhash(data+i+1)&(stbiw__ZHASH-1); 907 | hlist = hash_table[h]; 908 | n = stbiw__sbcount(hlist); 909 | for (j=0; j < n; ++j) { 910 | if (hlist[j]-data > i-32767) { 911 | int e = stbiw__zlib_countm(hlist[j], data+i+1, data_len-i-1); 912 | if (e > best) { // if next match is better, bail on current match 913 | bestloc = NULL; 914 | break; 915 | } 916 | } 917 | } 918 | } 919 | 920 | if (bestloc) { 921 | int d = (int) (data+i - bestloc); // distance back 922 | STBIW_ASSERT(d <= 32767 && best <= 258); 923 | for (j=0; best > lengthc[j+1]-1; ++j); 924 | stbiw__zlib_huff(j+257); 925 | if (lengtheb[j]) stbiw__zlib_add(best - lengthc[j], lengtheb[j]); 926 | for (j=0; d > distc[j+1]-1; ++j); 927 | stbiw__zlib_add(stbiw__zlib_bitrev(j,5),5); 928 | if (disteb[j]) stbiw__zlib_add(d - distc[j], disteb[j]); 929 | i += best; 930 | } else { 931 | stbiw__zlib_huffb(data[i]); 932 | ++i; 933 | } 934 | } 935 | // write out final bytes 936 | for (;i < data_len; ++i) 937 | stbiw__zlib_huffb(data[i]); 938 | stbiw__zlib_huff(256); // end of block 939 | // pad with 0 bits to byte boundary 940 | while (bitcount) 941 | stbiw__zlib_add(0,1); 942 | 943 | for (i=0; i < stbiw__ZHASH; ++i) 944 | (void) stbiw__sbfree(hash_table[i]); 945 | STBIW_FREE(hash_table); 946 | 947 | { 948 | // compute adler32 on input 949 | unsigned int s1=1, s2=0; 950 | int blocklen = (int) (data_len % 5552); 951 | j=0; 952 | while (j < data_len) { 953 | for (i=0; i < blocklen; ++i) { s1 += data[j+i]; s2 += s1; } 954 | s1 %= 65521; s2 %= 65521; 955 | j += blocklen; 956 | blocklen = 5552; 957 | } 958 | stbiw__sbpush(out, STBIW_UCHAR(s2 >> 8)); 959 | stbiw__sbpush(out, STBIW_UCHAR(s2)); 960 | stbiw__sbpush(out, STBIW_UCHAR(s1 >> 8)); 961 | stbiw__sbpush(out, STBIW_UCHAR(s1)); 962 | } 963 | *out_len = stbiw__sbn(out); 964 | // make returned pointer freeable 965 | STBIW_MEMMOVE(stbiw__sbraw(out), out, *out_len); 966 | return (unsigned char *) stbiw__sbraw(out); 967 | #endif // STBIW_ZLIB_COMPRESS 968 | } 969 | 970 | static unsigned int stbiw__crc32(unsigned char *buffer, int len) 971 | { 972 | #ifdef STBIW_CRC32 973 | return STBIW_CRC32(buffer, len); 974 | #else 975 | static unsigned int crc_table[256] = 976 | { 977 | 0x00000000, 0x77073096, 0xEE0E612C, 0x990951BA, 0x076DC419, 0x706AF48F, 0xE963A535, 0x9E6495A3, 978 | 0x0eDB8832, 0x79DCB8A4, 0xE0D5E91E, 0x97D2D988, 0x09B64C2B, 0x7EB17CBD, 0xE7B82D07, 0x90BF1D91, 979 | 0x1DB71064, 0x6AB020F2, 0xF3B97148, 0x84BE41DE, 0x1ADAD47D, 0x6DDDE4EB, 0xF4D4B551, 0x83D385C7, 980 | 0x136C9856, 0x646BA8C0, 0xFD62F97A, 0x8A65C9EC, 0x14015C4F, 0x63066CD9, 0xFA0F3D63, 0x8D080DF5, 981 | 0x3B6E20C8, 0x4C69105E, 0xD56041E4, 0xA2677172, 0x3C03E4D1, 0x4B04D447, 0xD20D85FD, 0xA50AB56B, 982 | 0x35B5A8FA, 0x42B2986C, 0xDBBBC9D6, 0xACBCF940, 0x32D86CE3, 0x45DF5C75, 0xDCD60DCF, 0xABD13D59, 983 | 0x26D930AC, 0x51DE003A, 0xC8D75180, 0xBFD06116, 0x21B4F4B5, 0x56B3C423, 0xCFBA9599, 0xB8BDA50F, 984 | 0x2802B89E, 0x5F058808, 0xC60CD9B2, 0xB10BE924, 0x2F6F7C87, 0x58684C11, 0xC1611DAB, 0xB6662D3D, 985 | 0x76DC4190, 0x01DB7106, 0x98D220BC, 0xEFD5102A, 0x71B18589, 0x06B6B51F, 0x9FBFE4A5, 0xE8B8D433, 986 | 0x7807C9A2, 0x0F00F934, 0x9609A88E, 0xE10E9818, 0x7F6A0DBB, 0x086D3D2D, 0x91646C97, 0xE6635C01, 987 | 0x6B6B51F4, 0x1C6C6162, 0x856530D8, 0xF262004E, 0x6C0695ED, 0x1B01A57B, 0x8208F4C1, 0xF50FC457, 988 | 0x65B0D9C6, 0x12B7E950, 0x8BBEB8EA, 0xFCB9887C, 0x62DD1DDF, 0x15DA2D49, 0x8CD37CF3, 0xFBD44C65, 989 | 0x4DB26158, 0x3AB551CE, 0xA3BC0074, 0xD4BB30E2, 0x4ADFA541, 0x3DD895D7, 0xA4D1C46D, 0xD3D6F4FB, 990 | 0x4369E96A, 0x346ED9FC, 0xAD678846, 0xDA60B8D0, 0x44042D73, 0x33031DE5, 0xAA0A4C5F, 0xDD0D7CC9, 991 | 0x5005713C, 0x270241AA, 0xBE0B1010, 0xC90C2086, 0x5768B525, 0x206F85B3, 0xB966D409, 0xCE61E49F, 992 | 0x5EDEF90E, 0x29D9C998, 0xB0D09822, 0xC7D7A8B4, 0x59B33D17, 0x2EB40D81, 0xB7BD5C3B, 0xC0BA6CAD, 993 | 0xEDB88320, 0x9ABFB3B6, 0x03B6E20C, 0x74B1D29A, 0xEAD54739, 0x9DD277AF, 0x04DB2615, 0x73DC1683, 994 | 0xE3630B12, 0x94643B84, 0x0D6D6A3E, 0x7A6A5AA8, 0xE40ECF0B, 0x9309FF9D, 0x0A00AE27, 0x7D079EB1, 995 | 0xF00F9344, 0x8708A3D2, 0x1E01F268, 0x6906C2FE, 0xF762575D, 0x806567CB, 0x196C3671, 0x6E6B06E7, 996 | 0xFED41B76, 0x89D32BE0, 0x10DA7A5A, 0x67DD4ACC, 0xF9B9DF6F, 0x8EBEEFF9, 0x17B7BE43, 0x60B08ED5, 997 | 0xD6D6A3E8, 0xA1D1937E, 0x38D8C2C4, 0x4FDFF252, 0xD1BB67F1, 0xA6BC5767, 0x3FB506DD, 0x48B2364B, 998 | 0xD80D2BDA, 0xAF0A1B4C, 0x36034AF6, 0x41047A60, 0xDF60EFC3, 0xA867DF55, 0x316E8EEF, 0x4669BE79, 999 | 0xCB61B38C, 0xBC66831A, 0x256FD2A0, 0x5268E236, 0xCC0C7795, 0xBB0B4703, 0x220216B9, 0x5505262F, 1000 | 0xC5BA3BBE, 0xB2BD0B28, 0x2BB45A92, 0x5CB36A04, 0xC2D7FFA7, 0xB5D0CF31, 0x2CD99E8B, 0x5BDEAE1D, 1001 | 0x9B64C2B0, 0xEC63F226, 0x756AA39C, 0x026D930A, 0x9C0906A9, 0xEB0E363F, 0x72076785, 0x05005713, 1002 | 0x95BF4A82, 0xE2B87A14, 0x7BB12BAE, 0x0CB61B38, 0x92D28E9B, 0xE5D5BE0D, 0x7CDCEFB7, 0x0BDBDF21, 1003 | 0x86D3D2D4, 0xF1D4E242, 0x68DDB3F8, 0x1FDA836E, 0x81BE16CD, 0xF6B9265B, 0x6FB077E1, 0x18B74777, 1004 | 0x88085AE6, 0xFF0F6A70, 0x66063BCA, 0x11010B5C, 0x8F659EFF, 0xF862AE69, 0x616BFFD3, 0x166CCF45, 1005 | 0xA00AE278, 0xD70DD2EE, 0x4E048354, 0x3903B3C2, 0xA7672661, 0xD06016F7, 0x4969474D, 0x3E6E77DB, 1006 | 0xAED16A4A, 0xD9D65ADC, 0x40DF0B66, 0x37D83BF0, 0xA9BCAE53, 0xDEBB9EC5, 0x47B2CF7F, 0x30B5FFE9, 1007 | 0xBDBDF21C, 0xCABAC28A, 0x53B39330, 0x24B4A3A6, 0xBAD03605, 0xCDD70693, 0x54DE5729, 0x23D967BF, 1008 | 0xB3667A2E, 0xC4614AB8, 0x5D681B02, 0x2A6F2B94, 0xB40BBE37, 0xC30C8EA1, 0x5A05DF1B, 0x2D02EF8D 1009 | }; 1010 | 1011 | unsigned int crc = ~0u; 1012 | int i; 1013 | for (i=0; i < len; ++i) 1014 | crc = (crc >> 8) ^ crc_table[buffer[i] ^ (crc & 0xff)]; 1015 | return ~crc; 1016 | #endif 1017 | } 1018 | 1019 | #define stbiw__wpng4(o,a,b,c,d) ((o)[0]=STBIW_UCHAR(a),(o)[1]=STBIW_UCHAR(b),(o)[2]=STBIW_UCHAR(c),(o)[3]=STBIW_UCHAR(d),(o)+=4) 1020 | #define stbiw__wp32(data,v) stbiw__wpng4(data, (v)>>24,(v)>>16,(v)>>8,(v)); 1021 | #define stbiw__wptag(data,s) stbiw__wpng4(data, s[0],s[1],s[2],s[3]) 1022 | 1023 | static void stbiw__wpcrc(unsigned char **data, int len) 1024 | { 1025 | unsigned int crc = stbiw__crc32(*data - len - 4, len+4); 1026 | stbiw__wp32(*data, crc); 1027 | } 1028 | 1029 | static unsigned char stbiw__paeth(int a, int b, int c) 1030 | { 1031 | int p = a + b - c, pa = abs(p-a), pb = abs(p-b), pc = abs(p-c); 1032 | if (pa <= pb && pa <= pc) return STBIW_UCHAR(a); 1033 | if (pb <= pc) return STBIW_UCHAR(b); 1034 | return STBIW_UCHAR(c); 1035 | } 1036 | 1037 | // @OPTIMIZE: provide an option that always forces left-predict or paeth predict 1038 | static void stbiw__encode_png_line(unsigned char *pixels, int stride_bytes, int width, int height, int y, int n, int filter_type, signed char *line_buffer) 1039 | { 1040 | static int mapping[] = { 0,1,2,3,4 }; 1041 | static int firstmap[] = { 0,1,0,5,6 }; 1042 | int *mymap = (y != 0) ? mapping : firstmap; 1043 | int i; 1044 | int type = mymap[filter_type]; 1045 | unsigned char *z = pixels + stride_bytes * (stbi__flip_vertically_on_write ? height-1-y : y); 1046 | int signed_stride = stbi__flip_vertically_on_write ? -stride_bytes : stride_bytes; 1047 | 1048 | if (type==0) { 1049 | memcpy(line_buffer, z, width*n); 1050 | return; 1051 | } 1052 | 1053 | // first loop isn't optimized since it's just one pixel 1054 | for (i = 0; i < n; ++i) { 1055 | switch (type) { 1056 | case 1: line_buffer[i] = z[i]; break; 1057 | case 2: line_buffer[i] = z[i] - z[i-signed_stride]; break; 1058 | case 3: line_buffer[i] = z[i] - (z[i-signed_stride]>>1); break; 1059 | case 4: line_buffer[i] = (signed char) (z[i] - stbiw__paeth(0,z[i-signed_stride],0)); break; 1060 | case 5: line_buffer[i] = z[i]; break; 1061 | case 6: line_buffer[i] = z[i]; break; 1062 | } 1063 | } 1064 | switch (type) { 1065 | case 1: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-n]; break; 1066 | case 2: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - z[i-signed_stride]; break; 1067 | case 3: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - ((z[i-n] + z[i-signed_stride])>>1); break; 1068 | case 4: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], z[i-signed_stride], z[i-signed_stride-n]); break; 1069 | case 5: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - (z[i-n]>>1); break; 1070 | case 6: for (i=n; i < width*n; ++i) line_buffer[i] = z[i] - stbiw__paeth(z[i-n], 0,0); break; 1071 | } 1072 | } 1073 | 1074 | STBIWDEF unsigned char *stbi_write_png_to_mem(const unsigned char *pixels, int stride_bytes, int x, int y, int n, int *out_len) 1075 | { 1076 | int force_filter = stbi_write_force_png_filter; 1077 | int ctype[5] = { -1, 0, 4, 2, 6 }; 1078 | unsigned char sig[8] = { 137,80,78,71,13,10,26,10 }; 1079 | unsigned char *out,*o, *filt, *zlib; 1080 | signed char *line_buffer; 1081 | int j,zlen; 1082 | 1083 | if (stride_bytes == 0) 1084 | stride_bytes = x * n; 1085 | 1086 | if (force_filter >= 5) { 1087 | force_filter = -1; 1088 | } 1089 | 1090 | filt = (unsigned char *) STBIW_MALLOC((x*n+1) * y); if (!filt) return 0; 1091 | line_buffer = (signed char *) STBIW_MALLOC(x * n); if (!line_buffer) { STBIW_FREE(filt); return 0; } 1092 | for (j=0; j < y; ++j) { 1093 | int filter_type; 1094 | if (force_filter > -1) { 1095 | filter_type = force_filter; 1096 | stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, force_filter, line_buffer); 1097 | } else { // Estimate the best filter by running through all of them: 1098 | int best_filter = 0, best_filter_val = 0x7fffffff, est, i; 1099 | for (filter_type = 0; filter_type < 5; filter_type++) { 1100 | stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, filter_type, line_buffer); 1101 | 1102 | // Estimate the entropy of the line using this filter; the less, the better. 1103 | est = 0; 1104 | for (i = 0; i < x*n; ++i) { 1105 | est += abs((signed char) line_buffer[i]); 1106 | } 1107 | if (est < best_filter_val) { 1108 | best_filter_val = est; 1109 | best_filter = filter_type; 1110 | } 1111 | } 1112 | if (filter_type != best_filter) { // If the last iteration already got us the best filter, don't redo it 1113 | stbiw__encode_png_line((unsigned char*)(pixels), stride_bytes, x, y, j, n, best_filter, line_buffer); 1114 | filter_type = best_filter; 1115 | } 1116 | } 1117 | // when we get here, filter_type contains the filter type, and line_buffer contains the data 1118 | filt[j*(x*n+1)] = (unsigned char) filter_type; 1119 | STBIW_MEMMOVE(filt+j*(x*n+1)+1, line_buffer, x*n); 1120 | } 1121 | STBIW_FREE(line_buffer); 1122 | zlib = stbi_zlib_compress(filt, y*( x*n+1), &zlen, stbi_write_png_compression_level); 1123 | STBIW_FREE(filt); 1124 | if (!zlib) return 0; 1125 | 1126 | // each tag requires 12 bytes of overhead 1127 | out = (unsigned char *) STBIW_MALLOC(8 + 12+13 + 12+zlen + 12); 1128 | if (!out) return 0; 1129 | *out_len = 8 + 12+13 + 12+zlen + 12; 1130 | 1131 | o=out; 1132 | STBIW_MEMMOVE(o,sig,8); o+= 8; 1133 | stbiw__wp32(o, 13); // header length 1134 | stbiw__wptag(o, "IHDR"); 1135 | stbiw__wp32(o, x); 1136 | stbiw__wp32(o, y); 1137 | *o++ = 8; 1138 | *o++ = STBIW_UCHAR(ctype[n]); 1139 | *o++ = 0; 1140 | *o++ = 0; 1141 | *o++ = 0; 1142 | stbiw__wpcrc(&o,13); 1143 | 1144 | stbiw__wp32(o, zlen); 1145 | stbiw__wptag(o, "IDAT"); 1146 | STBIW_MEMMOVE(o, zlib, zlen); 1147 | o += zlen; 1148 | STBIW_FREE(zlib); 1149 | stbiw__wpcrc(&o, zlen); 1150 | 1151 | stbiw__wp32(o,0); 1152 | stbiw__wptag(o, "IEND"); 1153 | stbiw__wpcrc(&o,0); 1154 | 1155 | STBIW_ASSERT(o == out + *out_len); 1156 | 1157 | return out; 1158 | } 1159 | 1160 | #ifndef STBI_WRITE_NO_STDIO 1161 | STBIWDEF int stbi_write_png(char const *filename, int x, int y, int comp, const void *data, int stride_bytes) 1162 | { 1163 | FILE *f; 1164 | int len; 1165 | unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len); 1166 | if (png == NULL) return 0; 1167 | 1168 | f = stbiw__fopen(filename, "wb"); 1169 | if (!f) { STBIW_FREE(png); return 0; } 1170 | fwrite(png, 1, len, f); 1171 | fclose(f); 1172 | STBIW_FREE(png); 1173 | return 1; 1174 | } 1175 | #endif 1176 | 1177 | STBIWDEF int stbi_write_png_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int stride_bytes) 1178 | { 1179 | int len; 1180 | unsigned char *png = stbi_write_png_to_mem((const unsigned char *) data, stride_bytes, x, y, comp, &len); 1181 | if (png == NULL) return 0; 1182 | func(context, png, len); 1183 | STBIW_FREE(png); 1184 | return 1; 1185 | } 1186 | 1187 | 1188 | /* *************************************************************************** 1189 | * 1190 | * JPEG writer 1191 | * 1192 | * This is based on Jon Olick's jo_jpeg.cpp: 1193 | * public domain Simple, Minimalistic JPEG writer - http://www.jonolick.com/code.html 1194 | */ 1195 | 1196 | static const unsigned char stbiw__jpg_ZigZag[] = { 0,1,5,6,14,15,27,28,2,4,7,13,16,26,29,42,3,8,12,17,25,30,41,43,9,11,18, 1197 | 24,31,40,44,53,10,19,23,32,39,45,52,54,20,22,33,38,46,51,55,60,21,34,37,47,50,56,59,61,35,36,48,49,57,58,62,63 }; 1198 | 1199 | static void stbiw__jpg_writeBits(stbi__write_context *s, int *bitBufP, int *bitCntP, const unsigned short *bs) { 1200 | int bitBuf = *bitBufP, bitCnt = *bitCntP; 1201 | bitCnt += bs[1]; 1202 | bitBuf |= bs[0] << (24 - bitCnt); 1203 | while(bitCnt >= 8) { 1204 | unsigned char c = (bitBuf >> 16) & 255; 1205 | stbiw__putc(s, c); 1206 | if(c == 255) { 1207 | stbiw__putc(s, 0); 1208 | } 1209 | bitBuf <<= 8; 1210 | bitCnt -= 8; 1211 | } 1212 | *bitBufP = bitBuf; 1213 | *bitCntP = bitCnt; 1214 | } 1215 | 1216 | static void stbiw__jpg_DCT(float *d0p, float *d1p, float *d2p, float *d3p, float *d4p, float *d5p, float *d6p, float *d7p) { 1217 | float d0 = *d0p, d1 = *d1p, d2 = *d2p, d3 = *d3p, d4 = *d4p, d5 = *d5p, d6 = *d6p, d7 = *d7p; 1218 | float z1, z2, z3, z4, z5, z11, z13; 1219 | 1220 | float tmp0 = d0 + d7; 1221 | float tmp7 = d0 - d7; 1222 | float tmp1 = d1 + d6; 1223 | float tmp6 = d1 - d6; 1224 | float tmp2 = d2 + d5; 1225 | float tmp5 = d2 - d5; 1226 | float tmp3 = d3 + d4; 1227 | float tmp4 = d3 - d4; 1228 | 1229 | // Even part 1230 | float tmp10 = tmp0 + tmp3; // phase 2 1231 | float tmp13 = tmp0 - tmp3; 1232 | float tmp11 = tmp1 + tmp2; 1233 | float tmp12 = tmp1 - tmp2; 1234 | 1235 | d0 = tmp10 + tmp11; // phase 3 1236 | d4 = tmp10 - tmp11; 1237 | 1238 | z1 = (tmp12 + tmp13) * 0.707106781f; // c4 1239 | d2 = tmp13 + z1; // phase 5 1240 | d6 = tmp13 - z1; 1241 | 1242 | // Odd part 1243 | tmp10 = tmp4 + tmp5; // phase 2 1244 | tmp11 = tmp5 + tmp6; 1245 | tmp12 = tmp6 + tmp7; 1246 | 1247 | // The rotator is modified from fig 4-8 to avoid extra negations. 1248 | z5 = (tmp10 - tmp12) * 0.382683433f; // c6 1249 | z2 = tmp10 * 0.541196100f + z5; // c2-c6 1250 | z4 = tmp12 * 1.306562965f + z5; // c2+c6 1251 | z3 = tmp11 * 0.707106781f; // c4 1252 | 1253 | z11 = tmp7 + z3; // phase 5 1254 | z13 = tmp7 - z3; 1255 | 1256 | *d5p = z13 + z2; // phase 6 1257 | *d3p = z13 - z2; 1258 | *d1p = z11 + z4; 1259 | *d7p = z11 - z4; 1260 | 1261 | *d0p = d0; *d2p = d2; *d4p = d4; *d6p = d6; 1262 | } 1263 | 1264 | static void stbiw__jpg_calcBits(int val, unsigned short bits[2]) { 1265 | int tmp1 = val < 0 ? -val : val; 1266 | val = val < 0 ? val-1 : val; 1267 | bits[1] = 1; 1268 | while(tmp1 >>= 1) { 1269 | ++bits[1]; 1270 | } 1271 | bits[0] = val & ((1<0)&&(DU[end0pos]==0); --end0pos) { 1314 | } 1315 | // end0pos = first element in reverse order !=0 1316 | if(end0pos == 0) { 1317 | stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB); 1318 | return DU[0]; 1319 | } 1320 | for(i = 1; i <= end0pos; ++i) { 1321 | int startpos = i; 1322 | int nrzeroes; 1323 | unsigned short bits[2]; 1324 | for (; DU[i]==0 && i<=end0pos; ++i) { 1325 | } 1326 | nrzeroes = i-startpos; 1327 | if ( nrzeroes >= 16 ) { 1328 | int lng = nrzeroes>>4; 1329 | int nrmarker; 1330 | for (nrmarker=1; nrmarker <= lng; ++nrmarker) 1331 | stbiw__jpg_writeBits(s, bitBuf, bitCnt, M16zeroes); 1332 | nrzeroes &= 15; 1333 | } 1334 | stbiw__jpg_calcBits(DU[i], bits); 1335 | stbiw__jpg_writeBits(s, bitBuf, bitCnt, HTAC[(nrzeroes<<4)+bits[1]]); 1336 | stbiw__jpg_writeBits(s, bitBuf, bitCnt, bits); 1337 | } 1338 | if(end0pos != 63) { 1339 | stbiw__jpg_writeBits(s, bitBuf, bitCnt, EOB); 1340 | } 1341 | return DU[0]; 1342 | } 1343 | 1344 | static int stbi_write_jpg_core(stbi__write_context *s, int width, int height, int comp, const void* data, int quality) { 1345 | // Constants that don't pollute global namespace 1346 | static const unsigned char std_dc_luminance_nrcodes[] = {0,0,1,5,1,1,1,1,1,1,0,0,0,0,0,0,0}; 1347 | static const unsigned char std_dc_luminance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11}; 1348 | static const unsigned char std_ac_luminance_nrcodes[] = {0,0,2,1,3,3,2,4,3,5,5,4,4,0,0,1,0x7d}; 1349 | static const unsigned char std_ac_luminance_values[] = { 1350 | 0x01,0x02,0x03,0x00,0x04,0x11,0x05,0x12,0x21,0x31,0x41,0x06,0x13,0x51,0x61,0x07,0x22,0x71,0x14,0x32,0x81,0x91,0xa1,0x08, 1351 | 0x23,0x42,0xb1,0xc1,0x15,0x52,0xd1,0xf0,0x24,0x33,0x62,0x72,0x82,0x09,0x0a,0x16,0x17,0x18,0x19,0x1a,0x25,0x26,0x27,0x28, 1352 | 0x29,0x2a,0x34,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58,0x59, 1353 | 0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x83,0x84,0x85,0x86,0x87,0x88,0x89, 1354 | 0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4,0xb5,0xb6, 1355 | 0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda,0xe1,0xe2, 1356 | 0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf1,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa 1357 | }; 1358 | static const unsigned char std_dc_chrominance_nrcodes[] = {0,0,3,1,1,1,1,1,1,1,1,1,0,0,0,0,0}; 1359 | static const unsigned char std_dc_chrominance_values[] = {0,1,2,3,4,5,6,7,8,9,10,11}; 1360 | static const unsigned char std_ac_chrominance_nrcodes[] = {0,0,2,1,2,4,4,3,4,7,5,4,4,0,1,2,0x77}; 1361 | static const unsigned char std_ac_chrominance_values[] = { 1362 | 0x00,0x01,0x02,0x03,0x11,0x04,0x05,0x21,0x31,0x06,0x12,0x41,0x51,0x07,0x61,0x71,0x13,0x22,0x32,0x81,0x08,0x14,0x42,0x91, 1363 | 0xa1,0xb1,0xc1,0x09,0x23,0x33,0x52,0xf0,0x15,0x62,0x72,0xd1,0x0a,0x16,0x24,0x34,0xe1,0x25,0xf1,0x17,0x18,0x19,0x1a,0x26, 1364 | 0x27,0x28,0x29,0x2a,0x35,0x36,0x37,0x38,0x39,0x3a,0x43,0x44,0x45,0x46,0x47,0x48,0x49,0x4a,0x53,0x54,0x55,0x56,0x57,0x58, 1365 | 0x59,0x5a,0x63,0x64,0x65,0x66,0x67,0x68,0x69,0x6a,0x73,0x74,0x75,0x76,0x77,0x78,0x79,0x7a,0x82,0x83,0x84,0x85,0x86,0x87, 1366 | 0x88,0x89,0x8a,0x92,0x93,0x94,0x95,0x96,0x97,0x98,0x99,0x9a,0xa2,0xa3,0xa4,0xa5,0xa6,0xa7,0xa8,0xa9,0xaa,0xb2,0xb3,0xb4, 1367 | 0xb5,0xb6,0xb7,0xb8,0xb9,0xba,0xc2,0xc3,0xc4,0xc5,0xc6,0xc7,0xc8,0xc9,0xca,0xd2,0xd3,0xd4,0xd5,0xd6,0xd7,0xd8,0xd9,0xda, 1368 | 0xe2,0xe3,0xe4,0xe5,0xe6,0xe7,0xe8,0xe9,0xea,0xf2,0xf3,0xf4,0xf5,0xf6,0xf7,0xf8,0xf9,0xfa 1369 | }; 1370 | // Huffman tables 1371 | static const unsigned short YDC_HT[256][2] = { {0,2},{2,3},{3,3},{4,3},{5,3},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9}}; 1372 | static const unsigned short UVDC_HT[256][2] = { {0,2},{1,2},{2,2},{6,3},{14,4},{30,5},{62,6},{126,7},{254,8},{510,9},{1022,10},{2046,11}}; 1373 | static const unsigned short YAC_HT[256][2] = { 1374 | {10,4},{0,2},{1,2},{4,3},{11,4},{26,5},{120,7},{248,8},{1014,10},{65410,16},{65411,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1375 | {12,4},{27,5},{121,7},{502,9},{2038,11},{65412,16},{65413,16},{65414,16},{65415,16},{65416,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1376 | {28,5},{249,8},{1015,10},{4084,12},{65417,16},{65418,16},{65419,16},{65420,16},{65421,16},{65422,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1377 | {58,6},{503,9},{4085,12},{65423,16},{65424,16},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1378 | {59,6},{1016,10},{65430,16},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1379 | {122,7},{2039,11},{65438,16},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1380 | {123,7},{4086,12},{65446,16},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1381 | {250,8},{4087,12},{65454,16},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1382 | {504,9},{32704,15},{65462,16},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1383 | {505,9},{65470,16},{65471,16},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1384 | {506,9},{65479,16},{65480,16},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1385 | {1017,10},{65488,16},{65489,16},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1386 | {1018,10},{65497,16},{65498,16},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1387 | {2040,11},{65506,16},{65507,16},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1388 | {65515,16},{65516,16},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{0,0},{0,0},{0,0},{0,0},{0,0}, 1389 | {2041,11},{65525,16},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0} 1390 | }; 1391 | static const unsigned short UVAC_HT[256][2] = { 1392 | {0,2},{1,2},{4,3},{10,4},{24,5},{25,5},{56,6},{120,7},{500,9},{1014,10},{4084,12},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1393 | {11,4},{57,6},{246,8},{501,9},{2038,11},{4085,12},{65416,16},{65417,16},{65418,16},{65419,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1394 | {26,5},{247,8},{1015,10},{4086,12},{32706,15},{65420,16},{65421,16},{65422,16},{65423,16},{65424,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1395 | {27,5},{248,8},{1016,10},{4087,12},{65425,16},{65426,16},{65427,16},{65428,16},{65429,16},{65430,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1396 | {58,6},{502,9},{65431,16},{65432,16},{65433,16},{65434,16},{65435,16},{65436,16},{65437,16},{65438,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1397 | {59,6},{1017,10},{65439,16},{65440,16},{65441,16},{65442,16},{65443,16},{65444,16},{65445,16},{65446,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1398 | {121,7},{2039,11},{65447,16},{65448,16},{65449,16},{65450,16},{65451,16},{65452,16},{65453,16},{65454,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1399 | {122,7},{2040,11},{65455,16},{65456,16},{65457,16},{65458,16},{65459,16},{65460,16},{65461,16},{65462,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1400 | {249,8},{65463,16},{65464,16},{65465,16},{65466,16},{65467,16},{65468,16},{65469,16},{65470,16},{65471,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1401 | {503,9},{65472,16},{65473,16},{65474,16},{65475,16},{65476,16},{65477,16},{65478,16},{65479,16},{65480,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1402 | {504,9},{65481,16},{65482,16},{65483,16},{65484,16},{65485,16},{65486,16},{65487,16},{65488,16},{65489,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1403 | {505,9},{65490,16},{65491,16},{65492,16},{65493,16},{65494,16},{65495,16},{65496,16},{65497,16},{65498,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1404 | {506,9},{65499,16},{65500,16},{65501,16},{65502,16},{65503,16},{65504,16},{65505,16},{65506,16},{65507,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1405 | {2041,11},{65508,16},{65509,16},{65510,16},{65511,16},{65512,16},{65513,16},{65514,16},{65515,16},{65516,16},{0,0},{0,0},{0,0},{0,0},{0,0},{0,0}, 1406 | {16352,14},{65517,16},{65518,16},{65519,16},{65520,16},{65521,16},{65522,16},{65523,16},{65524,16},{65525,16},{0,0},{0,0},{0,0},{0,0},{0,0}, 1407 | {1018,10},{32707,15},{65526,16},{65527,16},{65528,16},{65529,16},{65530,16},{65531,16},{65532,16},{65533,16},{65534,16},{0,0},{0,0},{0,0},{0,0},{0,0} 1408 | }; 1409 | static const int YQT[] = {16,11,10,16,24,40,51,61,12,12,14,19,26,58,60,55,14,13,16,24,40,57,69,56,14,17,22,29,51,87,80,62,18,22, 1410 | 37,56,68,109,103,77,24,35,55,64,81,104,113,92,49,64,78,87,103,121,120,101,72,92,95,98,112,100,103,99}; 1411 | static const int UVQT[] = {17,18,24,47,99,99,99,99,18,21,26,66,99,99,99,99,24,26,56,99,99,99,99,99,47,66,99,99,99,99,99,99, 1412 | 99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99,99}; 1413 | static const float aasf[] = { 1.0f * 2.828427125f, 1.387039845f * 2.828427125f, 1.306562965f * 2.828427125f, 1.175875602f * 2.828427125f, 1414 | 1.0f * 2.828427125f, 0.785694958f * 2.828427125f, 0.541196100f * 2.828427125f, 0.275899379f * 2.828427125f }; 1415 | 1416 | int row, col, i, k, subsample; 1417 | float fdtbl_Y[64], fdtbl_UV[64]; 1418 | unsigned char YTable[64], UVTable[64]; 1419 | 1420 | if(!data || !width || !height || comp > 4 || comp < 1) { 1421 | return 0; 1422 | } 1423 | 1424 | quality = quality ? quality : 90; 1425 | subsample = quality <= 90 ? 1 : 0; 1426 | quality = quality < 1 ? 1 : quality > 100 ? 100 : quality; 1427 | quality = quality < 50 ? 5000 / quality : 200 - quality * 2; 1428 | 1429 | for(i = 0; i < 64; ++i) { 1430 | int uvti, yti = (YQT[i]*quality+50)/100; 1431 | YTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (yti < 1 ? 1 : yti > 255 ? 255 : yti); 1432 | uvti = (UVQT[i]*quality+50)/100; 1433 | UVTable[stbiw__jpg_ZigZag[i]] = (unsigned char) (uvti < 1 ? 1 : uvti > 255 ? 255 : uvti); 1434 | } 1435 | 1436 | for(row = 0, k = 0; row < 8; ++row) { 1437 | for(col = 0; col < 8; ++col, ++k) { 1438 | fdtbl_Y[k] = 1 / (YTable [stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]); 1439 | fdtbl_UV[k] = 1 / (UVTable[stbiw__jpg_ZigZag[k]] * aasf[row] * aasf[col]); 1440 | } 1441 | } 1442 | 1443 | // Write Headers 1444 | { 1445 | static const unsigned char head0[] = { 0xFF,0xD8,0xFF,0xE0,0,0x10,'J','F','I','F',0,1,1,0,0,1,0,1,0,0,0xFF,0xDB,0,0x84,0 }; 1446 | static const unsigned char head2[] = { 0xFF,0xDA,0,0xC,3,1,0,2,0x11,3,0x11,0,0x3F,0 }; 1447 | const unsigned char head1[] = { 0xFF,0xC0,0,0x11,8,(unsigned char)(height>>8),STBIW_UCHAR(height),(unsigned char)(width>>8),STBIW_UCHAR(width), 1448 | 3,1,(unsigned char)(subsample?0x22:0x11),0,2,0x11,1,3,0x11,1,0xFF,0xC4,0x01,0xA2,0 }; 1449 | s->func(s->context, (void*)head0, sizeof(head0)); 1450 | s->func(s->context, (void*)YTable, sizeof(YTable)); 1451 | stbiw__putc(s, 1); 1452 | s->func(s->context, UVTable, sizeof(UVTable)); 1453 | s->func(s->context, (void*)head1, sizeof(head1)); 1454 | s->func(s->context, (void*)(std_dc_luminance_nrcodes+1), sizeof(std_dc_luminance_nrcodes)-1); 1455 | s->func(s->context, (void*)std_dc_luminance_values, sizeof(std_dc_luminance_values)); 1456 | stbiw__putc(s, 0x10); // HTYACinfo 1457 | s->func(s->context, (void*)(std_ac_luminance_nrcodes+1), sizeof(std_ac_luminance_nrcodes)-1); 1458 | s->func(s->context, (void*)std_ac_luminance_values, sizeof(std_ac_luminance_values)); 1459 | stbiw__putc(s, 1); // HTUDCinfo 1460 | s->func(s->context, (void*)(std_dc_chrominance_nrcodes+1), sizeof(std_dc_chrominance_nrcodes)-1); 1461 | s->func(s->context, (void*)std_dc_chrominance_values, sizeof(std_dc_chrominance_values)); 1462 | stbiw__putc(s, 0x11); // HTUACinfo 1463 | s->func(s->context, (void*)(std_ac_chrominance_nrcodes+1), sizeof(std_ac_chrominance_nrcodes)-1); 1464 | s->func(s->context, (void*)std_ac_chrominance_values, sizeof(std_ac_chrominance_values)); 1465 | s->func(s->context, (void*)head2, sizeof(head2)); 1466 | } 1467 | 1468 | // Encode 8x8 macroblocks 1469 | { 1470 | static const unsigned short fillBits[] = {0x7F, 7}; 1471 | int DCY=0, DCU=0, DCV=0; 1472 | int bitBuf=0, bitCnt=0; 1473 | // comp == 2 is grey+alpha (alpha is ignored) 1474 | int ofsG = comp > 2 ? 1 : 0, ofsB = comp > 2 ? 2 : 0; 1475 | const unsigned char *dataR = (const unsigned char *)data; 1476 | const unsigned char *dataG = dataR + ofsG; 1477 | const unsigned char *dataB = dataR + ofsB; 1478 | int x, y, pos; 1479 | if(subsample) { 1480 | for(y = 0; y < height; y += 16) { 1481 | for(x = 0; x < width; x += 16) { 1482 | float Y[256], U[256], V[256]; 1483 | for(row = y, pos = 0; row < y+16; ++row) { 1484 | // row >= height => use last input row 1485 | int clamped_row = (row < height) ? row : height - 1; 1486 | int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp; 1487 | for(col = x; col < x+16; ++col, ++pos) { 1488 | // if col >= width => use pixel from last input column 1489 | int p = base_p + ((col < width) ? col : (width-1))*comp; 1490 | float r = dataR[p], g = dataG[p], b = dataB[p]; 1491 | Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128; 1492 | U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b; 1493 | V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b; 1494 | } 1495 | } 1496 | DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+0, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT); 1497 | DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+8, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT); 1498 | DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+128, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT); 1499 | DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y+136, 16, fdtbl_Y, DCY, YDC_HT, YAC_HT); 1500 | 1501 | // subsample U,V 1502 | { 1503 | float subU[64], subV[64]; 1504 | int yy, xx; 1505 | for(yy = 0, pos = 0; yy < 8; ++yy) { 1506 | for(xx = 0; xx < 8; ++xx, ++pos) { 1507 | int j = yy*32+xx*2; 1508 | subU[pos] = (U[j+0] + U[j+1] + U[j+16] + U[j+17]) * 0.25f; 1509 | subV[pos] = (V[j+0] + V[j+1] + V[j+16] + V[j+17]) * 0.25f; 1510 | } 1511 | } 1512 | DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subU, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT); 1513 | DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, subV, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT); 1514 | } 1515 | } 1516 | } 1517 | } else { 1518 | for(y = 0; y < height; y += 8) { 1519 | for(x = 0; x < width; x += 8) { 1520 | float Y[64], U[64], V[64]; 1521 | for(row = y, pos = 0; row < y+8; ++row) { 1522 | // row >= height => use last input row 1523 | int clamped_row = (row < height) ? row : height - 1; 1524 | int base_p = (stbi__flip_vertically_on_write ? (height-1-clamped_row) : clamped_row)*width*comp; 1525 | for(col = x; col < x+8; ++col, ++pos) { 1526 | // if col >= width => use pixel from last input column 1527 | int p = base_p + ((col < width) ? col : (width-1))*comp; 1528 | float r = dataR[p], g = dataG[p], b = dataB[p]; 1529 | Y[pos]= +0.29900f*r + 0.58700f*g + 0.11400f*b - 128; 1530 | U[pos]= -0.16874f*r - 0.33126f*g + 0.50000f*b; 1531 | V[pos]= +0.50000f*r - 0.41869f*g - 0.08131f*b; 1532 | } 1533 | } 1534 | 1535 | DCY = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, Y, 8, fdtbl_Y, DCY, YDC_HT, YAC_HT); 1536 | DCU = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, U, 8, fdtbl_UV, DCU, UVDC_HT, UVAC_HT); 1537 | DCV = stbiw__jpg_processDU(s, &bitBuf, &bitCnt, V, 8, fdtbl_UV, DCV, UVDC_HT, UVAC_HT); 1538 | } 1539 | } 1540 | } 1541 | 1542 | // Do the bit alignment of the EOI marker 1543 | stbiw__jpg_writeBits(s, &bitBuf, &bitCnt, fillBits); 1544 | } 1545 | 1546 | // EOI 1547 | stbiw__putc(s, 0xFF); 1548 | stbiw__putc(s, 0xD9); 1549 | 1550 | return 1; 1551 | } 1552 | 1553 | STBIWDEF int stbi_write_jpg_to_func(stbi_write_func *func, void *context, int x, int y, int comp, const void *data, int quality) 1554 | { 1555 | stbi__write_context s; 1556 | stbi__start_write_callbacks(&s, func, context); 1557 | return stbi_write_jpg_core(&s, x, y, comp, (void *) data, quality); 1558 | } 1559 | 1560 | 1561 | #ifndef STBI_WRITE_NO_STDIO 1562 | STBIWDEF int stbi_write_jpg(char const *filename, int x, int y, int comp, const void *data, int quality) 1563 | { 1564 | stbi__write_context s; 1565 | if (stbi__start_write_file(&s,filename)) { 1566 | int r = stbi_write_jpg_core(&s, x, y, comp, data, quality); 1567 | stbi__end_write_file(&s); 1568 | return r; 1569 | } else 1570 | return 0; 1571 | } 1572 | #endif 1573 | 1574 | #endif // STB_IMAGE_WRITE_IMPLEMENTATION 1575 | 1576 | /* Revision history 1577 | 1.14 (2020-02-02) updated JPEG writer to downsample chroma channels 1578 | 1.13 1579 | 1.12 1580 | 1.11 (2019-08-11) 1581 | 1582 | 1.10 (2019-02-07) 1583 | support utf8 filenames in Windows; fix warnings and platform ifdefs 1584 | 1.09 (2018-02-11) 1585 | fix typo in zlib quality API, improve STB_I_W_STATIC in C++ 1586 | 1.08 (2018-01-29) 1587 | add stbi__flip_vertically_on_write, external zlib, zlib quality, choose PNG filter 1588 | 1.07 (2017-07-24) 1589 | doc fix 1590 | 1.06 (2017-07-23) 1591 | writing JPEG (using Jon Olick's code) 1592 | 1.05 ??? 1593 | 1.04 (2017-03-03) 1594 | monochrome BMP expansion 1595 | 1.03 ??? 1596 | 1.02 (2016-04-02) 1597 | avoid allocating large structures on the stack 1598 | 1.01 (2016-01-16) 1599 | STBIW_REALLOC_SIZED: support allocators with no realloc support 1600 | avoid race-condition in crc initialization 1601 | minor compile issues 1602 | 1.00 (2015-09-14) 1603 | installable file IO function 1604 | 0.99 (2015-09-13) 1605 | warning fixes; TGA rle support 1606 | 0.98 (2015-04-08) 1607 | added STBIW_MALLOC, STBIW_ASSERT etc 1608 | 0.97 (2015-01-18) 1609 | fixed HDR asserts, rewrote HDR rle logic 1610 | 0.96 (2015-01-17) 1611 | add HDR output 1612 | fix monochrome BMP 1613 | 0.95 (2014-08-17) 1614 | add monochrome TGA output 1615 | 0.94 (2014-05-31) 1616 | rename private functions to avoid conflicts with stb_image.h 1617 | 0.93 (2014-05-27) 1618 | warning fixes 1619 | 0.92 (2010-08-01) 1620 | casts to unsigned char to fix warnings 1621 | 0.91 (2010-07-17) 1622 | first public release 1623 | 0.90 first internal release 1624 | */ 1625 | 1626 | /* 1627 | ------------------------------------------------------------------------------ 1628 | This software is available under 2 licenses -- choose whichever you prefer. 1629 | ------------------------------------------------------------------------------ 1630 | ALTERNATIVE A - MIT License 1631 | Copyright (c) 2017 Sean Barrett 1632 | Permission is hereby granted, free of charge, to any person obtaining a copy of 1633 | this software and associated documentation files (the "Software"), to deal in 1634 | the Software without restriction, including without limitation the rights to 1635 | use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies 1636 | of the Software, and to permit persons to whom the Software is furnished to do 1637 | so, subject to the following conditions: 1638 | The above copyright notice and this permission notice shall be included in all 1639 | copies or substantial portions of the Software. 1640 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 1641 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 1642 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 1643 | AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER 1644 | LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, 1645 | OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE 1646 | SOFTWARE. 1647 | ------------------------------------------------------------------------------ 1648 | ALTERNATIVE B - Public Domain (www.unlicense.org) 1649 | This is free and unencumbered software released into the public domain. 1650 | Anyone is free to copy, modify, publish, use, compile, sell, or distribute this 1651 | software, either in source code form or as a compiled binary, for any purpose, 1652 | commercial or non-commercial, and by any means. 1653 | In jurisdictions that recognize copyright laws, the author or authors of this 1654 | software dedicate any and all copyright interest in the software to the public 1655 | domain. We make this dedication for the benefit of the public at large and to 1656 | the detriment of our heirs and successors. We intend this dedication to be an 1657 | overt act of relinquishment in perpetuity of all present and future rights to 1658 | this software under copyright law. 1659 | THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR 1660 | IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, 1661 | FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE 1662 | AUTHORS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN 1663 | ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION 1664 | WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE. 1665 | ------------------------------------------------------------------------------ 1666 | */ 1667 | -------------------------------------------------------------------------------- /Luma.sln: -------------------------------------------------------------------------------- 1 |  2 | Microsoft Visual Studio Solution File, Format Version 12.00 3 | # Visual Studio Version 16 4 | VisualStudioVersion = 16.0.29905.134 5 | MinimumVisualStudioVersion = 10.0.40219.1 6 | Project("{8BC9CEB8-8B4A-11D0-8D11-00A0C91BC942}") = "Luma", "Luma.vcxproj", "{7D2B5C1C-FDA0-464E-87D2-B333BA0C5886}" 7 | EndProject 8 | Global 9 | GlobalSection(SolutionConfigurationPlatforms) = preSolution 10 | Debug|x64 = Debug|x64 11 | Release|x64 = Release|x64 12 | EndGlobalSection 13 | GlobalSection(ProjectConfigurationPlatforms) = postSolution 14 | {7D2B5C1C-FDA0-464E-87D2-B333BA0C5886}.Debug|x64.ActiveCfg = Debug|x64 15 | {7D2B5C1C-FDA0-464E-87D2-B333BA0C5886}.Debug|x64.Build.0 = Debug|x64 16 | {7D2B5C1C-FDA0-464E-87D2-B333BA0C5886}.Release|x64.ActiveCfg = Release|x64 17 | {7D2B5C1C-FDA0-464E-87D2-B333BA0C5886}.Release|x64.Build.0 = Release|x64 18 | EndGlobalSection 19 | GlobalSection(SolutionProperties) = preSolution 20 | HideSolutionNode = FALSE 21 | EndGlobalSection 22 | GlobalSection(ExtensibilityGlobals) = postSolution 23 | SolutionGuid = {3DDA0FC8-6979-44CA-8BE9-83B34ACF31EE} 24 | EndGlobalSection 25 | EndGlobal 26 | -------------------------------------------------------------------------------- /Luma.vcxproj: -------------------------------------------------------------------------------- 1 | 2 | 3 | 4 | 5 | Debug 6 | x64 7 | 8 | 9 | Release 10 | x64 11 | 12 | 13 | 14 | 16.0 15 | {7D2B5C1C-FDA0-464E-87D2-B333BA0C5886} 16 | Luma 17 | 10.0 18 | 19 | 20 | 21 | Application 22 | true 23 | v142 24 | Unicode 25 | 26 | 27 | Application 28 | false 29 | v142 30 | true 31 | Unicode 32 | 33 | 34 | 35 | 36 | 37 | 38 | 39 | 40 | 41 | 42 | 43 | 44 | 45 | 46 | true 47 | $(SolutionDir)Build\$(Configuration)\ 48 | $(SolutionDir)Build\$(Configuration)\$(ProjectName)\ 49 | 50 | 51 | false 52 | $(SolutionDir)Build\$(Configuration)\ 53 | $(SolutionDir)Build\$(Configuration)\$(ProjectName)\ 54 | 55 | 56 | 57 | Level3 58 | true 59 | _DEBUG;_CONSOLE;%(PreprocessorDefinitions) 60 | true 61 | Externals 62 | Use 63 | pch.h 64 | 65 | 66 | Console 67 | true 68 | 69 | 70 | 71 | 72 | Level3 73 | true 74 | true 75 | true 76 | NDEBUG;_CONSOLE;%(PreprocessorDefinitions) 77 | true 78 | Externals 79 | Use 80 | pch.h 81 | 82 | 83 | Console 84 | true 85 | true 86 | true 87 | 88 | 89 | 90 | 91 | 92 | Create 93 | Create 94 | 95 | 96 | 97 | 98 | 99 | 100 | 101 | 102 | 103 | 104 | 105 | 106 | 107 | 108 | 109 | -------------------------------------------------------------------------------- /Luma.vcxproj.filters: -------------------------------------------------------------------------------- 1 |  2 | 3 | 4 | 5 | {4FC737F1-C7A5-4376-A066-2A32D752A2FF} 6 | cpp;c;cc;cxx;def;odl;idl;hpj;bat;asm;asmx 7 | 8 | 9 | {93995380-89BD-4b04-88EB-625FBE52EBFB} 10 | h;hh;hpp;hxx;hm;inl;inc;ipp;xsd 11 | 12 | 13 | {67DA6AB6-F800-4c08-8B7A-83BB121AAD01} 14 | rc;ico;cur;bmp;dlg;rc2;rct;bin;rgs;gif;jpg;jpeg;jpe;resx;tiff;tif;png;wav;mfcribbon-ms 15 | 16 | 17 | 18 | 19 | Source Files 20 | 21 | 22 | Source Files 23 | 24 | 25 | 26 | 27 | Header Files 28 | 29 | 30 | Header Files 31 | 32 | 33 | Header Files 34 | 35 | 36 | Header Files 37 | 38 | 39 | Header Files 40 | 41 | 42 | Header Files 43 | 44 | 45 | Header Files 46 | 47 | 48 | Header Files 49 | 50 | 51 | -------------------------------------------------------------------------------- /README.md: -------------------------------------------------------------------------------- 1 | # Luma 2 | 3 | This is a simple ray tracer which uses [Ray Tracing in One Weekend](https://raytracing.github.io/books/RayTracingInOneWeekend.html) as a starting point. This code has (many) more comments than the code on that site, so I hope you can learn from it as well. 4 | 5 | Confession time: I am using this as a way to (finally!) write a ray tracer from scratch. I have modified ray tracers, integrated ray tracers, and used ray tracing frameworks (like DirectX Raytracing), but somehow never wrote a ray tracer from scratch, not even in school. 6 | 7 | This is written as a Visual Studio 2019 project. Just open the solution file, and build the Debug or Release configuration. The Release configuration is _much_ faster here. 8 | 9 | Currently the code covers up to and including section 8 of _Ray Tracing in One Weekend_, "Diffuse Materials." It will render the image below (or one close to it, depending on settings). 10 | 11 | ![Sample Image](Doc/sample.png) 12 | _Rendered at 960x540 with 128 samples per pixel in 7.5 seconds on a Core i7-8700 CPU with 12 threads._ 13 | -------------------------------------------------------------------------------- /Source/Camera.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | #include "Ray.h" 4 | #include "Vec3.h" 5 | 6 | namespace Luma { 7 | 8 | // A camera for generating primary rays for rendering. 9 | class Camera 10 | { 11 | public: 12 | // Constructor. 13 | Camera(float aspect) : m_aspect(aspect) {} 14 | 15 | // Computes a ray from the camera with the specified U (horizontal) and V (vertical) offsets in 16 | // the camera image plane. 17 | Ray getRay(float u, float v) const 18 | { 19 | // Define the camera with an eye position at the origin, and facing the -Z axis. 20 | // TODO: The camera is currently hardcoded, but will later support standard properties. 21 | static Vec3 origin(0.0f, 0.0f, 0.0f); 22 | static Vec3 start(-m_aspect, -1.0f, -1.0f); 23 | static float viewWidth = m_aspect * 2.0f; 24 | static float viewHeight = 2.0f; 25 | 26 | // Prepare a ray direction offset from the start (lower left) corner of the image plane. 27 | Vec3 direction( 28 | start.x() + u * viewWidth, 29 | start.y() + v * viewHeight, 30 | start.z()); 31 | direction.normalize(); 32 | 33 | // Return a ray with the direction and the camera's origin. 34 | return Ray(origin, direction); 35 | } 36 | 37 | private: 38 | float m_aspect; 39 | }; 40 | 41 | } // namespace Luma 42 | -------------------------------------------------------------------------------- /Source/Image.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | #define STB_IMAGE_WRITE_IMPLEMENTATION 4 | #include "stb_image/stb_image_write.h" 5 | 6 | namespace Luma { 7 | 8 | class Image 9 | { 10 | public: 11 | // Constructor. 12 | Image(uint16_t width, uint16_t height) : m_width(width), m_height(height) 13 | { 14 | // Create the image buffer. 15 | size_t bufferSize = m_width * m_height * NUM_COMPONENTS; 16 | m_pImageData = new uint8_t[bufferSize]; 17 | } 18 | 19 | // Destructor. 20 | ~Image() 21 | { 22 | delete[] m_pImageData; 23 | m_pImageData = nullptr; 24 | } 25 | 26 | // Returns the image data buffer. 27 | uint8_t* getImageData() { return m_pImageData; } 28 | 29 | // Saves the image as a PNG file to the specified path, with an optional scale to enlarge the 30 | // image. 31 | void savePNG(string sFilePath, uint8_t scale = 1) 32 | { 33 | uint8_t* pImageData = m_pImageData; 34 | uint16_t width = m_width; 35 | uint16_t height = m_height; 36 | 37 | // Resize the image by the specified scale if one is specified. 38 | if (scale > 1) 39 | { 40 | pImageData = scaleImage(scale); 41 | width *= scale; 42 | height *= scale; 43 | } 44 | 45 | // Write the image buffer to the output file path, and delete the image buffer. 46 | ::stbi_write_png( 47 | sFilePath.c_str(), width, height, NUM_COMPONENTS, pImageData, width * NUM_COMPONENTS); 48 | 49 | // Delete the scaled image buffer if it was created. 50 | if (scale > 1) 51 | { 52 | delete[] pImageData; 53 | } 54 | } 55 | 56 | private: 57 | static const uint8_t NUM_COMPONENTS = 3; 58 | 59 | uint8_t* m_pImageData = nullptr; 60 | uint16_t m_width; 61 | uint16_t m_height; 62 | 63 | // Scales (enlarges) an image buffer by the specified scale factor, returning a new buffer. 64 | uint8_t* scaleImage(uint8_t scale) 65 | { 66 | // Create the destination buffer, as a multiple of the source buffer, e.g. 240x135 with a 67 | // scale of 8 becomes 1920x1080. 68 | uint16_t destWidth = m_width * scale; 69 | uint16_t destHeight = m_height * scale; 70 | size_t destBufferSize = destWidth * destHeight * NUM_COMPONENTS; 71 | uint8_t* pDestData = new uint8_t[destBufferSize]; 72 | 73 | // Iterate the destination image pixels, copying the appropriate source image pixel for each. 74 | uint8_t* pDst = pDestData; 75 | for (uint16_t y = 0; y < destHeight; y++) 76 | { 77 | // Get the start of the relevant scanline from the source image, i.e. advance the source 78 | // scanline every "scale" scanlines (e.g. 8) of the destination image. 79 | const uint8_t* pSrc = &m_pImageData[(y / scale) * m_width * NUM_COMPONENTS]; 80 | 81 | // Iterate the pixels of the destination image, copying the pixels of source image. The 82 | // source image pointer is advanced every "scale" (e.g. 8) pixels of the destination image. 83 | for (uint16_t x = 0; x < destWidth; x++) 84 | { 85 | ::memcpy(pDst, pSrc, NUM_COMPONENTS); 86 | pDst += NUM_COMPONENTS; 87 | pSrc += (x + 1) % scale == 0 ? NUM_COMPONENTS : 0; 88 | } 89 | } 90 | 91 | return pDestData; 92 | } 93 | }; 94 | 95 | } // namespace Luma 96 | -------------------------------------------------------------------------------- /Source/Ray.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | #include "Vec3.h" 4 | 5 | namespace Luma { 6 | 7 | // Ray with origin, direction, and min/max range. 8 | class Ray 9 | { 10 | public: 11 | // Constructor. 12 | Ray(const Vec3& origin, const Vec3& direction, float tMin = 0.0f, float tMax = INF) : 13 | m_origin(origin), m_direction(direction), m_tMin(tMin), m_tMax(tMax) {} 14 | 15 | // Returns the origin (start point) of the ray. 16 | const Vec3& origin() const { return m_origin; } 17 | 18 | // Returns the direction of the ray. 19 | const Vec3& direction() const { return m_direction; } 20 | 21 | // Returns the minimum distance from the origin for ray intersections. 22 | // 23 | // NOTE: The distance is a multiple of the direction length. 24 | const float tMin() const { return m_tMin; } 25 | 26 | // Returns the maximum distance from the origin for ray intersections. 27 | // 28 | // NOTE: The distance is a multiple of the direction length. 29 | const float tMax() const { return m_tMax; } 30 | 31 | // Computes a point along the ray at the specified distance. 32 | // 33 | // NOTE: The distance is a multiple of the direction length. 34 | const Vec3 at(float t) const { return m_origin + m_direction * t; } 35 | 36 | private: 37 | Vec3 m_origin; 38 | Vec3 m_direction; 39 | float m_tMin; 40 | float m_tMax; 41 | }; 42 | 43 | } // namespace Luma 44 | -------------------------------------------------------------------------------- /Source/Scene.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | #include "Utils.h" 4 | 5 | namespace Luma { 6 | 7 | // A structure storing the data for a hit (ray-element intersection). 8 | struct Hit 9 | { 10 | float t; 11 | Vec3 position; 12 | Vec3 normal; 13 | }; 14 | 15 | // An interface for any element that can be intersected by a ray. 16 | class Element 17 | { 18 | public: 19 | // Intersects the ray with the element, returns whether an intersection was found. If so, the 20 | // hit value is update with properties of the intersection. 21 | virtual bool intersect(const Ray& ray, Hit& hit) const = 0; 22 | }; 23 | 24 | // A scene consisting of multiple elements suitable for rendering. 25 | class Scene : public Element 26 | { 27 | public: 28 | // Adds an element to the scene. 29 | void add(shared_ptr pElement) 30 | { 31 | m_elements.push_back(pElement); 32 | } 33 | 34 | // Overrides Element.Intersect(). 35 | virtual bool intersect(const Ray& ray, Hit& hit) const override 36 | { 37 | // Initialize the closest hit with the rays TMax value. 38 | bool anyHit = false; 39 | Hit closestHit; 40 | closestHit.t = ray.tMax(); 41 | 42 | // Iterate the elements, finding the closest intersection with the ray. 43 | for (auto element : m_elements) 44 | { 45 | // If the ray intersects the element, and the hit is closer that the closest one so far, 46 | // record it as the closest hit. 47 | Hit nextHit; 48 | if (element->intersect(ray, nextHit) && nextHit.t < closestHit.t) 49 | { 50 | anyHit = true; 51 | closestHit = nextHit; 52 | } 53 | } 54 | 55 | // If there was a hit, record that for the caller. 56 | if (anyHit) 57 | { 58 | assert(closestHit.t <= ray.tMax()); 59 | hit = closestHit; 60 | } 61 | 62 | return anyHit; 63 | } 64 | 65 | private: 66 | vector> m_elements; 67 | }; 68 | 69 | } // namespace Luma 70 | -------------------------------------------------------------------------------- /Source/Sphere.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | #include "Ray.h" 4 | #include "Vec3.h" 5 | #include "Scene.h" 6 | 7 | namespace Luma { 8 | 9 | // A sphere with a center and radius. 10 | class Sphere : public Element 11 | { 12 | public: 13 | // Constructor. 14 | Sphere(const Vec3& center, float radius) : m_center(center), m_radius(radius) {} 15 | 16 | // Override's Element.Intersect(). 17 | virtual bool intersect(const Ray& ray, Hit& hit) const override 18 | { 19 | // Compute the components needed to solve the quadratic equation, using the quadratic 20 | // formula: (-b ± √(b² - 4ac)) / 2a. 21 | // 22 | // NOTE: Search online for "ray sphere intersection" for a complete derivation. 23 | Vec3 delta = ray.origin() - m_center; 24 | float a = dot(ray.direction(), ray.direction()); 25 | float b = 2.0f * dot(ray.direction(), delta); 26 | float c = dot(delta, delta) - m_radius * m_radius; 27 | 28 | // Compute the discriminant. If the value is less than zero, then a square root is not 29 | // possible and the ray misses the sphere. 30 | float discriminant = b * b - 4 * a * c; 31 | if (discriminant < 0.0f) 32 | { 33 | return false; 34 | } 35 | 36 | // Evaluate the quadratic formula to get the closer possible intersection point (lower t). 37 | // If it exceeds the TMax value, the sphere is beyond the ray. If it is less than the TMin 38 | // value, try the farther possible intersection point (higher t). If this is still not in 39 | // the ray bounds, return false. 40 | float t = (-b - sqrt(discriminant)) / (2.0f * a); 41 | if (t > ray.tMax()) 42 | { 43 | return false; 44 | } 45 | else if (t < ray.tMin()) 46 | { 47 | t = (-b + sqrt(discriminant)) / (2.0f * a); 48 | if (t < ray.tMin() || t > ray.tMax()) 49 | { 50 | return false; 51 | } 52 | } 53 | 54 | // The sphere was hit, so update the hit record with the t parameter, hit position, and 55 | // (normalized) normal at the hit position. 56 | hit.t = t; 57 | hit.position = ray.at(t); 58 | hit.normal = (hit.position - m_center) / m_radius; 59 | 60 | return true; 61 | } 62 | 63 | private: 64 | Vec3 m_center; 65 | float m_radius; 66 | }; 67 | 68 | } // namespace Luma 69 | -------------------------------------------------------------------------------- /Source/Utils.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | namespace Luma { 4 | 5 | // Linearly interpolates between two values. 6 | template 7 | inline T lerp(T a, T b, float t) 8 | { 9 | return a + t * (b - a); 10 | } 11 | 12 | // Clamps a value to the specified range. 13 | template 14 | inline T clamp(T x, T min, T max) 15 | { 16 | if (x < min) return min; 17 | if (x > max) return max; 18 | return x; 19 | } 20 | 21 | // Generates a random 32-bit integer from a seed value, using an LCG. 22 | // 23 | // NOTE: Based on http://www.reedbeta.com/blog/quick-and-easy-gpu-random-numbers-in-d3d11. 24 | uint32_t randomLCG(uint32_t seed) 25 | { 26 | return 1664525 * seed + 1013904223; 27 | } 28 | 29 | // Generates a random 32-bit integer from a seed value, using XOR shifts. 30 | // 31 | // NOTE: Based on http://www.reedbeta.com/blog/quick-and-easy-gpu-random-numbers-in-d3d11. 32 | uint32_t randomXORShift(uint32_t x) 33 | { 34 | x ^= (x << 13); 35 | x ^= (x >> 17); 36 | x ^= (x << 5); 37 | 38 | return x; 39 | } 40 | 41 | // Hashes a 32-bit integer, which can be used to randomize a seed for an RNG, or directly as an RNG. 42 | // 43 | // NOTE: Based on http://www.reedbeta.com/blog/quick-and-easy-gpu-random-numbers-in-d3d11. 44 | uint32_t wangHash(uint32_t x) 45 | { 46 | x = (x ^ 61) ^ (x >> 16); 47 | x *= 9; 48 | x ^= x >> 4; 49 | x *= 0x27d4eb2dU; 50 | x ^= x >> 15; 51 | 52 | return x; 53 | } 54 | 55 | // Hashes a 32-bit integer, which can be used to randomize a seed for an RNG, or directly as an RNG. 56 | // 57 | // NOTE: Based on https://nullprogram.com/blog/2018/07/31. 58 | uint32_t lowBias32Hash(uint32_t x) 59 | { 60 | x ^= x >> 16; 61 | x *= 0x7feb352dU; 62 | x ^= x >> 15; 63 | x *= 0x846ca68bU; 64 | x ^= x >> 16; 65 | 66 | return x; 67 | } 68 | 69 | // Computes the entry in the base 2 Halton sequence at the specified index. 70 | // 71 | // NOTE: Based on PBRT at https://github.com/mmp/pbrt-v3/blob/master/src/core/lowdiscrepancy.h. 72 | inline float halton2(uint32_t index) 73 | { 74 | index = (index << 16) | (index >> 16); 75 | index = ((index & 0x00ff00ff) << 8) | ((index & 0xff00ff00) >> 8); 76 | index = ((index & 0x0f0f0f0f) << 4) | ((index & 0xf0f0f0f0) >> 4); 77 | index = ((index & 0x33333333) << 2) | ((index & 0xcccccccc) >> 2); 78 | index = ((index & 0x55555555) << 1) | ((index & 0xaaaaaaaa) >> 1); 79 | 80 | return index * 1.0f / (1ull << 32); 81 | } 82 | 83 | // Computes the entry in the base 3 Halton sequence at the specified index. 84 | inline float halton3(uint32_t index) 85 | { 86 | float result = 0.0f; 87 | float scale = 1.0f; 88 | while (index != 0) 89 | { 90 | scale /= 3; 91 | result += (index % 3) * scale; 92 | index /= 3; 93 | } 94 | 95 | return result; 96 | } 97 | 98 | // Generates a uniformly distributed pseudorandom random number in the range [0.0, 1.0) using the 99 | // Mersenne Twister. This gives very high quality numbers. 100 | float randomMT() 101 | { 102 | static std::mt19937 generator; 103 | static std::uniform_real_distribution distribution; 104 | static std::function func = std::bind(distribution, generator); 105 | 106 | return func(); 107 | } 108 | 109 | // Get two uniformly distributed quasirandom numbers in the range [0.0, 1.0), using Halton (2,3) 110 | // sequences with the specified index. 111 | // 112 | // NOTE: The use of *quasirandom* (low discrepancy) numbers can substantially improve the rate of 113 | // convergence for path tracing, compared to *pseudorandom* numbers. Try using MT nubmers here to 114 | // see the difference. See PBRT and https://en.wikipedia.org/wiki/Halton_sequence for more 115 | // information. 116 | inline void getRandom2D(float& u1, float& u2, uint32_t& index) 117 | { 118 | u1 = halton2(index); // randomMT(); 119 | u2 = halton3(index); // randomMT(); 120 | } 121 | 122 | // Generates a random direction in the cosine-weighted hemisphere above the specified normal. This 123 | // provides a PDF value ("probability density function") which is the *relative* probability that 124 | // the returned direction will be chosen. 125 | Vec3 randomDirection(float u1, float u2, const Vec3& normal, float& pdf) 126 | { 127 | // Create a point on the unit sphere, i.e. a direction, from the uniform random variables. 128 | float cosTheta = 1.0f - 2.0f * u2; 129 | float sinTheta = sqrt(1.0f - cosTheta * cosTheta); 130 | float phi = 2.0f * PI * u1; 131 | Vec3 direction(sinTheta * cos(phi), sinTheta * sin(phi), cosTheta); 132 | 133 | // To transform that into a sample from a cosine-weighted hemisphere over the normal, treat the 134 | // sphere as tangent to the surface: add the normal to the direction and normalize it. The PDF 135 | // is cos(theta) / PI, so use a dot product to compute cos(theta). See "Ray Tracing in One 136 | // Weekend" for details. 137 | direction = (normal + direction).normalize(); 138 | pdf = dot(normal, direction) / PI; 139 | 140 | // To test a uniform hemisphere instead, comment the block above and uncomment the block below. 141 | // This simply flips the uniform sphere direction if it is on the opposite side of the normal. 142 | // While this is still correct, for cosine-weighted operations like ambient occlusion or 143 | // computing outgoing radiance, this requires about twice as many samples to achieve the same 144 | // level of variance (noise) as the cosine-weighted direction computed above. See "Ray Tracing 145 | // Gems" chapter 15 for details. 146 | // 147 | // if (Dot(normal, direction) < 0.0) 148 | // { 149 | // direction = -direction; 150 | // } 151 | // pdf = 1.0f / (2.0f * M_PI_F); 152 | 153 | return direction; 154 | } 155 | 156 | // Reports the specified progress on the console, as a progress bar. 157 | void updateProgress(float progress) 158 | { 159 | const int PROGRESS_SIZE = 75; 160 | std::cout << "["; 161 | for (int i = 0; i < PROGRESS_SIZE; i++) 162 | { 163 | std::cout << (static_cast(i) / PROGRESS_SIZE < progress ? "#" : " "); 164 | } 165 | std::cout << "] " << static_cast(progress * 100.0f) << "%\r" << std::flush; 166 | } 167 | 168 | } // namespace Luma 169 | -------------------------------------------------------------------------------- /Source/Vec3.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | namespace Luma { 4 | 5 | // Vector with three components. 6 | class Vec3 7 | { 8 | public: 9 | // Constructors. 10 | Vec3(): m_val{ 0.0f, 0.0f, 0.0f } {} 11 | Vec3(float x, float y, float z) { m_val[0] = x; m_val[1] = y; m_val[2] = z; } 12 | 13 | // Accessors. 14 | float x() const { return m_val[0]; } 15 | float y() const { return m_val[1]; } 16 | float z() const { return m_val[2]; } 17 | float r() const { return m_val[0]; } 18 | float g() const { return m_val[1]; } 19 | float b() const { return m_val[2]; } 20 | 21 | // Operator overloads. 22 | Vec3 operator-() const { return Vec3(-m_val[0], -m_val[1], -m_val[2]); } 23 | Vec3& operator+=(const Vec3& a) { m_val[0] += a.m_val[0]; m_val[1] += a.m_val[1]; m_val[2] += a.m_val[2]; return *this; } 24 | Vec3& operator-=(const Vec3& a) { m_val[0] -= a.m_val[0]; m_val[1] -= a.m_val[1]; m_val[2] -= a.m_val[2]; return *this; } 25 | Vec3& operator*=(float a) { m_val[0] *= a; m_val[1] *= a; m_val[2] *= a; return *this; } 26 | Vec3& operator/=(float a) { m_val[0] /= a; m_val[1] /= a; m_val[2] /= a; return *this; } 27 | 28 | // Computes the length of the vector. 29 | float length() const { return sqrt(x() * x() + y() * y() + z() * z()); } 30 | 31 | // Normalizes the vector, i.e. with unit length. 32 | Vec3& normalize() { *this /= length(); return *this; } 33 | 34 | // Linearizes the vector (as a color) in the sRGB color space. 35 | // 36 | // NOTE: Colors should be linearized for rendering computations to work correctly. Linearization 37 | // has the effect of darkening the color. See this chapter for "GPU Gems 3" for details: 38 | // https://developer.nvidia.com/gpugems/gpugems3/part-iv-image-effects/chapter-24-importance-being-linear 39 | Vec3& sRGBToLinear() 40 | { 41 | static const float LINEARIZE = 2.2f; 42 | m_val[0] = pow(m_val[0], LINEARIZE); 43 | m_val[1] = pow(m_val[1], LINEARIZE); 44 | m_val[2] = pow(m_val[2], LINEARIZE); 45 | return *this; 46 | } 47 | 48 | // Gamma corrects the vector (as a color) in the sRGB color space. 49 | // 50 | // NOTE: Colors computed in rendering (linearized) should be gamma corrected immediately before 51 | // display or saving to most image file formats. Gamma correction has the effect of darkening 52 | // the color. 53 | Vec3& linearTosRGB() 54 | { 55 | static const float GAMMA = 1 / 2.2f; 56 | m_val[0] = pow(m_val[0], GAMMA); 57 | m_val[1] = pow(m_val[1], GAMMA); 58 | m_val[2] = pow(m_val[2], GAMMA); 59 | return *this; 60 | } 61 | 62 | private: 63 | float m_val[3]; 64 | }; 65 | 66 | // Computes the dot product of two vectors. 67 | inline float dot(const Vec3& a, const Vec3& b) 68 | { 69 | return a.x() * b.x() + a.y() * b.y() + a.z() * b.z(); 70 | } 71 | 72 | // Overloads the + operator for two vectors. 73 | inline Vec3 operator+(const Vec3& a, const Vec3& b) 74 | { 75 | return Vec3(a.x() + b.x(), a.y() + b.y(), a.z() + b.z()); 76 | } 77 | 78 | // Overloads the - operator for two vectors. 79 | inline Vec3 operator-(const Vec3& a, const Vec3& b) 80 | { 81 | return Vec3(a.x() - b.x(), a.y() - b.y(), a.z() - b.z()); 82 | } 83 | 84 | // Overloads the * operator for two vectors. 85 | inline Vec3 operator*(const Vec3& a, const Vec3& b) 86 | { 87 | return Vec3(a.x() * b.x(), a.y() * b.y(), a.z() * b.z()); 88 | } 89 | 90 | // Overloads the * operator for a scalar and a vector. 91 | inline Vec3 operator*(float a, const Vec3& b) 92 | { 93 | return Vec3(a * b.x(), a * b.y(), a * b.z()); 94 | } 95 | 96 | // Overloads the * operator for a vector and a scalar. 97 | inline Vec3 operator*(const Vec3& a, float b) 98 | { 99 | return Vec3(a.x() * b, a.y() * b, a.z() * b); 100 | } 101 | 102 | // Overloads the / operator for a vector and a scalar. 103 | inline Vec3 operator/(const Vec3& a, float b) 104 | { 105 | return Vec3(a.x() / b, a.y() / b, a.z() / b); 106 | } 107 | 108 | } // namespace Luma 109 | -------------------------------------------------------------------------------- /Source/main.cpp: -------------------------------------------------------------------------------- 1 | #include "pch.h" 2 | 3 | #include "Camera.h" 4 | #include "Image.h" 5 | #include "Ray.h" 6 | #include "Scene.h" 7 | #include "Sphere.h" 8 | #include "Vec3.h" 9 | #include "Utils.h" 10 | using namespace Luma; 11 | 12 | #include 13 | 14 | // Computes the radiance incident along the specified ray, for the specified element. 15 | Vec3 radiance(const Ray& ray, const Element& element, int depth, uint32_t& index) 16 | { 17 | // If the trace depth has been exhausted, simply return black. 18 | if (depth == 0) 19 | { 20 | return Vec3(); 21 | } 22 | 23 | // Intersect the sphere with the ray, and shade with the hit record if there was an 24 | // intersection. Otherwise shade with a (vertical) background gradient. 25 | Vec3 radiance; 26 | Hit hit; 27 | if (element.intersect(ray, hit)) 28 | { 29 | // Generate a random direction in the hemisphere above the normal. 30 | float u1 = 0.0f, u2 = 0.0f; 31 | float pdf = 1.0f; 32 | getRandom2D(u1, u2, index); 33 | Vec3 direction = randomDirection(u1, u2, hit.normal, pdf); 34 | float cosTheta = dot(hit.normal, direction); 35 | assert(cosTheta > 0.0f); 36 | 37 | // Compute the Lambertian BRDF, i.e. the amount of light reflected by the material. 38 | static const Vec3 materialColor(Vec3(0.75f, 0.75f, 0.75f).sRGBToLinear()); 39 | Vec3 brdf = materialColor / PI; 40 | 41 | // Compute the radiance incident from the direction, i.e. the incident light. 42 | // 43 | // NOTE: As this is recursive, this renders global illumination (indirect light) which is 44 | // very difficult to achieve with rasterization on GPUs. 45 | // 46 | // NOTE: A small ray offset is used to avoid self-intersection. 47 | static const float RAY_OFFSET = 1e-4f; 48 | Ray ray(hit.position, direction, RAY_OFFSET); 49 | Vec3 light = ::radiance(ray, element, depth - 1, index); 50 | 51 | // Compute the outgoing radiance, as defined by the rendering equation. 52 | radiance = brdf * light * cosTheta / pdf; 53 | 54 | // DIRECT LIGHTING: Uncomment this to perform simple direct shading and shadowing with a 55 | // directional light. As there is no random sampling, this will have no noise. 56 | // 57 | // Hit shadowHit; 58 | // static const Vec3 lightDirection(Vec3(1.0f, 1.0f, 1.0f).normalize()); 59 | // Ray shadowRay(hit.position, lightDirection, RAY_OFFSET); 60 | // float visibility = element.intersect(shadowRay, shadowHit) ? 0.1f : 1.0f; 61 | // radiance = brdf * visibility * std::max(dot(hit.normal, lightDirection), 0.0f); 62 | 63 | // AMBIENT OCCLUSION: Uncomment this to render ambient occlusion, i.e. the amount by which a 64 | // point can see the environment. 65 | // 66 | // Vec3 visibility = element.intersect(ray, hit) ? Vec3() : Vec3(1.0f, 1.0f, 1.0f); 67 | // radiance = visibility * cosTheta / PI / pdf; 68 | 69 | // NORMALS: Uncomment this to render the surface normals as colors. 70 | // 71 | // radiance = (0.5f * (hit.normal + Vec3(1.0f, 1.0f, 1.0f))).Linearize(); 72 | } 73 | else 74 | { 75 | static const Vec3 topColor(Vec3(0.5f, 0.7f, 1.0f).sRGBToLinear()); 76 | static const Vec3 bottomColor(Vec3(1.0f, 1.0f, 1.0f).sRGBToLinear()); 77 | 78 | float gradientFactor = (ray.direction().y() + 1.0f) * 0.5f; 79 | radiance = lerp(bottomColor, topColor, gradientFactor); 80 | } 81 | 82 | return radiance; 83 | } 84 | 85 | // Computes the radiance for all the pixels in the image buffer with the specified properties, using 86 | // the specified element (scene) and camera. 87 | void render( 88 | const Element& element, const Camera& camera, uint8_t* pImageData, 89 | uint16_t width, uint16_t height, uint16_t samples) 90 | { 91 | // Report the rendering parameters. 92 | unsigned int threadCount = std::thread::hardware_concurrency(); 93 | std::cout 94 | << "Rendering " << width << "x" << height 95 | << " at " << samples << " samples per pixel on " 96 | << threadCount << " threads..." << std::endl; 97 | 98 | // Record the start time. 99 | auto startTime = std::chrono::high_resolution_clock::now(); 100 | auto prevTime = startTime; 101 | 102 | // Iterate the image pixels, starting from the top left (U = 0.0, Y = 1.0) corner, and computing 103 | // the incident radiance for each one. A parallel for loop is used here to support thread 104 | // concurrency. 105 | // 106 | // NOTE: Ray tracing is a naturally parallel algorithm: there is no read / write contention for 107 | // memory, with the exception of progress reporting. 108 | static const uint8_t NUM_COMPONENTS = 3; 109 | const size_t stride = width * NUM_COMPONENTS; 110 | std::mutex progressMutex; 111 | std::atomic completedLines(0); 112 | Concurrency::parallel_for(uint16_t(0), height, [&](uint16_t line) 113 | { 114 | // Get a pointer to the start of the current line. 115 | uint16_t y = height - line - 1; 116 | uint8_t* pPixel = pImageData + stride * line; 117 | 118 | // Iterate the pixels of line, computing radiance for each one. 119 | for (uint16_t x = 0; x < width; x++) 120 | { 121 | // Create an index for a sequence of *quasirandom* numbers. Such numbers are used for 122 | // "random" sampling while path tracing, e.g. selecting a random direction in a 123 | // hemisphere. The sequence index starts with a unique value for each pixel in the image 124 | // which is then randomized with a hash. 125 | // 126 | // NOTE: Using a constant sequence index leads to total aliasing, but will still 127 | // converge to the correct result with enough samples. Using only the unique per-pixel 128 | // starting index will reduce aliasing, but still yields substantial correlation 129 | // artifacts. Finally, hashing that index yields less objectionable noise, but still 130 | // with better convergence than using *pseudorandom* numbers. 131 | // 132 | // IMPORTANT: For now the same index is used for all random numbers in this pixel 133 | // sample. This strangely works quite well, but will likely need to be revisited. The 134 | // index is only incremented once, when the sample is complete. 135 | uint32_t sequenceIndex = 0; 136 | sequenceIndex = samples * (line * height + x); 137 | sequenceIndex = lowBias32Hash(sequenceIndex); 138 | 139 | // Accumulate radiance samples for each pixel. 140 | Vec3 radiance; 141 | for (uint16_t sample = 0; sample < samples; sample++) 142 | { 143 | // Compute the sample position, using a random offset for each sample. If only one 144 | // sample is being taken, use the pixel center. 145 | // 146 | // NOTE: This uses pseudorandom (MT) numbers because using the quasirandom sequence 147 | // with the same index as the radiance sampling yields minor edge artifacts. 148 | float rand_x = samples == 1 ? 0.5f : randomMT(); 149 | float rand_y = samples == 1 ? 0.5f : randomMT(); 150 | float u = (x + rand_x) / width; 151 | float v = (y - rand_y) / height; 152 | 153 | // Compute a camera ray direction based on the current pixel's UV coordinates. 154 | Ray ray = camera.getRay(u, v); 155 | 156 | // Compute a color for the ray, i.e. the scene radiance from that direction and add 157 | // it to the accumulated radiance. 158 | static const int MAX_DEPTH = 10; 159 | radiance += ::radiance(ray, element, MAX_DEPTH, sequenceIndex); 160 | 161 | // Increment the sequence index, for the next sample. 162 | // 163 | // NOTE: See the "IMPORTANT" note above. 164 | sequenceIndex++; 165 | } 166 | 167 | // Compute the average of the radiance samples to yield the pixel radiance. 168 | radiance /= samples; 169 | 170 | // Gamma correct the radiance and store it in the image buffer. 171 | radiance.linearTosRGB(); 172 | const float COMPONENT_SCALE = 255.99f; 173 | uint8_t color[] = 174 | { 175 | static_cast(clamp(radiance.r(), 0.0f, 1.0f) * COMPONENT_SCALE), 176 | static_cast(clamp(radiance.g(), 0.0f, 1.0f) * COMPONENT_SCALE), 177 | static_cast(clamp(radiance.b(), 0.0f, 1.0f) * COMPONENT_SCALE), 178 | }; 179 | ::memcpy(pPixel, color, 3); 180 | pPixel += 3; 181 | } 182 | 183 | // Increment the (atomic) number of completed lines. 184 | completedLines++; 185 | 186 | // Update the progress if more than one second has elapsed since the last update. 187 | // 188 | // NOTE: A mutex is used to avoid a race condition with multiple threads. 189 | auto nextTime = std::chrono::high_resolution_clock::now(); 190 | auto elapsed = std::chrono::duration_cast(nextTime - prevTime).count(); 191 | if (elapsed >= 1) 192 | { 193 | progressMutex.lock(); 194 | float progress = static_cast(completedLines) / height; 195 | updateProgress(progress); 196 | prevTime = nextTime; 197 | progressMutex.unlock(); 198 | } 199 | }); 200 | 201 | // Finish progress updates. 202 | ::updateProgress(1.0f); 203 | std::cout << std::endl; 204 | 205 | // Report the image dimensions and time spent rendering. 206 | auto endTime = std::chrono::high_resolution_clock::now(); 207 | auto elapsedTime = std::chrono::duration_cast(endTime - startTime).count(); 208 | std::cout 209 | << std::setprecision(3) 210 | << "Completed in " << elapsedTime / 1000.0f << " seconds." << std::endl; 211 | } 212 | 213 | // Main entry point. 214 | int main() 215 | { 216 | // Enable memory leak detection. This will output a memory leak report when the process exits, 217 | // if there are any detected memory leaks. The report starts with "Detected memory leaks!" 218 | _CrtSetDbgFlag(_CRTDBG_ALLOC_MEM_DF | _CRTDBG_LEAK_CHECK_DF); 219 | 220 | // Create scene geometry. 221 | auto pCenter = make_shared(Vec3(0.0f, 0.0f, -1.0f), 0.5f); 222 | auto pGround = make_shared(Vec3(0.0f, -100.5f, -1.0f), 100.0f); 223 | Scene scene; 224 | scene.add(pCenter); 225 | scene.add(pGround); 226 | 227 | // Create the output image. 228 | // 229 | // NOTE: The image can be rendered at a lower resolution and scaled up to the desired image 230 | // size to make it easier to see the individual pixels and for faster rendering. The settings 231 | // here take about 2.1 seconds to render with a Debug build, but only about 300 ms with a 232 | // Release build on a Core i7-8700 CPU with 12 threads. 233 | static const uint8_t SCALE = 8; 234 | static const uint16_t OUTPUT_WIDTH = 3840; 235 | static const uint16_t OUTPUT_HEIGHT = 2160; 236 | static const uint16_t WIDTH = OUTPUT_WIDTH / SCALE; 237 | static const uint16_t HEIGHT = OUTPUT_HEIGHT / SCALE; 238 | static const uint16_t SPP = 16; 239 | Image image(WIDTH, HEIGHT); 240 | 241 | // Create a camera. 242 | // TODO: This will eventually accept typical camera properties: position, direction, FOV, etc. 243 | Camera camera(static_cast(WIDTH) / HEIGHT); 244 | 245 | // Render the scene with the camera, to the image buffer with the specified properties. 246 | ::render(scene, camera, image.getImageData(), WIDTH, HEIGHT, SPP); 247 | 248 | // Save the image. 249 | image.savePNG("output.png", SCALE); 250 | } 251 | -------------------------------------------------------------------------------- /Source/pch.cpp: -------------------------------------------------------------------------------- 1 | #include "pch.h" 2 | -------------------------------------------------------------------------------- /Source/pch.h: -------------------------------------------------------------------------------- 1 | #pragma once 2 | 3 | // STL headers. 4 | #define _USE_MATH_DEFINES 5 | #include 6 | #include 7 | #include 8 | #include 9 | #include 10 | #include 11 | #include 12 | #include 13 | #include 14 | #include 15 | #include 16 | #include 17 | #include 18 | #include 19 | #include 20 | #include 21 | 22 | // Make certain names from the std namespace accessible. 23 | using std::make_shared; 24 | using std::shared_ptr; 25 | using std::string; 26 | using std::vector; 27 | 28 | // Math constants. 29 | const float PI = static_cast(M_PI); 30 | const float INF = std::numeric_limits::infinity(); 31 | --------------------------------------------------------------------------------