diff --git a/CMakeLists.txt b/CMakeLists.txt index d3d976c..41dc21e 100644 --- a/CMakeLists.txt +++ b/CMakeLists.txt @@ -72,7 +72,7 @@ if(${CMAKE_SYSTEM_NAME} MATCHES "Darwin") endif() include_directories(.) -#add_subdirectory(stream_compaction) # TODO: uncomment if using your stream compaction +add_subdirectory(stream_compaction) # TODO: uncomment if using your stream compaction add_subdirectory(src) cuda_add_executable(${CMAKE_PROJECT_NAME} @@ -82,7 +82,7 @@ cuda_add_executable(${CMAKE_PROJECT_NAME} target_link_libraries(${CMAKE_PROJECT_NAME} src - #stream_compaction # TODO: uncomment if using your stream compaction + stream_compaction # TODO: uncomment if using your stream compaction ${CORELIBS} ) diff --git a/README.md b/README.md index 110697c..2701fe6 100644 --- a/README.md +++ b/README.md @@ -3,11 +3,95 @@ CUDA Path Tracer **University of Pennsylvania, CIS 565: GPU Programming and Architecture, Project 3** -* (TODO) YOUR NAME HERE -* Tested on: (TODO) Windows 22, i7-2222 @ 2.22GHz 22GB, GTX 222 222MB (Moore 2222 Lab) +* Siyu Zheng +* Tested on: Windows 10, i7-8750 @ 2.20GHz 16GB, GTX 1060 6GB, Visual Studio 2015, CUDA 8.0(Personal Laptop) -### (TODO: Your README) +## CUDA Path Tracer -*DO NOT* leave the README to the last minute! It is a crucial part of the -project, and we will not be able to grade you without a good README. +![](img/regular5000.png) +### Anti-aliasing + +Used Stochastic Sampled Antialiasing. Stochastic sampling is a discrete antialiasing method which antialiases scenes through the use of a special sample distribution which replaces most aliased image artefacts with noise of the correct average intensity. + +![](img/stochastic.png) + + +![](img/antialiasing_5000.png) + +| No Anti-aliasing | Anti-aliasing | +| ------------- |:-------------:| +| ![](img/no_antialiasing.png) | ![](img/antialiasing.png) | + +### Refraction + +Refraction [PBRT 8.2] with Frensel effects using Schlick's approximation, which is a formula for approximating the contribution of the Fresnel factor in the specular reflection of light from a non-conducting interface between two media. + +| Reflection | Refraction | +| ------------- |:-------------:| +| ![](img/regular5000.png) | ![](img/refraction.png) | + +### Depth of field + +Physically-based depth-of-field. Offset ray origin on a lens. Under the thin lens approximation, parallel incident rays passing through the lens focus at the focal point. If the film plane is placed at a distance equal to the focal length behind the lens, then objects infinitely far away will be in focus, as they image to a single point on the film. + +![](img/thinlens.png) + +| Focal length 7 | Focal length 12 | +| ------------- |:-------------:| +| ![](img/dof_7_0.4.png) | ![](img/dof_12_1.png) | + +### Motion Blur + +Add a motion vector to the translation vector on the model. So model may have a slightly different transformation for each ray. + +| No Motion Blur | Motion Blur | +| ------------- |:-------------:| +| ![](img/regular5000.png) | ![](img/motion_blur.png) | + + + + +### Arbitrary mesh loading and rendering +A mesh loading of wahoo.obj. +Used tinyObj to read data. Used triangle intersection function glm::intersectRayTriangle to calculate intersections of the mesh object. (Need improvement with better data structure or aabb) + +![](img/mesh_object.png) + + + +## Performance Analysis + +Sorted the rays/path segments so that rays/paths interacting with the same material are contiguous in memory before shading. + +![](img/performance_sort.png) + +Compared the performance w/o sorting in defferent trace depth. The non-sorting version is much faster than sorting version. This might because thrust::sort_by_key took longer time than we saved using contiguous memory accessing. Another reason might be the scene is originally composed of same material, so the sorting did not improve a lot. If the scene become more complicated, we might take advantage of material sorting. + +Next, cached the first bounces data , so that we don't need to recalculate first bounce path segment after first iteration. + +![](img/performance_cach.png) + +According to the graph, cach first version has better performance since it save the time for recalculation for first bounce path segment after first iteration. + +In each iteration, I used stream compaction to terminate path segment which has no more remain bounces. Below is a graph shows how many rays left after each depth. + +![](img/open_closed.png) + +As we see from the graph, stream compaction terminate a lot of paths. Compare scenes which are open (like the given cornell box) and closed (i.e. no light can escape the scene), we can see that since rays in closed scene can not escape, there were more rays remain in closed scene. Also, at first few depth open scenes has more rays terminated, later since many rays have already escape from the scene, the curve has a slower downward trend. + +## Blooper + +thrust::remove_if problem in stream compaction. +![](img/compaction_blooper.png) + +Miscalculate refraction ray costheta. +![](img/refraction_blooper.png) + +## Reference + +[tinyObj](http://syoyo.github.io/tinyobjloader/) to bring the data from object file into C++. + +Physically-based depth-of-field (by jittering rays within an aperture) [PBRT 6.2.3]. + +Stochastic Sampled Antialiasing from [Paul Bourke's note](http://paulbourke.net/miscellaneous/raytracing/) \ No newline at end of file diff --git a/external/include/tiny_obj_loader.h b/external/include/tiny_obj_loader.h new file mode 100644 index 0000000..f627e94 --- /dev/null +++ b/external/include/tiny_obj_loader.h @@ -0,0 +1,2518 @@ +/* +The MIT License (MIT) + +Copyright (c) 2012-2018 Syoyo Fujita and many contributors. + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in +all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +THE SOFTWARE. +*/ + +// +// version 1.2.3 : Added color space extension('-colorspace') to tex opts. +// version 1.2.2 : Parse multiple group names. +// version 1.2.1 : Added initial support for line('l') primitive(PR #178) +// version 1.2.0 : Hardened implementation(#175) +// version 1.1.1 : Support smoothing groups(#162) +// version 1.1.0 : Support parsing vertex color(#144) +// version 1.0.8 : Fix parsing `g` tag just after `usemtl`(#138) +// version 1.0.7 : Support multiple tex options(#126) +// version 1.0.6 : Add TINYOBJLOADER_USE_DOUBLE option(#124) +// version 1.0.5 : Ignore `Tr` when `d` exists in MTL(#43) +// version 1.0.4 : Support multiple filenames for 'mtllib'(#112) +// version 1.0.3 : Support parsing texture options(#85) +// version 1.0.2 : Improve parsing speed by about a factor of 2 for large +// files(#105) +// version 1.0.1 : Fixes a shape is lost if obj ends with a 'usemtl'(#104) +// version 1.0.0 : Change data structure. Change license from BSD to MIT. +// + +// +// Use this in *one* .cc +// #define TINYOBJLOADER_IMPLEMENTATION +// #include "tiny_obj_loader.h" +// + +#ifndef TINY_OBJ_LOADER_H_ +#define TINY_OBJ_LOADER_H_ + +#include +#include +#include + +namespace tinyobj { + +#ifdef __clang__ +#pragma clang diagnostic push +#if __has_warning("-Wzero-as-null-pointer-constant") +#pragma clang diagnostic ignored "-Wzero-as-null-pointer-constant" +#endif + +#pragma clang diagnostic ignored "-Wpadded" + +#endif + +// https://en.wikipedia.org/wiki/Wavefront_.obj_file says ... +// +// -blendu on | off # set horizontal texture blending +// (default on) +// -blendv on | off # set vertical texture blending +// (default on) +// -boost real_value # boost mip-map sharpness +// -mm base_value gain_value # modify texture map values (default +// 0 1) +// # base_value = brightness, +// gain_value = contrast +// -o u [v [w]] # Origin offset (default +// 0 0 0) +// -s u [v [w]] # Scale (default +// 1 1 1) +// -t u [v [w]] # Turbulence (default +// 0 0 0) +// -texres resolution # texture resolution to create +// -clamp on | off # only render texels in the clamped +// 0-1 range (default off) +// # When unclamped, textures are +// repeated across a surface, +// # when clamped, only texels which +// fall within the 0-1 +// # range are rendered. +// -bm mult_value # bump multiplier (for bump maps +// only) +// +// -imfchan r | g | b | m | l | z # specifies which channel of the file +// is used to +// # create a scalar or bump texture. +// r:red, g:green, +// # b:blue, m:matte, l:luminance, +// z:z-depth.. +// # (the default for bump is 'l' and +// for decal is 'm') +// bump -imfchan r bumpmap.tga # says to use the red channel of +// bumpmap.tga as the bumpmap +// +// For reflection maps... +// +// -type sphere # specifies a sphere for a "refl" +// reflection map +// -type cube_top | cube_bottom | # when using a cube map, the texture +// file for each +// cube_front | cube_back | # side of the cube is specified +// separately +// cube_left | cube_right +// +// TinyObjLoader extension. +// +// -colorspace SPACE # Color space of the texture. e.g. 'sRGB` or 'linear' +// + +#ifdef TINYOBJLOADER_USE_DOUBLE +//#pragma message "using double" +typedef double real_t; +#else +//#pragma message "using float" +typedef float real_t; +#endif + +typedef enum { + TEXTURE_TYPE_NONE, // default + TEXTURE_TYPE_SPHERE, + TEXTURE_TYPE_CUBE_TOP, + TEXTURE_TYPE_CUBE_BOTTOM, + TEXTURE_TYPE_CUBE_FRONT, + TEXTURE_TYPE_CUBE_BACK, + TEXTURE_TYPE_CUBE_LEFT, + TEXTURE_TYPE_CUBE_RIGHT +} texture_type_t; + +typedef struct { + texture_type_t type; // -type (default TEXTURE_TYPE_NONE) + real_t sharpness; // -boost (default 1.0?) + real_t brightness; // base_value in -mm option (default 0) + real_t contrast; // gain_value in -mm option (default 1) + real_t origin_offset[3]; // -o u [v [w]] (default 0 0 0) + real_t scale[3]; // -s u [v [w]] (default 1 1 1) + real_t turbulence[3]; // -t u [v [w]] (default 0 0 0) + // int texture_resolution; // -texres resolution (default = ?) TODO + bool clamp; // -clamp (default false) + char imfchan; // -imfchan (the default for bump is 'l' and for decal is 'm') + bool blendu; // -blendu (default on) + bool blendv; // -blendv (default on) + real_t bump_multiplier; // -bm (for bump maps only, default 1.0) + + // extension + std::string colorspace; // Explicitly specify color space of stored value. Usually `sRGB` or `linear` (default empty). +} texture_option_t; + +typedef struct { + std::string name; + + real_t ambient[3]; + real_t diffuse[3]; + real_t specular[3]; + real_t transmittance[3]; + real_t emission[3]; + real_t shininess; + real_t ior; // index of refraction + real_t dissolve; // 1 == opaque; 0 == fully transparent + // illumination model (see http://www.fileformat.info/format/material/) + int illum; + + int dummy; // Suppress padding warning. + + std::string ambient_texname; // map_Ka + std::string diffuse_texname; // map_Kd + std::string specular_texname; // map_Ks + std::string specular_highlight_texname; // map_Ns + std::string bump_texname; // map_bump, map_Bump, bump + std::string displacement_texname; // disp + std::string alpha_texname; // map_d + std::string reflection_texname; // refl + + texture_option_t ambient_texopt; + texture_option_t diffuse_texopt; + texture_option_t specular_texopt; + texture_option_t specular_highlight_texopt; + texture_option_t bump_texopt; + texture_option_t displacement_texopt; + texture_option_t alpha_texopt; + texture_option_t reflection_texopt; + + // PBR extension + // http://exocortex.com/blog/extending_wavefront_mtl_to_support_pbr + real_t roughness; // [0, 1] default 0 + real_t metallic; // [0, 1] default 0 + real_t sheen; // [0, 1] default 0 + real_t clearcoat_thickness; // [0, 1] default 0 + real_t clearcoat_roughness; // [0, 1] default 0 + real_t anisotropy; // aniso. [0, 1] default 0 + real_t anisotropy_rotation; // anisor. [0, 1] default 0 + real_t pad0; + std::string roughness_texname; // map_Pr + std::string metallic_texname; // map_Pm + std::string sheen_texname; // map_Ps + std::string emissive_texname; // map_Ke + std::string normal_texname; // norm. For normal mapping. + + texture_option_t roughness_texopt; + texture_option_t metallic_texopt; + texture_option_t sheen_texopt; + texture_option_t emissive_texopt; + texture_option_t normal_texopt; + + int pad2; + + std::map unknown_parameter; +} material_t; + +typedef struct { + std::string name; + + std::vector intValues; + std::vector floatValues; + std::vector stringValues; +} tag_t; + +// Index struct to support different indices for vtx/normal/texcoord. +// -1 means not used. +typedef struct { + int vertex_index; + int normal_index; + int texcoord_index; +} index_t; + +typedef struct { + std::vector indices; + std::vector num_face_vertices; // The number of vertices per + // face. 3 = polygon, 4 = quad, + // ... Up to 255. + std::vector material_ids; // per-face material ID + std::vector smoothing_group_ids; // per-face smoothing group + // ID(0 = off. positive value + // = group id) + std::vector tags; // SubD tag +} mesh_t; + +typedef struct { + std::vector indices; // pairs of indices for lines +} path_t; + +typedef struct { + std::string name; + mesh_t mesh; + path_t path; +} shape_t; + +// Vertex attributes +typedef struct { + std::vector vertices; // 'v' + std::vector normals; // 'vn' + std::vector texcoords; // 'vt' + std::vector colors; // extension: vertex colors +} attrib_t; + +typedef struct callback_t_ { + // W is optional and set to 1 if there is no `w` item in `v` line + void (*vertex_cb)(void *user_data, real_t x, real_t y, real_t z, real_t w); + void (*normal_cb)(void *user_data, real_t x, real_t y, real_t z); + + // y and z are optional and set to 0 if there is no `y` and/or `z` item(s) in + // `vt` line. + void (*texcoord_cb)(void *user_data, real_t x, real_t y, real_t z); + + // called per 'f' line. num_indices is the number of face indices(e.g. 3 for + // triangle, 4 for quad) + // 0 will be passed for undefined index in index_t members. + void (*index_cb)(void *user_data, index_t *indices, int num_indices); + // `name` material name, `material_id` = the array index of material_t[]. -1 + // if + // a material not found in .mtl + void (*usemtl_cb)(void *user_data, const char *name, int material_id); + // `materials` = parsed material data. + void (*mtllib_cb)(void *user_data, const material_t *materials, + int num_materials); + // There may be multiple group names + void (*group_cb)(void *user_data, const char **names, int num_names); + void (*object_cb)(void *user_data, const char *name); + + callback_t_() + : vertex_cb(NULL), + normal_cb(NULL), + texcoord_cb(NULL), + index_cb(NULL), + usemtl_cb(NULL), + mtllib_cb(NULL), + group_cb(NULL), + object_cb(NULL) {} +} callback_t; + +class MaterialReader { + public: + MaterialReader() {} + virtual ~MaterialReader(); + + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) = 0; +}; + +class MaterialFileReader : public MaterialReader { + public: + explicit MaterialFileReader(const std::string &mtl_basedir) + : m_mtlBaseDir(mtl_basedir) {} + virtual ~MaterialFileReader() {} + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, std::string *err); + + private: + std::string m_mtlBaseDir; +}; + +class MaterialStreamReader : public MaterialReader { + public: + explicit MaterialStreamReader(std::istream &inStream) + : m_inStream(inStream) {} + virtual ~MaterialStreamReader() {} + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, std::string *err); + + private: + std::istream &m_inStream; +}; + +/// Loads .obj from a file. +/// 'attrib', 'shapes' and 'materials' will be filled with parsed shape data +/// 'shapes' will be filled with parsed shape data +/// Returns true when loading .obj become success. +/// Returns warning and error message into `err` +/// 'mtl_basedir' is optional, and used for base directory for .mtl file. +/// In default(`NULL'), .mtl file is searched from an application's working +/// directory. +/// 'triangulate' is optional, and used whether triangulate polygon face in .obj +/// or not. +/// Option 'default_vcols_fallback' specifies whether vertex colors should +/// always be defined, even if no colors are given (fallback to white). +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + const char *filename, const char *mtl_basedir = NULL, + bool triangulate = true, bool default_vcols_fallback = true); + +/// Loads .obj from a file with custom user callback. +/// .mtl is loaded as usual and parsed material_t data will be passed to +/// `callback.mtllib_cb`. +/// Returns true when loading .obj/.mtl become success. +/// Returns warning and error message into `err` +/// See `examples/callback_api/` for how to use this function. +bool LoadObjWithCallback(std::istream &inStream, const callback_t &callback, + void *user_data = NULL, + MaterialReader *readMatFn = NULL, + std::string *err = NULL); + +/// Loads object from a std::istream, uses GetMtlIStreamFn to retrieve +/// std::istream for materials. +/// Returns true when loading .obj become success. +/// Returns warning and error message into `err` +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + std::istream *inStream, MaterialReader *readMatFn = NULL, + bool triangulate = true, bool default_vcols_fallback = true); + +/// Loads materials into std::map +void LoadMtl(std::map *material_map, + std::vector *materials, std::istream *inStream, + std::string *warning); + +} // namespace tinyobj + +#endif // TINY_OBJ_LOADER_H_ + +#ifdef TINYOBJLOADER_IMPLEMENTATION +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +namespace tinyobj { + +MaterialReader::~MaterialReader() {} + +struct vertex_index_t { + int v_idx, vt_idx, vn_idx; + vertex_index_t() : v_idx(-1), vt_idx(-1), vn_idx(-1) {} + explicit vertex_index_t(int idx) : v_idx(idx), vt_idx(idx), vn_idx(idx) {} + vertex_index_t(int vidx, int vtidx, int vnidx) + : v_idx(vidx), vt_idx(vtidx), vn_idx(vnidx) {} +}; + +// Internal data structure for face representation +// index + smoothing group. +struct face_t { + unsigned int + smoothing_group_id; // smoothing group id. 0 = smoothing groupd is off. + int pad_; + std::vector vertex_indices; // face vertex indices. + + face_t() : smoothing_group_id(0) {} +}; + +struct line_t { + int idx0; + int idx1; +}; + +struct tag_sizes { + tag_sizes() : num_ints(0), num_reals(0), num_strings(0) {} + int num_ints; + int num_reals; + int num_strings; +}; + +struct obj_shape { + std::vector v; + std::vector vn; + std::vector vt; +}; + +// See +// http://stackoverflow.com/questions/6089231/getting-std-ifstream-to-handle-lf-cr-and-crlf +static std::istream &safeGetline(std::istream &is, std::string &t) { + t.clear(); + + // The characters in the stream are read one-by-one using a std::streambuf. + // That is faster than reading them one-by-one using the std::istream. + // Code that uses streambuf this way must be guarded by a sentry object. + // The sentry object performs various tasks, + // such as thread synchronization and updating the stream state. + + std::istream::sentry se(is, true); + std::streambuf *sb = is.rdbuf(); + + if (se) { + for (;;) { + int c = sb->sbumpc(); + switch (c) { + case '\n': + return is; + case '\r': + if (sb->sgetc() == '\n') sb->sbumpc(); + return is; + case EOF: + // Also handle the case when the last line has no line ending + if (t.empty()) is.setstate(std::ios::eofbit); + return is; + default: + t += static_cast(c); + } + } + } + + return is; +} + +#define IS_SPACE(x) (((x) == ' ') || ((x) == '\t')) +#define IS_DIGIT(x) \ + (static_cast((x) - '0') < static_cast(10)) +#define IS_NEW_LINE(x) (((x) == '\r') || ((x) == '\n') || ((x) == '\0')) + +// Make index zero-base, and also support relative index. +static inline bool fixIndex(int idx, int n, int *ret) { + if (!ret) { + return false; + } + + if (idx > 0) { + (*ret) = idx - 1; + return true; + } + + if (idx == 0) { + // zero is not allowed according to the spec. + return false; + } + + if (idx < 0) { + (*ret) = n + idx; // negative value = relative + return true; + } + + return false; // never reach here. +} + +static inline std::string parseString(const char **token) { + std::string s; + (*token) += strspn((*token), " \t"); + size_t e = strcspn((*token), " \t\r"); + s = std::string((*token), &(*token)[e]); + (*token) += e; + return s; +} + +static inline int parseInt(const char **token) { + (*token) += strspn((*token), " \t"); + int i = atoi((*token)); + (*token) += strcspn((*token), " \t\r"); + return i; +} + +// Tries to parse a floating point number located at s. +// +// s_end should be a location in the string where reading should absolutely +// stop. For example at the end of the string, to prevent buffer overflows. +// +// Parses the following EBNF grammar: +// sign = "+" | "-" ; +// END = ? anything not in digit ? +// digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ; +// integer = [sign] , digit , {digit} ; +// decimal = integer , ["." , integer] ; +// float = ( decimal , END ) | ( decimal , ("E" | "e") , integer , END ) ; +// +// Valid strings are for example: +// -0 +3.1417e+2 -0.0E-3 1.0324 -1.41 11e2 +// +// If the parsing is a success, result is set to the parsed value and true +// is returned. +// +// The function is greedy and will parse until any of the following happens: +// - a non-conforming character is encountered. +// - s_end is reached. +// +// The following situations triggers a failure: +// - s >= s_end. +// - parse failure. +// +static bool tryParseDouble(const char *s, const char *s_end, double *result) { + if (s >= s_end) { + return false; + } + + double mantissa = 0.0; + // This exponent is base 2 rather than 10. + // However the exponent we parse is supposed to be one of ten, + // thus we must take care to convert the exponent/and or the + // mantissa to a * 2^E, where a is the mantissa and E is the + // exponent. + // To get the final double we will use ldexp, it requires the + // exponent to be in base 2. + int exponent = 0; + + // NOTE: THESE MUST BE DECLARED HERE SINCE WE ARE NOT ALLOWED + // TO JUMP OVER DEFINITIONS. + char sign = '+'; + char exp_sign = '+'; + char const *curr = s; + + // How many characters were read in a loop. + int read = 0; + // Tells whether a loop terminated due to reaching s_end. + bool end_not_reached = false; + + /* + BEGIN PARSING. + */ + + // Find out what sign we've got. + if (*curr == '+' || *curr == '-') { + sign = *curr; + curr++; + } else if (IS_DIGIT(*curr)) { /* Pass through. */ + } else { + goto fail; + } + + // Read the integer part. + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + mantissa *= 10; + mantissa += static_cast(*curr - 0x30); + curr++; + read++; + end_not_reached = (curr != s_end); + } + + // We must make sure we actually got something. + if (read == 0) goto fail; + // We allow numbers of form "#", "###" etc. + if (!end_not_reached) goto assemble; + + // Read the decimal part. + if (*curr == '.') { + curr++; + read = 1; + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + static const double pow_lut[] = { + 1.0, 0.1, 0.01, 0.001, 0.0001, 0.00001, 0.000001, 0.0000001, + }; + const int lut_entries = sizeof pow_lut / sizeof pow_lut[0]; + + // NOTE: Don't use powf here, it will absolutely murder precision. + mantissa += static_cast(*curr - 0x30) * + (read < lut_entries ? pow_lut[read] : std::pow(10.0, -read)); + read++; + curr++; + end_not_reached = (curr != s_end); + } + } else if (*curr == 'e' || *curr == 'E') { + } else { + goto assemble; + } + + if (!end_not_reached) goto assemble; + + // Read the exponent part. + if (*curr == 'e' || *curr == 'E') { + curr++; + // Figure out if a sign is present and if it is. + end_not_reached = (curr != s_end); + if (end_not_reached && (*curr == '+' || *curr == '-')) { + exp_sign = *curr; + curr++; + } else if (IS_DIGIT(*curr)) { /* Pass through. */ + } else { + // Empty E is not allowed. + goto fail; + } + + read = 0; + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + exponent *= 10; + exponent += static_cast(*curr - 0x30); + curr++; + read++; + end_not_reached = (curr != s_end); + } + exponent *= (exp_sign == '+' ? 1 : -1); + if (read == 0) goto fail; + } + +assemble: + *result = (sign == '+' ? 1 : -1) * + (exponent ? std::ldexp(mantissa * std::pow(5.0, exponent), exponent) + : mantissa); + return true; +fail: + return false; +} + +static inline real_t parseReal(const char **token, double default_value = 0.0) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + double val = default_value; + tryParseDouble((*token), end, &val); + real_t f = static_cast(val); + (*token) = end; + return f; +} + +static inline bool parseReal(const char **token, real_t *out) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + double val; + bool ret = tryParseDouble((*token), end, &val); + if (ret) { + real_t f = static_cast(val); + (*out) = f; + } + (*token) = end; + return ret; +} + +static inline void parseReal2(real_t *x, real_t *y, const char **token, + const double default_x = 0.0, + const double default_y = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); +} + +static inline void parseReal3(real_t *x, real_t *y, real_t *z, + const char **token, const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); +} + +static inline void parseV(real_t *x, real_t *y, real_t *z, real_t *w, + const char **token, const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0, + const double default_w = 1.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); + (*w) = parseReal(token, default_w); +} + +// Extension: parse vertex with colors(6 items) +static inline bool parseVertexWithColor(real_t *x, real_t *y, real_t *z, + real_t *r, real_t *g, real_t *b, + const char **token, + const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); + + const bool found_color = parseReal(token, r) && parseReal(token, g) && parseReal(token, b); + + if (!found_color) { + (*r) = (*g) = (*b) = 1.0; + } + + return found_color; +} + +static inline bool parseOnOff(const char **token, bool default_value = true) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + + bool ret = default_value; + if ((0 == strncmp((*token), "on", 2))) { + ret = true; + } else if ((0 == strncmp((*token), "off", 3))) { + ret = false; + } + + (*token) = end; + return ret; +} + +static inline texture_type_t parseTextureType( + const char **token, texture_type_t default_value = TEXTURE_TYPE_NONE) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + texture_type_t ty = default_value; + + if ((0 == strncmp((*token), "cube_top", strlen("cube_top")))) { + ty = TEXTURE_TYPE_CUBE_TOP; + } else if ((0 == strncmp((*token), "cube_bottom", strlen("cube_bottom")))) { + ty = TEXTURE_TYPE_CUBE_BOTTOM; + } else if ((0 == strncmp((*token), "cube_left", strlen("cube_left")))) { + ty = TEXTURE_TYPE_CUBE_LEFT; + } else if ((0 == strncmp((*token), "cube_right", strlen("cube_right")))) { + ty = TEXTURE_TYPE_CUBE_RIGHT; + } else if ((0 == strncmp((*token), "cube_front", strlen("cube_front")))) { + ty = TEXTURE_TYPE_CUBE_FRONT; + } else if ((0 == strncmp((*token), "cube_back", strlen("cube_back")))) { + ty = TEXTURE_TYPE_CUBE_BACK; + } else if ((0 == strncmp((*token), "sphere", strlen("sphere")))) { + ty = TEXTURE_TYPE_SPHERE; + } + + (*token) = end; + return ty; +} + +static tag_sizes parseTagTriple(const char **token) { + tag_sizes ts; + + (*token) += strspn((*token), " \t"); + ts.num_ints = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return ts; + } + + (*token)++; // Skip '/' + + (*token) += strspn((*token), " \t"); + ts.num_reals = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return ts; + } + (*token)++; // Skip '/' + + ts.num_strings = parseInt(token); + + return ts; +} + +// Parse triples with index offsets: i, i/j/k, i//k, i/j +static bool parseTriple(const char **token, int vsize, int vnsize, int vtsize, + vertex_index_t *ret) { + if (!ret) { + return false; + } + + vertex_index_t vi(-1); + + if (!fixIndex(atoi((*token)), vsize, &(vi.v_idx))) { + return false; + } + + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + (*ret) = vi; + return true; + } + (*token)++; + + // i//k + if ((*token)[0] == '/') { + (*token)++; + if (!fixIndex(atoi((*token)), vnsize, &(vi.vn_idx))) { + return false; + } + (*token) += strcspn((*token), "/ \t\r"); + (*ret) = vi; + return true; + } + + // i/j/k or i/j + if (!fixIndex(atoi((*token)), vtsize, &(vi.vt_idx))) { + return false; + } + + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + (*ret) = vi; + return true; + } + + // i/j/k + (*token)++; // skip '/' + if (!fixIndex(atoi((*token)), vnsize, &(vi.vn_idx))) { + return false; + } + (*token) += strcspn((*token), "/ \t\r"); + + (*ret) = vi; + + return true; +} + +// Parse raw triples: i, i/j/k, i//k, i/j +static vertex_index_t parseRawTriple(const char **token) { + vertex_index_t vi(static_cast(0)); // 0 is an invalid index in OBJ + + vi.v_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return vi; + } + (*token)++; + + // i//k + if ((*token)[0] == '/') { + (*token)++; + vi.vn_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + return vi; + } + + // i/j/k or i/j + vi.vt_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return vi; + } + + // i/j/k + (*token)++; // skip '/' + vi.vn_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + return vi; +} + +static bool ParseTextureNameAndOption(std::string *texname, + texture_option_t *texopt, + const char *linebuf, const bool is_bump) { + // @todo { write more robust lexer and parser. } + bool found_texname = false; + std::string texture_name; + + // Fill with default value for texopt. + if (is_bump) { + texopt->imfchan = 'l'; + } else { + texopt->imfchan = 'm'; + } + texopt->bump_multiplier = static_cast(1.0); + texopt->clamp = false; + texopt->blendu = true; + texopt->blendv = true; + texopt->sharpness = static_cast(1.0); + texopt->brightness = static_cast(0.0); + texopt->contrast = static_cast(1.0); + texopt->origin_offset[0] = static_cast(0.0); + texopt->origin_offset[1] = static_cast(0.0); + texopt->origin_offset[2] = static_cast(0.0); + texopt->scale[0] = static_cast(1.0); + texopt->scale[1] = static_cast(1.0); + texopt->scale[2] = static_cast(1.0); + texopt->turbulence[0] = static_cast(0.0); + texopt->turbulence[1] = static_cast(0.0); + texopt->turbulence[2] = static_cast(0.0); + texopt->type = TEXTURE_TYPE_NONE; + + const char *token = linebuf; // Assume line ends with NULL + + while (!IS_NEW_LINE((*token))) { + token += strspn(token, " \t"); // skip space + if ((0 == strncmp(token, "-blendu", 7)) && IS_SPACE((token[7]))) { + token += 8; + texopt->blendu = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-blendv", 7)) && IS_SPACE((token[7]))) { + token += 8; + texopt->blendv = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-clamp", 6)) && IS_SPACE((token[6]))) { + token += 7; + texopt->clamp = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-boost", 6)) && IS_SPACE((token[6]))) { + token += 7; + texopt->sharpness = parseReal(&token, 1.0); + } else if ((0 == strncmp(token, "-bm", 3)) && IS_SPACE((token[3]))) { + token += 4; + texopt->bump_multiplier = parseReal(&token, 1.0); + } else if ((0 == strncmp(token, "-o", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->origin_offset[0]), &(texopt->origin_offset[1]), + &(texopt->origin_offset[2]), &token); + } else if ((0 == strncmp(token, "-s", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->scale[0]), &(texopt->scale[1]), &(texopt->scale[2]), + &token, 1.0, 1.0, 1.0); + } else if ((0 == strncmp(token, "-t", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->turbulence[0]), &(texopt->turbulence[1]), + &(texopt->turbulence[2]), &token); + } else if ((0 == strncmp(token, "-type", 5)) && IS_SPACE((token[5]))) { + token += 5; + texopt->type = parseTextureType((&token), TEXTURE_TYPE_NONE); + } else if ((0 == strncmp(token, "-imfchan", 8)) && IS_SPACE((token[8]))) { + token += 9; + token += strspn(token, " \t"); + const char *end = token + strcspn(token, " \t\r"); + if ((end - token) == 1) { // Assume one char for -imfchan + texopt->imfchan = (*token); + } + token = end; + } else if ((0 == strncmp(token, "-mm", 3)) && IS_SPACE((token[3]))) { + token += 4; + parseReal2(&(texopt->brightness), &(texopt->contrast), &token, 0.0, 1.0); + } else if ((0 == strncmp(token, "-colorspace", 11)) && IS_SPACE((token[11]))) { + token += 12; + texopt->colorspace = parseString(&token); + } else { + // Assume texture filename +#if 0 + size_t len = strcspn(token, " \t\r"); // untile next space + texture_name = std::string(token, token + len); + token += len; + + token += strspn(token, " \t"); // skip space +#else + // Read filename until line end to parse filename containing whitespace + // TODO(syoyo): Support parsing texture option flag after the filename. + texture_name = std::string(token); + token += texture_name.length(); +#endif + + found_texname = true; + } + } + + if (found_texname) { + (*texname) = texture_name; + return true; + } else { + return false; + } +} + +static void InitMaterial(material_t *material) { + material->name = ""; + material->ambient_texname = ""; + material->diffuse_texname = ""; + material->specular_texname = ""; + material->specular_highlight_texname = ""; + material->bump_texname = ""; + material->displacement_texname = ""; + material->reflection_texname = ""; + material->alpha_texname = ""; + for (int i = 0; i < 3; i++) { + material->ambient[i] = static_cast(0.0); + material->diffuse[i] = static_cast(0.0); + material->specular[i] = static_cast(0.0); + material->transmittance[i] = static_cast(0.0); + material->emission[i] = static_cast(0.0); + } + material->illum = 0; + material->dissolve = static_cast(1.0); + material->shininess = static_cast(1.0); + material->ior = static_cast(1.0); + + material->roughness = static_cast(0.0); + material->metallic = static_cast(0.0); + material->sheen = static_cast(0.0); + material->clearcoat_thickness = static_cast(0.0); + material->clearcoat_roughness = static_cast(0.0); + material->anisotropy_rotation = static_cast(0.0); + material->anisotropy = static_cast(0.0); + material->roughness_texname = ""; + material->metallic_texname = ""; + material->sheen_texname = ""; + material->emissive_texname = ""; + material->normal_texname = ""; + + material->unknown_parameter.clear(); +} + +// code from https://wrf.ecse.rpi.edu//Research/Short_Notes/pnpoly.html +template +static int pnpoly(int nvert, T *vertx, T *verty, T testx, T testy) { + int i, j, c = 0; + for (i = 0, j = nvert - 1; i < nvert; j = i++) { + if (((verty[i] > testy) != (verty[j] > testy)) && + (testx < + (vertx[j] - vertx[i]) * (testy - verty[i]) / (verty[j] - verty[i]) + + vertx[i])) + c = !c; + } + return c; +} + +// TODO(syoyo): refactor function. +static bool exportGroupsToShape(shape_t *shape, + const std::vector &faceGroup, + std::vector &lineGroup, + const std::vector &tags, + const int material_id, const std::string &name, + bool triangulate, + const std::vector &v) { + if (faceGroup.empty() && lineGroup.empty()) { + return false; + } + + if (!faceGroup.empty()) { + // Flatten vertices and indices + for (size_t i = 0; i < faceGroup.size(); i++) { + const face_t &face = faceGroup[i]; + + size_t npolys = face.vertex_indices.size(); + + if (npolys < 3) { + // Face must have 3+ vertices. + continue; + } + + vertex_index_t i0 = face.vertex_indices[0]; + vertex_index_t i1(-1); + vertex_index_t i2 = face.vertex_indices[1]; + + if (triangulate) { + // find the two axes to work in + size_t axes[2] = {1, 2}; + for (size_t k = 0; k < npolys; ++k) { + i0 = face.vertex_indices[(k + 0) % npolys]; + i1 = face.vertex_indices[(k + 1) % npolys]; + i2 = face.vertex_indices[(k + 2) % npolys]; + size_t vi0 = size_t(i0.v_idx); + size_t vi1 = size_t(i1.v_idx); + size_t vi2 = size_t(i2.v_idx); + + if (((3 * vi0 + 2) >= v.size()) || ((3 * vi1 + 2) >= v.size()) || + ((3 * vi2 + 2) >= v.size())) { + // Invalid triangle. + // FIXME(syoyo): Is it ok to simply skip this invalid triangle? + continue; + } + real_t v0x = v[vi0 * 3 + 0]; + real_t v0y = v[vi0 * 3 + 1]; + real_t v0z = v[vi0 * 3 + 2]; + real_t v1x = v[vi1 * 3 + 0]; + real_t v1y = v[vi1 * 3 + 1]; + real_t v1z = v[vi1 * 3 + 2]; + real_t v2x = v[vi2 * 3 + 0]; + real_t v2y = v[vi2 * 3 + 1]; + real_t v2z = v[vi2 * 3 + 2]; + real_t e0x = v1x - v0x; + real_t e0y = v1y - v0y; + real_t e0z = v1z - v0z; + real_t e1x = v2x - v1x; + real_t e1y = v2y - v1y; + real_t e1z = v2z - v1z; + real_t cx = std::fabs(e0y * e1z - e0z * e1y); + real_t cy = std::fabs(e0z * e1x - e0x * e1z); + real_t cz = std::fabs(e0x * e1y - e0y * e1x); + const real_t epsilon = std::numeric_limits::epsilon(); + if (cx > epsilon || cy > epsilon || cz > epsilon) { + // found a corner + if (cx > cy && cx > cz) { + } else { + axes[0] = 0; + if (cz > cx && cz > cy) axes[1] = 1; + } + break; + } + } + + real_t area = 0; + for (size_t k = 0; k < npolys; ++k) { + i0 = face.vertex_indices[(k + 0) % npolys]; + i1 = face.vertex_indices[(k + 1) % npolys]; + size_t vi0 = size_t(i0.v_idx); + size_t vi1 = size_t(i1.v_idx); + if (((vi0 * 3 + axes[0]) >= v.size()) || + ((vi0 * 3 + axes[1]) >= v.size()) || + ((vi1 * 3 + axes[0]) >= v.size()) || + ((vi1 * 3 + axes[1]) >= v.size())) { + // Invalid index. + continue; + } + real_t v0x = v[vi0 * 3 + axes[0]]; + real_t v0y = v[vi0 * 3 + axes[1]]; + real_t v1x = v[vi1 * 3 + axes[0]]; + real_t v1y = v[vi1 * 3 + axes[1]]; + area += (v0x * v1y - v0y * v1x) * static_cast(0.5); + } + + int maxRounds = 10; // arbitrary max loop count to protect against + // unexpected errors + + face_t remainingFace = face; // copy + size_t guess_vert = 0; + vertex_index_t ind[3]; + real_t vx[3]; + real_t vy[3]; + while (remainingFace.vertex_indices.size() > 3 && maxRounds > 0) { + npolys = remainingFace.vertex_indices.size(); + if (guess_vert >= npolys) { + maxRounds -= 1; + guess_vert -= npolys; + } + for (size_t k = 0; k < 3; k++) { + ind[k] = remainingFace.vertex_indices[(guess_vert + k) % npolys]; + size_t vi = size_t(ind[k].v_idx); + if (((vi * 3 + axes[0]) >= v.size()) || + ((vi * 3 + axes[1]) >= v.size())) { + // ??? + vx[k] = static_cast(0.0); + vy[k] = static_cast(0.0); + } else { + vx[k] = v[vi * 3 + axes[0]]; + vy[k] = v[vi * 3 + axes[1]]; + } + } + real_t e0x = vx[1] - vx[0]; + real_t e0y = vy[1] - vy[0]; + real_t e1x = vx[2] - vx[1]; + real_t e1y = vy[2] - vy[1]; + real_t cross = e0x * e1y - e0y * e1x; + // if an internal angle + if (cross * area < static_cast(0.0)) { + guess_vert += 1; + continue; + } + + // check all other verts in case they are inside this triangle + bool overlap = false; + for (size_t otherVert = 3; otherVert < npolys; ++otherVert) { + size_t idx = (guess_vert + otherVert) % npolys; + + if (idx >= remainingFace.vertex_indices.size()) { + // ??? + continue; + } + + size_t ovi = size_t(remainingFace.vertex_indices[idx].v_idx); + + if (((ovi * 3 + axes[0]) >= v.size()) || + ((ovi * 3 + axes[1]) >= v.size())) { + // ??? + continue; + } + real_t tx = v[ovi * 3 + axes[0]]; + real_t ty = v[ovi * 3 + axes[1]]; + if (pnpoly(3, vx, vy, tx, ty)) { + overlap = true; + break; + } + } + + if (overlap) { + guess_vert += 1; + continue; + } + + // this triangle is an ear + { + index_t idx0, idx1, idx2; + idx0.vertex_index = ind[0].v_idx; + idx0.normal_index = ind[0].vn_idx; + idx0.texcoord_index = ind[0].vt_idx; + idx1.vertex_index = ind[1].v_idx; + idx1.normal_index = ind[1].vn_idx; + idx1.texcoord_index = ind[1].vt_idx; + idx2.vertex_index = ind[2].v_idx; + idx2.normal_index = ind[2].vn_idx; + idx2.texcoord_index = ind[2].vt_idx; + + shape->mesh.indices.push_back(idx0); + shape->mesh.indices.push_back(idx1); + shape->mesh.indices.push_back(idx2); + + shape->mesh.num_face_vertices.push_back(3); + shape->mesh.material_ids.push_back(material_id); + shape->mesh.smoothing_group_ids.push_back(face.smoothing_group_id); + } + + // remove v1 from the list + size_t removed_vert_index = (guess_vert + 1) % npolys; + while (removed_vert_index + 1 < npolys) { + remainingFace.vertex_indices[removed_vert_index] = + remainingFace.vertex_indices[removed_vert_index + 1]; + removed_vert_index += 1; + } + remainingFace.vertex_indices.pop_back(); + } + + if (remainingFace.vertex_indices.size() == 3) { + i0 = remainingFace.vertex_indices[0]; + i1 = remainingFace.vertex_indices[1]; + i2 = remainingFace.vertex_indices[2]; + { + index_t idx0, idx1, idx2; + idx0.vertex_index = i0.v_idx; + idx0.normal_index = i0.vn_idx; + idx0.texcoord_index = i0.vt_idx; + idx1.vertex_index = i1.v_idx; + idx1.normal_index = i1.vn_idx; + idx1.texcoord_index = i1.vt_idx; + idx2.vertex_index = i2.v_idx; + idx2.normal_index = i2.vn_idx; + idx2.texcoord_index = i2.vt_idx; + + shape->mesh.indices.push_back(idx0); + shape->mesh.indices.push_back(idx1); + shape->mesh.indices.push_back(idx2); + + shape->mesh.num_face_vertices.push_back(3); + shape->mesh.material_ids.push_back(material_id); + shape->mesh.smoothing_group_ids.push_back(face.smoothing_group_id); + } + } + } else { + for (size_t k = 0; k < npolys; k++) { + index_t idx; + idx.vertex_index = face.vertex_indices[k].v_idx; + idx.normal_index = face.vertex_indices[k].vn_idx; + idx.texcoord_index = face.vertex_indices[k].vt_idx; + shape->mesh.indices.push_back(idx); + } + + shape->mesh.num_face_vertices.push_back( + static_cast(npolys)); + shape->mesh.material_ids.push_back(material_id); // per face + shape->mesh.smoothing_group_ids.push_back( + face.smoothing_group_id); // per face + } + } + + shape->name = name; + shape->mesh.tags = tags; + } + + if (!lineGroup.empty()) { + shape->path.indices.swap(lineGroup); + } + + return true; +} + +// Split a string with specified delimiter character. +// http://stackoverflow.com/questions/236129/split-a-string-in-c +static void SplitString(const std::string &s, char delim, + std::vector &elems) { + std::stringstream ss; + ss.str(s); + std::string item; + while (std::getline(ss, item, delim)) { + elems.push_back(item); + } +} + +void LoadMtl(std::map *material_map, + std::vector *materials, std::istream *inStream, + std::string *warning) { + // Create a default material anyway. + material_t material; + InitMaterial(&material); + + // Issue 43. `d` wins against `Tr` since `Tr` is not in the MTL specification. + bool has_d = false; + bool has_tr = false; + + std::stringstream ss; + + std::string linebuf; + while (inStream->peek() != -1) { + safeGetline(*inStream, linebuf); + + // Trim trailing whitespace. + if (linebuf.size() > 0) { + linebuf = linebuf.substr(0, linebuf.find_last_not_of(" \t") + 1); + } + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // new mtl + if ((0 == strncmp(token, "newmtl", 6)) && IS_SPACE((token[6]))) { + // flush previous material. + if (!material.name.empty()) { + material_map->insert(std::pair( + material.name, static_cast(materials->size()))); + materials->push_back(material); + } + + // initial temporary material + InitMaterial(&material); + + has_d = false; + has_tr = false; + + // set new mtl name + token += 7; + { + std::stringstream sstr; + sstr << token; + material.name = sstr.str(); + } + continue; + } + + // ambient + if (token[0] == 'K' && token[1] == 'a' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.ambient[0] = r; + material.ambient[1] = g; + material.ambient[2] = b; + continue; + } + + // diffuse + if (token[0] == 'K' && token[1] == 'd' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.diffuse[0] = r; + material.diffuse[1] = g; + material.diffuse[2] = b; + continue; + } + + // specular + if (token[0] == 'K' && token[1] == 's' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.specular[0] = r; + material.specular[1] = g; + material.specular[2] = b; + continue; + } + + // transmittance + if ((token[0] == 'K' && token[1] == 't' && IS_SPACE((token[2]))) || + (token[0] == 'T' && token[1] == 'f' && IS_SPACE((token[2])))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.transmittance[0] = r; + material.transmittance[1] = g; + material.transmittance[2] = b; + continue; + } + + // ior(index of refraction) + if (token[0] == 'N' && token[1] == 'i' && IS_SPACE((token[2]))) { + token += 2; + material.ior = parseReal(&token); + continue; + } + + // emission + if (token[0] == 'K' && token[1] == 'e' && IS_SPACE(token[2])) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.emission[0] = r; + material.emission[1] = g; + material.emission[2] = b; + continue; + } + + // shininess + if (token[0] == 'N' && token[1] == 's' && IS_SPACE(token[2])) { + token += 2; + material.shininess = parseReal(&token); + continue; + } + + // illum model + if (0 == strncmp(token, "illum", 5) && IS_SPACE(token[5])) { + token += 6; + material.illum = parseInt(&token); + continue; + } + + // dissolve + if ((token[0] == 'd' && IS_SPACE(token[1]))) { + token += 1; + material.dissolve = parseReal(&token); + + if (has_tr) { + ss << "WARN: Both `d` and `Tr` parameters defined for \"" + << material.name << "\". Use the value of `d` for dissolve." + << std::endl; + } + has_d = true; + continue; + } + if (token[0] == 'T' && token[1] == 'r' && IS_SPACE(token[2])) { + token += 2; + if (has_d) { + // `d` wins. Ignore `Tr` value. + ss << "WARN: Both `d` and `Tr` parameters defined for \"" + << material.name << "\". Use the value of `d` for dissolve." + << std::endl; + } else { + // We invert value of Tr(assume Tr is in range [0, 1]) + // NOTE: Interpretation of Tr is application(exporter) dependent. For + // some application(e.g. 3ds max obj exporter), Tr = d(Issue 43) + material.dissolve = static_cast(1.0) - parseReal(&token); + } + has_tr = true; + continue; + } + + // PBR: roughness + if (token[0] == 'P' && token[1] == 'r' && IS_SPACE(token[2])) { + token += 2; + material.roughness = parseReal(&token); + continue; + } + + // PBR: metallic + if (token[0] == 'P' && token[1] == 'm' && IS_SPACE(token[2])) { + token += 2; + material.metallic = parseReal(&token); + continue; + } + + // PBR: sheen + if (token[0] == 'P' && token[1] == 's' && IS_SPACE(token[2])) { + token += 2; + material.sheen = parseReal(&token); + continue; + } + + // PBR: clearcoat thickness + if (token[0] == 'P' && token[1] == 'c' && IS_SPACE(token[2])) { + token += 2; + material.clearcoat_thickness = parseReal(&token); + continue; + } + + // PBR: clearcoat roughness + if ((0 == strncmp(token, "Pcr", 3)) && IS_SPACE(token[3])) { + token += 4; + material.clearcoat_roughness = parseReal(&token); + continue; + } + + // PBR: anisotropy + if ((0 == strncmp(token, "aniso", 5)) && IS_SPACE(token[5])) { + token += 6; + material.anisotropy = parseReal(&token); + continue; + } + + // PBR: anisotropy rotation + if ((0 == strncmp(token, "anisor", 6)) && IS_SPACE(token[6])) { + token += 7; + material.anisotropy_rotation = parseReal(&token); + continue; + } + + // ambient texture + if ((0 == strncmp(token, "map_Ka", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.ambient_texname), + &(material.ambient_texopt), token, + /* is_bump */ false); + continue; + } + + // diffuse texture + if ((0 == strncmp(token, "map_Kd", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.diffuse_texname), + &(material.diffuse_texopt), token, + /* is_bump */ false); + continue; + } + + // specular texture + if ((0 == strncmp(token, "map_Ks", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.specular_texname), + &(material.specular_texopt), token, + /* is_bump */ false); + continue; + } + + // specular highlight texture + if ((0 == strncmp(token, "map_Ns", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.specular_highlight_texname), + &(material.specular_highlight_texopt), token, + /* is_bump */ false); + continue; + } + + // bump texture + if ((0 == strncmp(token, "map_bump", 8)) && IS_SPACE(token[8])) { + token += 9; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // bump texture + if ((0 == strncmp(token, "map_Bump", 8)) && IS_SPACE(token[8])) { + token += 9; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // bump texture + if ((0 == strncmp(token, "bump", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // alpha texture + if ((0 == strncmp(token, "map_d", 5)) && IS_SPACE(token[5])) { + token += 6; + material.alpha_texname = token; + ParseTextureNameAndOption(&(material.alpha_texname), + &(material.alpha_texopt), token, + /* is_bump */ false); + continue; + } + + // displacement texture + if ((0 == strncmp(token, "disp", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.displacement_texname), + &(material.displacement_texopt), token, + /* is_bump */ false); + continue; + } + + // reflection map + if ((0 == strncmp(token, "refl", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.reflection_texname), + &(material.reflection_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: roughness texture + if ((0 == strncmp(token, "map_Pr", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.roughness_texname), + &(material.roughness_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: metallic texture + if ((0 == strncmp(token, "map_Pm", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.metallic_texname), + &(material.metallic_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: sheen texture + if ((0 == strncmp(token, "map_Ps", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.sheen_texname), + &(material.sheen_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: emissive texture + if ((0 == strncmp(token, "map_Ke", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.emissive_texname), + &(material.emissive_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: normal map texture + if ((0 == strncmp(token, "norm", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption( + &(material.normal_texname), &(material.normal_texopt), token, + /* is_bump */ false); // @fixme { is_bump will be true? } + continue; + } + + // unknown parameter + const char *_space = strchr(token, ' '); + if (!_space) { + _space = strchr(token, '\t'); + } + if (_space) { + std::ptrdiff_t len = _space - token; + std::string key(token, static_cast(len)); + std::string value = _space + 1; + material.unknown_parameter.insert( + std::pair(key, value)); + } + } + // flush last material. + material_map->insert(std::pair( + material.name, static_cast(materials->size()))); + materials->push_back(material); + + if (warning) { + (*warning) = ss.str(); + } +} + +bool MaterialFileReader::operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) { + std::string filepath; + + if (!m_mtlBaseDir.empty()) { + filepath = std::string(m_mtlBaseDir) + matId; + } else { + filepath = matId; + } + + std::ifstream matIStream(filepath.c_str()); + if (!matIStream) { + std::stringstream ss; + ss << "WARN: Material file [ " << filepath << " ] not found." << std::endl; + if (err) { + (*err) += ss.str(); + } + return false; + } + + std::string warning; + LoadMtl(matMap, materials, &matIStream, &warning); + + if (!warning.empty()) { + if (err) { + (*err) += warning; + } + } + + return true; +} + +bool MaterialStreamReader::operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) { + (void)matId; + if (!m_inStream) { + std::stringstream ss; + ss << "WARN: Material stream in error state. " << std::endl; + if (err) { + (*err) += ss.str(); + } + return false; + } + + std::string warning; + LoadMtl(matMap, materials, &m_inStream, &warning); + + if (!warning.empty()) { + if (err) { + (*err) += warning; + } + } + + return true; +} + +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + const char *filename, const char *mtl_basedir, + bool trianglulate, bool default_vcols_fallback) { + attrib->vertices.clear(); + attrib->normals.clear(); + attrib->texcoords.clear(); + attrib->colors.clear(); + shapes->clear(); + + std::stringstream errss; + + std::ifstream ifs(filename); + if (!ifs) { + errss << "Cannot open file [" << filename << "]" << std::endl; + if (err) { + (*err) = errss.str(); + } + return false; + } + + std::string baseDir = mtl_basedir ? mtl_basedir : ""; + if (!baseDir.empty()) { +#ifndef _WIN32 + const char dirsep = '/'; +#else + const char dirsep = '\\'; +#endif + if (baseDir[baseDir.length() - 1] != dirsep) baseDir += dirsep; + } + MaterialFileReader matFileReader(baseDir); + + return LoadObj(attrib, shapes, materials, err, &ifs, &matFileReader, + trianglulate, default_vcols_fallback); +} + +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + std::istream *inStream, MaterialReader *readMatFn /*= NULL*/, + bool triangulate, bool default_vcols_fallback) { + std::stringstream errss; + + std::vector v; + std::vector vn; + std::vector vt; + std::vector vc; + std::vector tags; + std::vector faceGroup; + std::vector lineGroup; + std::string name; + + // material + std::map material_map; + int material = -1; + + // smoothing group id + unsigned int current_smoothing_id = + 0; // Initial value. 0 means no smoothing. + + int greatest_v_idx = -1; + int greatest_vn_idx = -1; + int greatest_vt_idx = -1; + + shape_t shape; + + bool found_all_colors = true; + + size_t line_num = 0; + std::string linebuf; + while (inStream->peek() != -1) { + safeGetline(*inStream, linebuf); + + line_num++; + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // vertex + if (token[0] == 'v' && IS_SPACE((token[1]))) { + token += 2; + real_t x, y, z; + real_t r, g, b; + + found_all_colors &= parseVertexWithColor(&x, &y, &z, &r, &g, &b, &token); + + v.push_back(x); + v.push_back(y); + v.push_back(z); + + if (found_all_colors || default_vcols_fallback) { + vc.push_back(r); + vc.push_back(g); + vc.push_back(b); + } + + continue; + } + + // normal + if (token[0] == 'v' && token[1] == 'n' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; + parseReal3(&x, &y, &z, &token); + vn.push_back(x); + vn.push_back(y); + vn.push_back(z); + continue; + } + + // texcoord + if (token[0] == 'v' && token[1] == 't' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y; + parseReal2(&x, &y, &token); + vt.push_back(x); + vt.push_back(y); + continue; + } + + // line + if (token[0] == 'l' && IS_SPACE((token[1]))) { + token += 2; + + line_t line_cache; + bool end_line_bit = 0; + while (!IS_NEW_LINE(token[0])) { + // get index from string + int idx; + fixIndex(parseInt(&token), 0, &idx); + + size_t n = strspn(token, " \t\r"); + token += n; + + if (!end_line_bit) { + line_cache.idx0 = idx; + } else { + line_cache.idx1 = idx; + lineGroup.push_back(line_cache.idx0); + lineGroup.push_back(line_cache.idx1); + line_cache = line_t(); + } + end_line_bit = !end_line_bit; + } + + continue; + } + // face + if (token[0] == 'f' && IS_SPACE((token[1]))) { + token += 2; + token += strspn(token, " \t"); + + face_t face; + + face.smoothing_group_id = current_smoothing_id; + face.vertex_indices.reserve(3); + + while (!IS_NEW_LINE(token[0])) { + vertex_index_t vi; + if (!parseTriple(&token, static_cast(v.size() / 3), + static_cast(vn.size() / 3), + static_cast(vt.size() / 2), &vi)) { + if (err) { + (*err) = "Failed parse `f' line(e.g. zero value for face index).\n"; + } + return false; + } + + greatest_v_idx = greatest_v_idx > vi.v_idx ? greatest_v_idx : vi.v_idx; + greatest_vn_idx = greatest_vn_idx > vi.vn_idx ? greatest_vn_idx : vi.vn_idx; + greatest_vt_idx = greatest_vt_idx > vi.vt_idx ? greatest_vt_idx : vi.vt_idx; + + face.vertex_indices.push_back(vi); + size_t n = strspn(token, " \t\r"); + token += n; + } + + // replace with emplace_back + std::move on C++11 + faceGroup.push_back(face); + + continue; + } + + // use mtl + if ((0 == strncmp(token, "usemtl", 6)) && IS_SPACE((token[6]))) { + token += 7; + std::stringstream ss; + ss << token; + std::string namebuf = ss.str(); + + int newMaterialId = -1; + if (material_map.find(namebuf) != material_map.end()) { + newMaterialId = material_map[namebuf]; + } else { + // { error!! material not found } + } + + if (newMaterialId != material) { + // Create per-face material. Thus we don't add `shape` to `shapes` at + // this time. + // just clear `faceGroup` after `exportGroupsToShape()` call. + exportGroupsToShape(&shape, faceGroup, lineGroup, tags, material, name, + triangulate, v); + faceGroup.clear(); + material = newMaterialId; + } + + continue; + } + + // load mtl + if ((0 == strncmp(token, "mtllib", 6)) && IS_SPACE((token[6]))) { + if (readMatFn) { + token += 7; + + std::vector filenames; + SplitString(std::string(token), ' ', filenames); + + if (filenames.empty()) { + if (err) { + (*err) += + "WARN: Looks like empty filename for mtllib. Use default " + "material. \n"; + } + } else { + bool found = false; + for (size_t s = 0; s < filenames.size(); s++) { + std::string err_mtl; + bool ok = (*readMatFn)(filenames[s].c_str(), materials, + &material_map, &err_mtl); + if (err && (!err_mtl.empty())) { + (*err) += err_mtl; // This should be warn message. + } + + if (ok) { + found = true; + break; + } + } + + if (!found) { + if (err) { + (*err) += + "WARN: Failed to load material file(s). Use default " + "material.\n"; + } + } + } + } + + continue; + } + + // group name + if (token[0] == 'g' && IS_SPACE((token[1]))) { + // flush previous face group. + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, + material, name, triangulate, v); + (void)ret; // return value not used. + + if (shape.mesh.indices.size() > 0) { + shapes->push_back(shape); + } + + shape = shape_t(); + + // material = -1; + faceGroup.clear(); + + std::vector names; + + while (!IS_NEW_LINE(token[0])) { + std::string str = parseString(&token); + names.push_back(str); + token += strspn(token, " \t\r"); // skip tag + } + + // names[0] must be 'g' + + if (names.size() < 2) { + // 'g' with empty names + if (err) { + std::stringstream ss; + ss << "WARN: Empty group name. line: " << line_num << "\n"; + (*err) += ss.str(); + name = ""; + } + } else { + + std::stringstream ss; + ss << names[1]; + + // tinyobjloader does not support multiple groups for a primitive. + // Currently we concatinate multiple group names with a space to get + // single group name. + + for (size_t i = 2; i < names.size(); i++) { + ss << " " << names[i]; + } + + name = ss.str(); + + } + + continue; + } + + // object name + if (token[0] == 'o' && IS_SPACE((token[1]))) { + // flush previous face group. + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, + material, name, triangulate, v); + if (ret) { + shapes->push_back(shape); + } + + // material = -1; + faceGroup.clear(); + shape = shape_t(); + + // @todo { multiple object name? } + token += 2; + std::stringstream ss; + ss << token; + name = ss.str(); + + continue; + } + + if (token[0] == 't' && IS_SPACE(token[1])) { + const int max_tag_nums = 8192; // FIXME(syoyo): Parameterize. + tag_t tag; + + token += 2; + + tag.name = parseString(&token); + + tag_sizes ts = parseTagTriple(&token); + + if (ts.num_ints < 0) { + ts.num_ints = 0; + } + if (ts.num_ints > max_tag_nums) { + ts.num_ints = max_tag_nums; + } + + if (ts.num_reals < 0) { + ts.num_reals = 0; + } + if (ts.num_reals > max_tag_nums) { + ts.num_reals = max_tag_nums; + } + + if (ts.num_strings < 0) { + ts.num_strings = 0; + } + if (ts.num_strings > max_tag_nums) { + ts.num_strings = max_tag_nums; + } + + tag.intValues.resize(static_cast(ts.num_ints)); + + for (size_t i = 0; i < static_cast(ts.num_ints); ++i) { + tag.intValues[i] = parseInt(&token); + } + + tag.floatValues.resize(static_cast(ts.num_reals)); + for (size_t i = 0; i < static_cast(ts.num_reals); ++i) { + tag.floatValues[i] = parseReal(&token); + } + + tag.stringValues.resize(static_cast(ts.num_strings)); + for (size_t i = 0; i < static_cast(ts.num_strings); ++i) { + tag.stringValues[i] = parseString(&token); + } + + tags.push_back(tag); + + continue; + } + + if (token[0] == 's' && IS_SPACE(token[1])) { + // smoothing group id + token += 2; + + // skip space. + token += strspn(token, " \t"); // skip space + + if (token[0] == '\0') { + continue; + } + + if (token[0] == '\r' || token[1] == '\n') { + continue; + } + + if (strlen(token) >= 3) { + if (token[0] == 'o' && token[1] == 'f' && token[2] == 'f') { + current_smoothing_id = 0; + } + } else { + // assume number + int smGroupId = parseInt(&token); + if (smGroupId < 0) { + // parse error. force set to 0. + // FIXME(syoyo): Report warning. + current_smoothing_id = 0; + } else { + current_smoothing_id = static_cast(smGroupId); + } + } + + continue; + } // smoothing group id + + // Ignore unknown command. + } + + // not all vertices have colors, no default colors desired? -> clear colors + if (!found_all_colors && !default_vcols_fallback) { + vc.clear(); + } + + if (greatest_v_idx >= static_cast(v.size() / 3)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + if (greatest_vn_idx >= static_cast(vn.size() / 3)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex normal indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + if (greatest_vt_idx >= static_cast(vt.size() / 2)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex texcoord indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, material, + name, triangulate, v); + // exportGroupsToShape return false when `usemtl` is called in the last + // line. + // we also add `shape` to `shapes` when `shape.mesh` has already some + // faces(indices) + if (ret || shape.mesh.indices.size()) { + shapes->push_back(shape); + } + faceGroup.clear(); // for safety + + if (err) { + (*err) += errss.str(); + } + + attrib->vertices.swap(v); + attrib->normals.swap(vn); + attrib->texcoords.swap(vt); + attrib->colors.swap(vc); + + return true; +} + +bool LoadObjWithCallback(std::istream &inStream, const callback_t &callback, + void *user_data /*= NULL*/, + MaterialReader *readMatFn /*= NULL*/, + std::string *err /*= NULL*/) { + std::stringstream errss; + + // material + std::map material_map; + int material_id = -1; // -1 = invalid + + std::vector indices; + std::vector materials; + std::vector names; + names.reserve(2); + std::vector names_out; + + std::string linebuf; + while (inStream.peek() != -1) { + safeGetline(inStream, linebuf); + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // vertex + if (token[0] == 'v' && IS_SPACE((token[1]))) { + token += 2; + // TODO(syoyo): Support parsing vertex color extension. + real_t x, y, z, w; // w is optional. default = 1.0 + parseV(&x, &y, &z, &w, &token); + if (callback.vertex_cb) { + callback.vertex_cb(user_data, x, y, z, w); + } + continue; + } + + // normal + if (token[0] == 'v' && token[1] == 'n' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; + parseReal3(&x, &y, &z, &token); + if (callback.normal_cb) { + callback.normal_cb(user_data, x, y, z); + } + continue; + } + + // texcoord + if (token[0] == 'v' && token[1] == 't' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; // y and z are optional. default = 0.0 + parseReal3(&x, &y, &z, &token); + if (callback.texcoord_cb) { + callback.texcoord_cb(user_data, x, y, z); + } + continue; + } + + // face + if (token[0] == 'f' && IS_SPACE((token[1]))) { + token += 2; + token += strspn(token, " \t"); + + indices.clear(); + while (!IS_NEW_LINE(token[0])) { + vertex_index_t vi = parseRawTriple(&token); + + index_t idx; + idx.vertex_index = vi.v_idx; + idx.normal_index = vi.vn_idx; + idx.texcoord_index = vi.vt_idx; + + indices.push_back(idx); + size_t n = strspn(token, " \t\r"); + token += n; + } + + if (callback.index_cb && indices.size() > 0) { + callback.index_cb(user_data, &indices.at(0), + static_cast(indices.size())); + } + + continue; + } + + // use mtl + if ((0 == strncmp(token, "usemtl", 6)) && IS_SPACE((token[6]))) { + token += 7; + std::stringstream ss; + ss << token; + std::string namebuf = ss.str(); + + int newMaterialId = -1; + if (material_map.find(namebuf) != material_map.end()) { + newMaterialId = material_map[namebuf]; + } else { + // { error!! material not found } + } + + if (newMaterialId != material_id) { + material_id = newMaterialId; + } + + if (callback.usemtl_cb) { + callback.usemtl_cb(user_data, namebuf.c_str(), material_id); + } + + continue; + } + + // load mtl + if ((0 == strncmp(token, "mtllib", 6)) && IS_SPACE((token[6]))) { + if (readMatFn) { + token += 7; + + std::vector filenames; + SplitString(std::string(token), ' ', filenames); + + if (filenames.empty()) { + if (err) { + (*err) += + "WARN: Looks like empty filename for mtllib. Use default " + "material. \n"; + } + } else { + bool found = false; + for (size_t s = 0; s < filenames.size(); s++) { + std::string err_mtl; + bool ok = (*readMatFn)(filenames[s].c_str(), &materials, + &material_map, &err_mtl); + if (err && (!err_mtl.empty())) { + (*err) += err_mtl; // This should be warn message. + } + + if (ok) { + found = true; + break; + } + } + + if (!found) { + if (err) { + (*err) += + "WARN: Failed to load material file(s). Use default " + "material.\n"; + } + } else { + if (callback.mtllib_cb) { + callback.mtllib_cb(user_data, &materials.at(0), + static_cast(materials.size())); + } + } + } + } + + continue; + } + + // group name + if (token[0] == 'g' && IS_SPACE((token[1]))) { + names.clear(); + + while (!IS_NEW_LINE(token[0])) { + std::string str = parseString(&token); + names.push_back(str); + token += strspn(token, " \t\r"); // skip tag + } + + assert(names.size() > 0); + + if (callback.group_cb) { + if (names.size() > 1) { + // create const char* array. + names_out.resize(names.size() - 1); + for (size_t j = 0; j < names_out.size(); j++) { + names_out[j] = names[j + 1].c_str(); + } + callback.group_cb(user_data, &names_out.at(0), + static_cast(names_out.size())); + + } else { + callback.group_cb(user_data, NULL, 0); + } + } + + continue; + } + + // object name + if (token[0] == 'o' && IS_SPACE((token[1]))) { + // @todo { multiple object name? } + token += 2; + + std::stringstream ss; + ss << token; + std::string object_name = ss.str(); + + if (callback.object_cb) { + callback.object_cb(user_data, object_name.c_str()); + } + + continue; + } + +#if 0 // @todo + if (token[0] == 't' && IS_SPACE(token[1])) { + tag_t tag; + + token += 2; + std::stringstream ss; + ss << token; + tag.name = ss.str(); + + token += tag.name.size() + 1; + + tag_sizes ts = parseTagTriple(&token); + + tag.intValues.resize(static_cast(ts.num_ints)); + + for (size_t i = 0; i < static_cast(ts.num_ints); ++i) { + tag.intValues[i] = atoi(token); + token += strcspn(token, "/ \t\r") + 1; + } + + tag.floatValues.resize(static_cast(ts.num_reals)); + for (size_t i = 0; i < static_cast(ts.num_reals); ++i) { + tag.floatValues[i] = parseReal(&token); + token += strcspn(token, "/ \t\r") + 1; + } + + tag.stringValues.resize(static_cast(ts.num_strings)); + for (size_t i = 0; i < static_cast(ts.num_strings); ++i) { + std::stringstream ss; + ss << token; + tag.stringValues[i] = ss.str(); + token += tag.stringValues[i].size() + 1; + } + + tags.push_back(tag); + } +#endif + + // Ignore unknown command. + } + + if (err) { + (*err) += errss.str(); + } + + return true; +} + +#ifdef __clang__ +#pragma clang diagnostic pop +#endif +} // namespace tinyobj + +#endif diff --git a/img/antialiasing.png b/img/antialiasing.png new file mode 100644 index 0000000..17c9111 Binary files /dev/null and b/img/antialiasing.png differ diff --git a/img/antialiasingBlooper.png b/img/antialiasingBlooper.png new file mode 100644 index 0000000..b004a83 Binary files /dev/null and b/img/antialiasingBlooper.png differ diff --git a/img/antialiasing_5000.png b/img/antialiasing_5000.png new file mode 100644 index 0000000..91f4490 Binary files /dev/null and b/img/antialiasing_5000.png differ diff --git a/img/compaction_blooper.png b/img/compaction_blooper.png new file mode 100644 index 0000000..14acf05 Binary files /dev/null and b/img/compaction_blooper.png differ diff --git a/img/cornell.2018-09-27_02-14-54z.1000sampNoaitialiasing.png b/img/cornell.2018-09-27_02-14-54z.1000sampNoaitialiasing.png new file mode 100644 index 0000000..fd8b60f Binary files /dev/null and b/img/cornell.2018-09-27_02-14-54z.1000sampNoaitialiasing.png differ diff --git a/img/cornell.2018-09-27_02-15-49z.1000sampAntialiasing.png b/img/cornell.2018-09-27_02-15-49z.1000sampAntialiasing.png new file mode 100644 index 0000000..ea701ec Binary files /dev/null and b/img/cornell.2018-09-27_02-15-49z.1000sampAntialiasing.png differ diff --git a/img/dof_12_1.png b/img/dof_12_1.png new file mode 100644 index 0000000..e1fae7f Binary files /dev/null and b/img/dof_12_1.png differ diff --git a/img/dof_14_1.png b/img/dof_14_1.png new file mode 100644 index 0000000..ff56905 Binary files /dev/null and b/img/dof_14_1.png differ diff --git a/img/dof_7_0.4.png b/img/dof_7_0.4.png new file mode 100644 index 0000000..455a824 Binary files /dev/null and b/img/dof_7_0.4.png differ diff --git a/img/mesh_object.png b/img/mesh_object.png new file mode 100644 index 0000000..f11c9fc Binary files /dev/null and b/img/mesh_object.png differ diff --git a/img/motion_blur.png b/img/motion_blur.png new file mode 100644 index 0000000..4fb8f1c Binary files /dev/null and b/img/motion_blur.png differ diff --git a/img/no_antialiasing.png b/img/no_antialiasing.png new file mode 100644 index 0000000..74f2ab5 Binary files /dev/null and b/img/no_antialiasing.png differ diff --git a/img/open_closed.png b/img/open_closed.png new file mode 100644 index 0000000..74887d6 Binary files /dev/null and b/img/open_closed.png differ diff --git a/img/performance_cach.png b/img/performance_cach.png new file mode 100644 index 0000000..92ea223 Binary files /dev/null and b/img/performance_cach.png differ diff --git a/img/performance_sort.png b/img/performance_sort.png new file mode 100644 index 0000000..060f55c Binary files /dev/null and b/img/performance_sort.png differ diff --git a/img/refraction.png b/img/refraction.png new file mode 100644 index 0000000..2ad351f Binary files /dev/null and b/img/refraction.png differ diff --git a/img/refraction_blooper.png b/img/refraction_blooper.png new file mode 100644 index 0000000..6514dfe Binary files /dev/null and b/img/refraction_blooper.png differ diff --git a/img/regular5000.png b/img/regular5000.png new file mode 100644 index 0000000..1bcaa68 Binary files /dev/null and b/img/regular5000.png differ diff --git a/img/stochastic.png b/img/stochastic.png new file mode 100644 index 0000000..f1ba285 Binary files /dev/null and b/img/stochastic.png differ diff --git a/img/thinlens.png b/img/thinlens.png new file mode 100644 index 0000000..5795bea Binary files /dev/null and b/img/thinlens.png differ diff --git a/scenes/closed.txt b/scenes/closed.txt new file mode 100644 index 0000000..7ff176b --- /dev/null +++ b/scenes/closed.txt @@ -0,0 +1,163 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular reflective purple +MATERIAL 4 +RGB .69 .14 .94 +SPECEX 0 +SPECRGB .69 .69 .69 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular transmissive yellow +MATERIAL 5 +RGB .98 .98 0 +SPECEX 0 +SPECRGB .98 .98 0 +REFL 0 +REFR 1 +REFRIOR 1.3 +EMITTANCE 0 + +// Specular reflective white + diffusive light blue +MATERIAL 6 +RGB .68 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 1 +DEPTH 16 +FILE cornell +EYE 0.0 5 8 +LOOKAT 0 5 0 +UP 0 1 0 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 20 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 20 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -10 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 20 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 20 + +//Specular transmissive yellow sphere +// Sphere +OBJECT 6 +sphere +material 5 +TRANS 0 4 -1 +ROTAT 0 0 0 +SCALE 5 5 5 + + +//Specular reflective purple sphere +OBJECT 7 +sphere +material 4 +TRANS 3 3 0 +ROTAT 0 0 0 +SCALE 2 2 2 + +//Specular reflective white + diffusive light blue Sphere +OBJECT 8 +sphere +material 6 +TRANS -2 4 1 +ROTAT 0 0 0 +SCALE 2 2 2 + +//Diffusive white wall closing up the scene +//OBJECT 9 +//cube +//material 1 +TRANS 0 5 10 +ROTAT 0 -90 0 +SCALE .01 10 10 \ No newline at end of file diff --git a/scenes/cornell.txt b/scenes/cornell.txt index 83ff820..65fea75 100644 --- a/scenes/cornell.txt +++ b/scenes/cornell.txt @@ -52,8 +52,8 @@ EMITTANCE 0 CAMERA RES 800 800 FOVY 45 -ITERATIONS 5000 -DEPTH 8 +ITERATIONS 1 +DEPTH 16 FILE cornell EYE 0.0 5 10.5 LOOKAT 0 5 0 @@ -114,4 +114,4 @@ sphere material 4 TRANS -1 4 -1 ROTAT 0 0 0 -SCALE 3 3 3 +SCALE 3 3 3 \ No newline at end of file diff --git a/scenes/cornell1.txt b/scenes/cornell1.txt new file mode 100644 index 0000000..cb818cf --- /dev/null +++ b/scenes/cornell1.txt @@ -0,0 +1,130 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// diffusive +MATERIAL 5 +RGB .85 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 1000 +DEPTH 20 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 +FOCALDIST 10 +LENSRAD 0.5 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 6 +mesh +material 4 +TRANS -1 2 0 +ROTAT 0 0 0 +SCALE 3 3 3 +FILE dodecahedron.obj \ No newline at end of file diff --git a/scenes/dodecahedron.txt b/scenes/dodecahedron.txt new file mode 100644 index 0000000..f96f09c --- /dev/null +++ b/scenes/dodecahedron.txt @@ -0,0 +1,138 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// diffusive +MATERIAL 5 +RGB .85 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 3000 +DEPTH 20 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 +FOCALDIST 10 +LENSRAD 0.5 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS -3 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Ceiling light +OBJECT 1 +cube +material 0 +TRANS 3 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 2 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 3 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 4 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 5 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 6 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 7 +mesh +material 4 +TRANS 5 -5 0 +ROTAT 0 0 0 +SCALE 10 10 10 +FILE dodecahedron.obj \ No newline at end of file diff --git a/scenes/dof.txt b/scenes/dof.txt new file mode 100644 index 0000000..ae54c50 --- /dev/null +++ b/scenes/dof.txt @@ -0,0 +1,172 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse yellow +MATERIAL 5 +RGB .85 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse +MATERIAL 6 +RGB .35 .35 .85 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse +MATERIAL 7 +RGB .85 .35 .85 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 5000 +DEPTH 8 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 +FOCALDIST 7 +LENSRAD 0.3 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 5 .3 5 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 6 +sphere +material 4 +TRANS -1 4 -1 +ROTAT 0 0 0 +SCALE 3 3 3 + +// Sphere +OBJECT 7 +cube +material 5 +TRANS -4 1 3 +ROTAT 0 0 0 +SCALE 2 2 2 + +OBJECT 8 +cube +material 6 +TRANS 2 2 4 +ROTAT 0 0 0 +SCALE 1 1 1 + +// Sphere +OBJECT 9 +sphere +material 7 +TRANS 4 4 -3 +ROTAT 0 0 0 +SCALE 2 2 2 diff --git a/scenes/motion_blur.txt b/scenes/motion_blur.txt new file mode 100644 index 0000000..02e4caf --- /dev/null +++ b/scenes/motion_blur.txt @@ -0,0 +1,118 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 5000 +DEPTH 8 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 6 +sphere +material 4 +TRANS -1 4 -1 +ROTAT 0 0 0 +SCALE 3 3 3 +MOTH 2.0 2.0 2.0 \ No newline at end of file diff --git a/scenes/refraction.txt b/scenes/refraction.txt new file mode 100644 index 0000000..e79827d --- /dev/null +++ b/scenes/refraction.txt @@ -0,0 +1,117 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0s + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular refract white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 0 +REFR 1 +REFRIOR 1.5 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 5000 +DEPTH 8 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 6 +sphere +material 4 +TRANS -1 4 -1 +ROTAT 0 0 0 +SCALE 3 3 3 \ No newline at end of file diff --git a/scenes/wahoo.mtl b/scenes/wahoo.mtl new file mode 100644 index 0000000..c89145d --- /dev/null +++ b/scenes/wahoo.mtl @@ -0,0 +1,6 @@ +newmtl initialShadingGroup +illum 4 +Kd 0.50 0.50 0.50 +Ka 0.00 0.00 0.00 +Tf 1.00 1.00 1.00 +Ni 1.00 diff --git a/scenes/wahoo.txt b/scenes/wahoo.txt new file mode 100644 index 0000000..5bcbfd0 --- /dev/null +++ b/scenes/wahoo.txt @@ -0,0 +1,130 @@ +// Emissive material (light) +MATERIAL 0 +RGB 1 1 1 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 5 + +// Diffuse white +MATERIAL 1 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse red +MATERIAL 2 +RGB .85 .35 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Diffuse green +MATERIAL 3 +RGB .35 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Specular white +MATERIAL 4 +RGB .98 .98 .98 +SPECEX 0 +SPECRGB .98 .98 .98 +REFL 1 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// diffusive +MATERIAL 5 +RGB .85 .85 .35 +SPECEX 0 +SPECRGB 0 0 0 +REFL 0 +REFR 0 +REFRIOR 0 +EMITTANCE 0 + +// Camera +CAMERA +RES 800 800 +FOVY 45 +ITERATIONS 3000 +DEPTH 20 +FILE cornell +EYE 0.0 5 10.5 +LOOKAT 0 5 0 +UP 0 1 0 +FOCALDIST 10 +LENSRAD 0.5 + + +// Ceiling light +OBJECT 0 +cube +material 0 +TRANS 0 10 0 +ROTAT 0 0 0 +SCALE 3 .3 3 + +// Floor +OBJECT 1 +cube +material 1 +TRANS 0 0 0 +ROTAT 0 0 0 +SCALE 10 .01 10 + +// Ceiling +OBJECT 2 +cube +material 1 +TRANS 0 10 0 +ROTAT 0 0 90 +SCALE .01 10 10 + +// Back wall +OBJECT 3 +cube +material 1 +TRANS 0 5 -5 +ROTAT 0 90 0 +SCALE .01 10 10 + +// Left wall +OBJECT 4 +cube +material 2 +TRANS -5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Right wall +OBJECT 5 +cube +material 3 +TRANS 5 5 0 +ROTAT 0 0 0 +SCALE .01 10 10 + +// Sphere +OBJECT 6 +mesh +material 4 +TRANS -5 -5 10 +ROTAT 0 0 0 +SCALE 1 1 1 +FILE wahoo.obj \ No newline at end of file diff --git a/src/CMakeLists.txt b/src/CMakeLists.txt index a1cb3fb..093a234 100644 --- a/src/CMakeLists.txt +++ b/src/CMakeLists.txt @@ -15,9 +15,11 @@ set(SOURCE_FILES "preview.cpp" "utilities.cpp" "utilities.h" + "tiny_obj_loader.cc" + "tiny_obj_loader.h" ) cuda_add_library(src ${SOURCE_FILES} - OPTIONS -arch=sm_20 + OPTIONS -arch=sm_30 ) diff --git a/src/interactions.h b/src/interactions.h index 5ce3628..7259d73 100644 --- a/src/interactions.h +++ b/src/interactions.h @@ -66,6 +66,8 @@ glm::vec3 calculateRandomDirectionInHemisphere( * * You may need to change the parameter list for your purposes! */ + +float shift = 0.001f; __host__ __device__ void scatterRay( PathSegment & pathSegment, @@ -76,4 +78,142 @@ void scatterRay( // TODO: implement this. // A basic implementation of pure-diffuse shading will just call the // calculateRandomDirectionInHemisphere defined above. + + + if (m.emittance > 0.0f) { + // emittance + pathSegment.remainingBounces = 0; + pathSegment.color *= (m.color * m.emittance); + return; + } + //// diffuse reflection + //pathSegment.color *= m.color; + //pathSegment.ray.direction = calculateRandomDirectionInHemisphere(normal, rng); + //pathSegment.ray.origin = intersect + normal * 0.001f; + //pathSegment.remainingBounces--; + + //if (m.hasReflective == 0.f && m.hasRefractive == 0.f) { + // pathSegment.color *= m.color; + // pathSegment.ray.direction = calculateRandomDirectionInHemisphere(normal, rng); + // pathSegment.ray.origin = intersect + normal * 0.001f; + // pathSegment.remainingBounces--; + // return; + //} + //int types = 1; + //if (m.hasReflective > 0.f) { + // types++; + //} + //if (m.hasRefractive > 0.f) { + // types++; + //} + //thrust::uniform_int_distribution uTypes(0, types - 1); + //int randomType = uTypes(rng); + //// diffusion + //if (randomType == 0) { + // pathSegment.color *= m.color; + // pathSegment.ray.direction = calculateRandomDirectionInHemisphere(normal, rng); + // pathSegment.ray.origin = intersect + normal * 0.001f; + // pathSegment.remainingBounces--; + // return; + //} + //// reflection + //if ((randomType == 1 && types == 2 && m.hasReflective) || (randomType == 1 && types == 3)) { + // pathSegment.ray.direction = glm::reflect(pathSegment.ray.direction, normal); + // pathSegment.color *= m.specular.color; + // pathSegment.color *= glm::abs(glm::dot(pathSegment.ray.direction, normal)) * m.color; + // pathSegment.ray.origin = intersect + normal * 0.001f; + // pathSegment.remainingBounces--; + //} + //// refraction + //else { + // glm::vec3 ray = pathSegment.ray.direction; + // float cosTheta = glm::dot(ray, normal); + // float etaIn, etaOut; + // if (cosTheta < 0) { + // etaIn = 1.f; + // etaOut = m.indexOfRefraction; + // } + // else { + // etaIn = m.indexOfRefraction; + // etaOut = 1.f; + // } + // //cosTheta = std::abs(cosTheta); + // float R0 = (etaIn - etaOut) / (etaIn + etaOut); + // R0 *= R0; + // float R = R0 + (1 - R0) * pow(1 - cosTheta, 5); + // thrust::uniform_real_distribution uRefr(0, 1); + // float refr = uRefr(rng); + // if (refr < 1) { + // pathSegment.ray.direction = glm::refract(pathSegment.ray.direction, normal, etaOut / etaIn); + // pathSegment.color *= m.color; + // pathSegment.ray.origin = intersect + normal * 0.001f; + // pathSegment.remainingBounces--; + // } + // else { + // pathSegment.ray.direction = glm::reflect(pathSegment.ray.direction, normal); + // pathSegment.color *= m.specular.color; + // pathSegment.color *= m.color; + // pathSegment.ray.origin = intersect + normal * 0.001f; + // pathSegment.remainingBounces--; + // } + //} + + if (m.hasReflective > 0.f) { + thrust::uniform_real_distribution u01(0, 1); + float probability = u01(rng); + if (probability > 0.5) { + pathSegment.ray.direction = glm::reflect(pathSegment.ray.direction, normal); + pathSegment.color *= m.specular.color; + pathSegment.color *= m.color; + pathSegment.ray.origin = intersect + normal * 0.001f; + pathSegment.remainingBounces--; + } + else { + pathSegment.color *= m.color; + pathSegment.ray.direction = calculateRandomDirectionInHemisphere(normal, rng); + pathSegment.ray.origin = intersect + normal * 0.001f; + pathSegment.remainingBounces--; + } + } + else if (m.hasRefractive > 0.f) { + glm::vec3 ray = pathSegment.ray.direction; + float cosTheta = glm::dot(normal, ray); + float etaIn, etaOut; + if (cosTheta > 0) { + etaIn = m.indexOfRefraction; + etaOut = 1.f; + } + else { + etaIn = 1.f; + etaOut = m.indexOfRefraction; + } + cosTheta = abs(cosTheta); + float R0 = (etaIn - etaOut) / (etaIn + etaOut); + R0 *= R0; + float R = R0 + (1 - R0) * pow(1 - cosTheta, 5); + thrust::uniform_real_distribution uRefr(0, 1); + float refr = uRefr(rng); + if (refr > R) { + pathSegment.ray.direction = glm::refract(pathSegment.ray.direction, normal, etaIn / etaOut); + pathSegment.color *= m.color; + pathSegment.ray.origin = intersect + normal * 0.001f; + pathSegment.remainingBounces--; + } + else { + pathSegment.ray.direction = glm::reflect(pathSegment.ray.direction, normal); + pathSegment.color *= m.specular.color; + pathSegment.color *= m.color; + pathSegment.ray.origin = intersect + normal * 0.001f; + pathSegment.remainingBounces--; + } + } + else { + pathSegment.color *= m.color; + pathSegment.ray.direction = calculateRandomDirectionInHemisphere(normal, rng); + pathSegment.ray.origin = intersect + normal * 0.001f; + pathSegment.remainingBounces--; + return; + } } + + diff --git a/src/intersections.h b/src/intersections.h index 6f23872..a71c892 100644 --- a/src/intersections.h +++ b/src/intersections.h @@ -142,3 +142,57 @@ __host__ __device__ float sphereIntersectionTest(Geom sphere, Ray r, return glm::length(r.origin - intersectionPoint); } + +__host__ __device__ float meshIntersectionTest(Geom geom, Ray r, + glm::vec3 &intersectionPoint, glm::vec3 &normal, bool &outside, Triangle *tris) { + + Ray q; + q.origin = multiplyMV(geom.inverseTransform, glm::vec4(r.origin, 1.0f)); + q.direction = glm::normalize(multiplyMV(geom.inverseTransform, glm::vec4(r.direction, 0.0f))); + + int startInd = geom.triStartIndex; + int endInd = geom.triEndIndex; + float dist = FLT_MAX; + for (int i = startInd; i <= endInd; i++) { + Triangle curT = tris[i]; + float curDist = FLT_MAX; + glm::vec3 position; + bool intersect = glm::intersectRayTriangle(r.origin, q.direction, curT.vertices[0], curT.vertices[1], curT.vertices[2], position); + glm::vec3 intersectionP = q.origin + q.direction * position.z; + if (intersect) { + curDist = position.z; + } + if (curDist < dist) { + dist = curDist; + normal = (curT.normals[0] + curT.normals[1] + curT.normals[2]) / 3.0f; + intersectionPoint = intersectionP; + } + + intersect = glm::intersectRayTriangle(r.origin, q.direction, curT.vertices[2], curT.vertices[1], curT.vertices[0], position); + intersectionP = q.origin + q.direction * position.z; + curDist = FLT_MAX; + if (intersect) { + curDist = position.z; + } + if (curDist < dist) { + dist = curDist; + normal = (curT.normals[0] + curT.normals[1] + curT.normals[2]) / 3.0f; + intersectionPoint = intersectionP; + } + } + if (dist == FLT_MAX) { + return -1; + } + + if (glm::dot(q.direction, normal) >= 0) { + outside = false; + normal = -normal; + } + else { + outside = true; + } + + intersectionPoint = multiplyMV(geom.transform, glm::vec4(normal, 1.f)); + normal = glm::normalize(multiplyMV(geom.invTranspose, glm::vec4(normal, 0.f))); + return glm::length(q.origin - intersectionPoint); +} diff --git a/src/pathtrace.cu b/src/pathtrace.cu index c1ec122..79d90f6 100644 --- a/src/pathtrace.cu +++ b/src/pathtrace.cu @@ -4,7 +4,9 @@ #include #include #include - +#include +#include +#include "stream_compaction/efficient.h" #include "sceneStructs.h" #include "scene.h" #include "glm/glm.hpp" @@ -13,58 +15,83 @@ #include "pathtrace.h" #include "intersections.h" #include "interactions.h" +#include #define ERRORCHECK 1 +#define TIMER 0 +#define SORT_BY_MATERIAL 0 +#define CACH_FIRST 0 +#define STREAM_COMPACTION_THRUST 1 +#define ANTIALIASING 0 +#define DEPTH_OF_FIELD 0 +#define MESH_LOADING 1 +#define MOTION_BLUR 0 #define FILENAME (strrchr(__FILE__, '/') ? strrchr(__FILE__, '/') + 1 : __FILE__) #define checkCUDAError(msg) checkCUDAErrorFn(msg, FILENAME, __LINE__) void checkCUDAErrorFn(const char *msg, const char *file, int line) { #if ERRORCHECK - cudaDeviceSynchronize(); - cudaError_t err = cudaGetLastError(); - if (cudaSuccess == err) { - return; - } - - fprintf(stderr, "CUDA error"); - if (file) { - fprintf(stderr, " (%s:%d)", file, line); - } - fprintf(stderr, ": %s: %s\n", msg, cudaGetErrorString(err)); + cudaDeviceSynchronize(); + cudaError_t err = cudaGetLastError(); + if (cudaSuccess == err) { + return; + } + + fprintf(stderr, "CUDA error"); + if (file) { + fprintf(stderr, " (%s:%d)", file, line); + } + fprintf(stderr, ": %s: %s\n", msg, cudaGetErrorString(err)); # ifdef _WIN32 - getchar(); + getchar(); # endif - exit(EXIT_FAILURE); + exit(EXIT_FAILURE); #endif } + +class path_alive { +public: + __host__ __device__ bool operator()(const PathSegment &pathSegments) { + return pathSegments.remainingBounces > 0; + } +}; + + +class MaterialComparator { +public: + __host__ __device__ bool operator() (const ShadeableIntersection& s1, const ShadeableIntersection& s2) { + return s1.materialId < s2.materialId; + } +}; + __host__ __device__ thrust::default_random_engine makeSeededRandomEngine(int iter, int index, int depth) { - int h = utilhash((1 << 31) | (depth << 22) | iter) ^ utilhash(index); - return thrust::default_random_engine(h); + int h = utilhash((1 << 31) | (depth << 22) | iter) ^ utilhash(index); + return thrust::default_random_engine(h); } //Kernel that writes the image to the OpenGL PBO directly. __global__ void sendImageToPBO(uchar4* pbo, glm::ivec2 resolution, - int iter, glm::vec3* image) { - int x = (blockIdx.x * blockDim.x) + threadIdx.x; - int y = (blockIdx.y * blockDim.y) + threadIdx.y; - - if (x < resolution.x && y < resolution.y) { - int index = x + (y * resolution.x); - glm::vec3 pix = image[index]; - - glm::ivec3 color; - color.x = glm::clamp((int) (pix.x / iter * 255.0), 0, 255); - color.y = glm::clamp((int) (pix.y / iter * 255.0), 0, 255); - color.z = glm::clamp((int) (pix.z / iter * 255.0), 0, 255); - - // Each thread writes one pixel location in the texture (textel) - pbo[index].w = 0; - pbo[index].x = color.x; - pbo[index].y = color.y; - pbo[index].z = color.z; - } + int iter, glm::vec3* image) { + int x = (blockIdx.x * blockDim.x) + threadIdx.x; + int y = (blockIdx.y * blockDim.y) + threadIdx.y; + + if (x < resolution.x && y < resolution.y) { + int index = x + (y * resolution.x); + glm::vec3 pix = image[index]; + + glm::ivec3 color; + color.x = glm::clamp((int)(pix.x / iter * 255.0), 0, 255); + color.y = glm::clamp((int)(pix.y / iter * 255.0), 0, 255); + color.z = glm::clamp((int)(pix.z / iter * 255.0), 0, 255); + + // Each thread writes one pixel location in the texture (textel) + pbo[index].w = 0; + pbo[index].x = color.x; + pbo[index].y = color.y; + pbo[index].z = color.z; + } } static Scene * hst_scene = NULL; @@ -75,40 +102,63 @@ static PathSegment * dev_paths = NULL; static ShadeableIntersection * dev_intersections = NULL; // TODO: static variables for device memory, any extra info you need, etc // ... +static ShadeableIntersection * dev_firstBounceIntersections = NULL; +static Triangle * dev_triangles = NULL; void pathtraceInit(Scene *scene) { - hst_scene = scene; - const Camera &cam = hst_scene->state.camera; - const int pixelcount = cam.resolution.x * cam.resolution.y; + hst_scene = scene; + const Camera &cam = hst_scene->state.camera; + const int pixelcount = cam.resolution.x * cam.resolution.y; - cudaMalloc(&dev_image, pixelcount * sizeof(glm::vec3)); - cudaMemset(dev_image, 0, pixelcount * sizeof(glm::vec3)); + cudaMalloc(&dev_image, pixelcount * sizeof(glm::vec3)); + cudaMemset(dev_image, 0, pixelcount * sizeof(glm::vec3)); - cudaMalloc(&dev_paths, pixelcount * sizeof(PathSegment)); + cudaMalloc(&dev_paths, pixelcount * sizeof(PathSegment)); - cudaMalloc(&dev_geoms, scene->geoms.size() * sizeof(Geom)); - cudaMemcpy(dev_geoms, scene->geoms.data(), scene->geoms.size() * sizeof(Geom), cudaMemcpyHostToDevice); + cudaMalloc(&dev_geoms, scene->geoms.size() * sizeof(Geom)); + cudaMemcpy(dev_geoms, scene->geoms.data(), scene->geoms.size() * sizeof(Geom), cudaMemcpyHostToDevice); - cudaMalloc(&dev_materials, scene->materials.size() * sizeof(Material)); - cudaMemcpy(dev_materials, scene->materials.data(), scene->materials.size() * sizeof(Material), cudaMemcpyHostToDevice); + cudaMalloc(&dev_materials, scene->materials.size() * sizeof(Material)); + cudaMemcpy(dev_materials, scene->materials.data(), scene->materials.size() * sizeof(Material), cudaMemcpyHostToDevice); - cudaMalloc(&dev_intersections, pixelcount * sizeof(ShadeableIntersection)); - cudaMemset(dev_intersections, 0, pixelcount * sizeof(ShadeableIntersection)); - - // TODO: initialize any extra device memeory you need + cudaMalloc(&dev_intersections, pixelcount * sizeof(ShadeableIntersection)); + cudaMemset(dev_intersections, 0, pixelcount * sizeof(ShadeableIntersection)); - checkCUDAError("pathtraceInit"); + // TODO: initialize any extra device memeory you need + cudaMalloc(&dev_firstBounceIntersections, pixelcount * sizeof(ShadeableIntersection)); + cudaMemset(dev_firstBounceIntersections, 0, pixelcount * sizeof(ShadeableIntersection)); + cudaMalloc(&dev_triangles, scene->tris.size() * sizeof(Triangle)); + cudaMemcpy(dev_triangles, scene->tris.data(), scene->tris.size() * sizeof(Triangle), cudaMemcpyHostToDevice); + checkCUDAError("pathtraceInit"); } void pathtraceFree() { - cudaFree(dev_image); // no-op if dev_image is null - cudaFree(dev_paths); - cudaFree(dev_geoms); - cudaFree(dev_materials); - cudaFree(dev_intersections); - // TODO: clean up any extra device memory you created - - checkCUDAError("pathtraceFree"); + cudaFree(dev_image); // no-op if dev_image is null + cudaFree(dev_paths); + cudaFree(dev_geoms); + cudaFree(dev_materials); + cudaFree(dev_intersections); + // TODO: clean up any extra device memory you created + cudaFree(dev_firstBounceIntersections); + cudaFree(dev_triangles); + checkCUDAError("pathtraceFree"); +} + +__device__ glm::vec2 ConcentricSampleDisk(glm::vec2 u) { + glm::vec2 uOffset = 2.f * u - glm::vec2(1.f, 1.f); + if (uOffset.x == 0 && uOffset.y == 0) { + return glm::vec2(0.f, 0.f); + } + float theta, r; + if (abs(uOffset.x) > abs(uOffset.y)) { + r = uOffset.x; + theta = PI / 4 * (uOffset.y / uOffset.x); + } + else { + r = uOffset.y; + theta = PI / 2 - PI / 4 * (uOffset.x / uOffset.y); + } + return r * glm::vec2(cos(theta), sin(theta)); } /** @@ -119,24 +169,43 @@ void pathtraceFree() { * motion blur - jitter rays "in time" * lens effect - jitter ray origin positions based on a lens */ -__global__ void generateRayFromCamera(Camera cam, int iter, int traceDepth, PathSegment* pathSegments) -{ +__global__ void generateRayFromCamera(Camera cam, int iter, int traceDepth, PathSegment* pathSegments) { int x = (blockIdx.x * blockDim.x) + threadIdx.x; int y = (blockIdx.y * blockDim.y) + threadIdx.y; - +#if ANTIALIASING || DEPTH_OF_FIELD + thrust::default_random_engine rng = makeSeededRandomEngine(iter, x, 0); + thrust::uniform_real_distribution u0505(-0.5, 0.5); + thrust::uniform_real_distribution u01(0, 1); +#endif if (x < cam.resolution.x && y < cam.resolution.y) { int index = x + (y * cam.resolution.x); PathSegment & segment = pathSegments[index]; segment.ray.origin = cam.position; - segment.color = glm::vec3(1.0f, 1.0f, 1.0f); + segment.color = glm::vec3(1.0f, 1.0f, 1.0f); // TODO: implement antialiasing by jittering the ray +#if ANTIALIASING + segment.ray.direction = glm::normalize(cam.view + - cam.right * cam.pixelLength.x * ((float)x + u0505(rng) - (float)cam.resolution.x * 0.5f) + - cam.up * cam.pixelLength.y * ((float)y + u0505(rng) - (float)cam.resolution.y * 0.5f) + ); +#else segment.ray.direction = glm::normalize(cam.view - cam.right * cam.pixelLength.x * ((float)x - (float)cam.resolution.x * 0.5f) - cam.up * cam.pixelLength.y * ((float)y - (float)cam.resolution.y * 0.5f) - ); + ); +#endif +#if DEPTH_OF_FIELD + if (cam.lensRadius > 0) { + glm::vec2 pLens = cam.lensRadius * ConcentricSampleDisk(glm::vec2(u01(rng), u01(rng))); + float ft = cam.focalLength / segment.ray.direction.z; + glm::vec3 pFocus = segment.ray.origin + segment.ray.direction * abs(ft); + segment.ray.origin += pLens.x * cam.right + pLens.y * cam.up; + segment.ray.direction = glm::normalize(pFocus - segment.ray.origin); + } +#endif segment.pixelIndex = index; segment.remainingBounces = traceDepth; } @@ -153,7 +222,9 @@ __global__ void computeIntersections( , Geom * geoms , int geoms_size , ShadeableIntersection * intersections - ) + , Triangle * dev_triangles + +) { int path_index = blockIdx.x * blockDim.x + threadIdx.x; @@ -176,7 +247,6 @@ __global__ void computeIntersections( for (int i = 0; i < geoms_size; i++) { Geom & geom = geoms[i]; - if (geom.type == CUBE) { t = boxIntersectionTest(geom, pathSegment.ray, tmp_intersect, tmp_normal, outside); @@ -186,7 +256,9 @@ __global__ void computeIntersections( t = sphereIntersectionTest(geom, pathSegment.ray, tmp_intersect, tmp_normal, outside); } // TODO: add more intersection tests here... triangle? metaball? CSG? - + else if (geom.type == MESH) { + t = meshIntersectionTest(geom, pathSegment.ray, tmp_intersect, tmp_normal, outside, dev_triangles); + } // Compute the minimum t from the intersection tests to determine what // scene geometry object was hit first. if (t > 0.0f && t_min > t) @@ -221,50 +293,56 @@ __global__ void computeIntersections( // Note that this shader does NOT do a BSDF evaluation! // Your shaders should handle that - this can allow techniques such as // bump mapping. -__global__ void shadeFakeMaterial ( - int iter - , int num_paths +__global__ void shadeFakeMaterial( + int iter + , int num_paths , ShadeableIntersection * shadeableIntersections , PathSegment * pathSegments , Material * materials - ) +) { - int idx = blockIdx.x * blockDim.x + threadIdx.x; - if (idx < num_paths) - { - ShadeableIntersection intersection = shadeableIntersections[idx]; - if (intersection.t > 0.0f) { // if the intersection exists... - // Set up the RNG - // LOOK: this is how you use thrust's RNG! Please look at - // makeSeededRandomEngine as well. - thrust::default_random_engine rng = makeSeededRandomEngine(iter, idx, 0); - thrust::uniform_real_distribution u01(0, 1); - - Material material = materials[intersection.materialId]; - glm::vec3 materialColor = material.color; - - // If the material indicates that the object was a light, "light" the ray - if (material.emittance > 0.0f) { - pathSegments[idx].color *= (materialColor * material.emittance); - } - // Otherwise, do some pseudo-lighting computation. This is actually more - // like what you would expect from shading in a rasterizer like OpenGL. - // TODO: replace this! you should be able to start with basically a one-liner - else { - float lightTerm = glm::dot(intersection.surfaceNormal, glm::vec3(0.0f, 1.0f, 0.0f)); - pathSegments[idx].color *= (materialColor * lightTerm) * 0.3f + ((1.0f - intersection.t * 0.02f) * materialColor) * 0.7f; - pathSegments[idx].color *= u01(rng); // apply some noise because why not - } - // If there was no intersection, color the ray black. - // Lots of renderers use 4 channel color, RGBA, where A = alpha, often - // used for opacity, in which case they can indicate "no opacity". - // This can be useful for post-processing and image compositing. - } else { - pathSegments[idx].color = glm::vec3(0.0f); - } - } + int idx = blockIdx.x * blockDim.x + threadIdx.x; + if (idx < num_paths) + { + ShadeableIntersection intersection = shadeableIntersections[idx]; + if (intersection.t > 0.0f) { // if the intersection exists... + // Set up the RNG + // LOOK: this is how you use thrust's RNG! Please look at + // makeSeededRandomEngine as well. + thrust::default_random_engine rng = makeSeededRandomEngine(iter, idx, 0); + thrust::uniform_real_distribution u01(0, 1); + + Material material = materials[intersection.materialId]; + glm::vec3 materialColor = material.color; + + // If the material indicates that the object was a light, "light" the ray + if (material.emittance > 0.0f) { + pathSegments[idx].color *= (materialColor * material.emittance); + pathSegments[idx].remainingBounces = 0; + } + // Otherwise, do some pseudo-lighting computation. This is actually more + // like what you would expect from shading in a rasterizer like OpenGL. + // TODO: replace this! you should be able to start with basically a one-liner + else { + glm::vec3 intersect = pathSegments[idx].ray.origin + pathSegments[idx].ray.direction * intersection.t; + scatterRay(pathSegments[idx], intersect, intersection.surfaceNormal, material, rng); + //float lightTerm = glm::dot(intersection.surfaceNormal, glm::vec3(0.0f, 1.0f, 0.0f)); + //pathSegments[idx].color *= (materialColor * lightTerm) * 0.3f + ((1.0f - intersection.t * 0.02f) * materialColor) * 0.7f; + //pathSegments[idx].color *= u01(rng); // apply some noise because why not + } + // If there was no intersection, color the ray black. + // Lots of renderers use 4 channel color, RGBA, where A = alpha, often + // used for opacity, in which case they can indicate "no opacity". + // This can be useful for post-processing and image compositing. + } + else { + pathSegments[idx].color = glm::vec3(0.0f); + pathSegments[idx].remainingBounces = 0; + } + } } + // Add the current iteration's output to the overall image __global__ void finalGather(int nPaths, glm::vec3 * image, PathSegment * iterationPaths) { @@ -277,117 +355,204 @@ __global__ void finalGather(int nPaths, glm::vec3 * image, PathSegment * iterati } } +void sort_by_material(int num_paths, ShadeableIntersection * shadeableIntersections, PathSegment * pathSegments) { + thrust::device_ptr dev_intersect_ptr(shadeableIntersections); + thrust::device_ptr dev_path_ptr(pathSegments); + thrust::sort_by_key(thrust::device, dev_intersect_ptr, dev_intersect_ptr + num_paths, dev_path_ptr, MaterialComparator()); +} + + + + /** - * Wrapper for the __global__ call that sets up the kernel calls and does a ton - * of memory management - */ +* Wrapper for the __global__ call that sets up the kernel calls and does a ton +* of memory management +*/ void pathtrace(uchar4 *pbo, int frame, int iter) { - const int traceDepth = hst_scene->state.traceDepth; - const Camera &cam = hst_scene->state.camera; - const int pixelcount = cam.resolution.x * cam.resolution.y; + const int traceDepth = hst_scene->state.traceDepth; + const Camera &cam = hst_scene->state.camera; + const int pixelcount = cam.resolution.x * cam.resolution.y; // 2D block for generating ray from camera - const dim3 blockSize2d(8, 8); - const dim3 blocksPerGrid2d( - (cam.resolution.x + blockSize2d.x - 1) / blockSize2d.x, - (cam.resolution.y + blockSize2d.y - 1) / blockSize2d.y); + const dim3 blockSize2d(8, 8); + const dim3 blocksPerGrid2d( + (cam.resolution.x + blockSize2d.x - 1) / blockSize2d.x, + (cam.resolution.y + blockSize2d.y - 1) / blockSize2d.y); // 1D block for path tracing const int blockSize1d = 128; - /////////////////////////////////////////////////////////////////////////// - - // Recap: - // * Initialize array of path rays (using rays that come out of the camera) - // * You can pass the Camera object to that kernel. - // * Each path ray must carry at minimum a (ray, color) pair, - // * where color starts as the multiplicative identity, white = (1, 1, 1). - // * This has already been done for you. - // * For each depth: - // * Compute an intersection in the scene for each path ray. - // A very naive version of this has been implemented for you, but feel - // free to add more primitives and/or a better algorithm. - // Currently, intersection distance is recorded as a parametric distance, - // t, or a "distance along the ray." t = -1.0 indicates no intersection. - // * Color is attenuated (multiplied) by reflections off of any object - // * TODO: Stream compact away all of the terminated paths. - // You may use either your implementation or `thrust::remove_if` or its - // cousins. - // * Note that you can't really use a 2D kernel launch any more - switch - // to 1D. - // * TODO: Shade the rays that intersected something or didn't bottom out. - // That is, color the ray by performing a color computation according - // to the shader, then generate a new ray to continue the ray path. - // We recommend just updating the ray's PathSegment in place. - // Note that this step may come before or after stream compaction, - // since some shaders you write may also cause a path to terminate. - // * Finally, add this iteration's results to the image. This has been done - // for you. - - // TODO: perform one iteration of path tracing - - generateRayFromCamera <<>>(cam, iter, traceDepth, dev_paths); + +#if MOTION_BLUR + for (int i = 0; i < hst_scene->geoms.size(); i++) { + float t = 0.0001; + Geom & motion_geom = hst_scene->geoms[i]; + if (motion_geom.hasMotion) { + motion_geom.translation += motion_geom.motion * t; + motion_geom.transform = utilityCore::buildTransformationMatrix(motion_geom.translation, motion_geom.rotation, motion_geom.scale); + motion_geom.inverseTransform = glm::inverse(motion_geom.transform); + motion_geom.invTranspose = glm::inverseTranspose(motion_geom.transform); + } + } + cudaMemcpy(dev_geoms, &(hst_scene->geoms)[0], hst_scene->geoms.size() * sizeof(Geom), cudaMemcpyHostToDevice); +#endif + + /////////////////////////////////////////////////////////////////////////// + + // Recap: + // * Initialize array of path rays (using rays that come out of the camera) + // * You can pass the Camera object to that kernel. + // * Each path ray must carry at minimum a (ray, color) pair, + // * where color starts as the multiplicative identity, white = (1, 1, 1). + // * This has already been done for you. + // * For each depth: + // * Compute an intersection in the scene for each path ray. + // A very naive version of this has been implemented for you, but feel + // free to add more primitives and/or a better algorithm. + // Currently, intersection distance is recorded as a parametric distance, + // t, or a "distance along the ray." t = -1.0 indicates no intersection. + // * Color is attenuated (multiplied) by reflections off of any object + // * TODO: Stream compact away all of the terminated paths. + // You may use either your implementation or `thrust::remove_if` or its + // cousins. + // * Note that you can't really use a 2D kernel launch any more - switch + // to 1D. + // * TODO: Shade the rays that intersected something or didn't bottom out. + // That is, color the ray by performing a color computation according + // to the shader, then generate a new ray to continue the ray path. + // We recommend just updating the ray's PathSegment in place. + // Note that this step may come before or after stream compaction, + // since some shaders you write may also cause a path to terminate. + // * Finally, add this iteration's results to the image. This has been done + // for you. + + // TODO: perform one iteration of path tracing +#if TIMER + StreamCompaction::Efficient::timer().startGpuTimer(); +#endif + generateRayFromCamera << > >(cam, iter, traceDepth, dev_paths); checkCUDAError("generate camera ray"); int depth = 0; PathSegment* dev_path_end = dev_paths + pixelcount; int num_paths = dev_path_end - dev_paths; + int total_path = num_paths; // --- PathSegment Tracing Stage --- // Shoot ray into scene, bounce between objects, push shading chunks - bool iterationComplete = false; + bool iterationComplete = false; while (!iterationComplete) { - // clean shading chunks - cudaMemset(dev_intersections, 0, pixelcount * sizeof(ShadeableIntersection)); + // clean shading chunks + cudaMemset(dev_intersections, 0, pixelcount * sizeof(ShadeableIntersection)); - // tracing - dim3 numblocksPathSegmentTracing = (num_paths + blockSize1d - 1) / blockSize1d; - computeIntersections <<>> ( - depth - , num_paths - , dev_paths - , dev_geoms - , hst_scene->geoms.size() - , dev_intersections - ); - checkCUDAError("trace one bounce"); - cudaDeviceSynchronize(); - depth++; - - - // TODO: - // --- Shading Stage --- - // Shade path segments based on intersections and generate new rays by - // evaluating the BSDF. - // Start off with just a big kernel that handles all the different - // materials you have in the scenefile. - // TODO: compare between directly shading the path segments and shading - // path segments that have been reshuffled to be contiguous in memory. - - shadeFakeMaterial<<>> ( - iter, - num_paths, - dev_intersections, - dev_paths, - dev_materials - ); - iterationComplete = true; // TODO: should be based off stream compaction results. + // tracing + dim3 numblocksPathSegmentTracing = (num_paths + blockSize1d - 1) / blockSize1d; + +#if CACH_FIRST + if (iter != 1 && depth == 0) { + cudaMemcpy(dev_intersections, dev_firstBounceIntersections, pixelcount * sizeof(ShadeableIntersection), cudaMemcpyDeviceToDevice); + checkCUDAError("cudaMemcpy from first bounce intersections"); + } + else if (iter == 1 && depth == 0) { + computeIntersections << > > ( + depth + , num_paths + , dev_paths + , dev_geoms + , hst_scene->geoms.size() + , dev_intersections + , dev_triangles + ); + checkCUDAError("trace one bounce"); + cudaDeviceSynchronize(); + cudaMemcpy(dev_firstBounceIntersections, dev_intersections, pixelcount * sizeof(ShadeableIntersection), cudaMemcpyDeviceToDevice); + checkCUDAError("cudaMemcpy to first bounce intersections"); + } + else { + computeIntersections << > > ( + depth + , num_paths + , dev_paths + , dev_geoms + , hst_scene->geoms.size() + , dev_intersections + , dev_triangles + ); + checkCUDAError("trace one bounce"); + cudaDeviceSynchronize(); + } +#else + computeIntersections << > > ( + depth + , num_paths + , dev_paths + , dev_geoms + , hst_scene->geoms.size() + , dev_intersections + , dev_triangles + ); + checkCUDAError("trace one bounce"); + cudaDeviceSynchronize(); +#endif + + depth++; + + + // TODO: + // --- Shading Stage --- + // Shade path segments based on intersections and generate new rays by + // evaluating the BSDF. + // Start off with just a big kernel that handles all the different + // materials you have in the scenefile. + // TODO: compare between directly shading the path segments and shading + // path segments that have been reshuffled to be contiguous in memory. +#if SORT_BY_MATERIAL + sort_by_material(num_paths, dev_intersections, dev_paths); +#endif + shadeFakeMaterial << > > ( + iter, + num_paths, + dev_intersections, + dev_paths, + dev_materials + ); + //iterationComplete = true; + + //PathSegment* iter_end = thrust::remove_if(thrust::device, dev_paths, dev_paths + num_paths, NoMoreBounce()); + //num_paths = thrust::count_if(thrust::device, dev_paths, dev_paths + num_paths, bounce_remaining());; + //iterationComplete = num_paths <= 0; // TODO: should be based off stream compaction results. +#if STREAM_COMPACTION_THRUST + PathSegment* iter_end = thrust::partition(thrust::device, dev_paths, dev_paths + num_paths, path_alive()); + num_paths = iter_end - dev_paths; + +#endif + + iterationComplete = num_paths <= 0 || depth >= traceDepth; } - // Assemble this iteration and apply it to the image - dim3 numBlocksPixels = (pixelcount + blockSize1d - 1) / blockSize1d; - finalGather<<>>(num_paths, dev_image, dev_paths); + // Assemble this iteration and apply it to the image + dim3 numBlocksPixels = (pixelcount + blockSize1d - 1) / blockSize1d; + finalGather << > >(total_path, dev_image, dev_paths); +#if TIMER + StreamCompaction::Efficient::timer().endGpuTimer(); + cout << StreamCompaction::Efficient::timer().getGpuElapsedTimeForPreviousOperation() << endl; +#endif + //ofstream outFile; + //outFile.open("data.csv, ios::out"); + //outFile << "iteration" << ',' << iter++ << ',' << "time" << ',' << StreamCompaction::Efficient::timer().getGpuElapsedTimeForPreviousOperation() << endl; + //outFile.close(); - /////////////////////////////////////////////////////////////////////////// + /////////////////////////////////////////////////////////////////////////// - // Send results to OpenGL buffer for rendering - sendImageToPBO<<>>(pbo, cam.resolution, iter, dev_image); + // Send results to OpenGL buffer for rendering + sendImageToPBO << > >(pbo, cam.resolution, iter, dev_image); - // Retrieve image from GPU - cudaMemcpy(hst_scene->state.image.data(), dev_image, - pixelcount * sizeof(glm::vec3), cudaMemcpyDeviceToHost); + // Retrieve image from GPU + cudaMemcpy(hst_scene->state.image.data(), dev_image, + pixelcount * sizeof(glm::vec3), cudaMemcpyDeviceToHost); - checkCUDAError("pathtrace"); + checkCUDAError("pathtrace"); } + diff --git a/src/pathtrace.h b/src/pathtrace.h index 1241227..1ad0f59 100644 --- a/src/pathtrace.h +++ b/src/pathtrace.h @@ -6,3 +6,4 @@ void pathtraceInit(Scene *scene); void pathtraceFree(); void pathtrace(uchar4 *pbo, int frame, int iteration); +void sort_by_material(int num_paths, ShadeableIntersection * shadeableIntersections, PathSegment * pathSegments); diff --git a/src/scene.cpp b/src/scene.cpp index cbae043..ca1e8d8 100644 --- a/src/scene.cpp +++ b/src/scene.cpp @@ -4,6 +4,9 @@ #include #include +#define TINYOBJLOADER_IMPLEMENTATION // define this in only *one* .cc +#include "tiny_obj_loader.h" + Scene::Scene(string filename) { cout << "Reading scene from " << filename << " ..." << endl; cout << " " << endl; @@ -51,7 +54,11 @@ int Scene::loadGeom(string objectid) { } else if (strcmp(line.c_str(), "cube") == 0) { cout << "Creating new cube..." << endl; newGeom.type = CUBE; - } + } + else if (strcmp(line.c_str(), "mesh") == 0) { + cout << "Creating new mesh object..." << endl; + newGeom.type = MESH; + } } //link material @@ -64,7 +71,8 @@ int Scene::loadGeom(string objectid) { //load transformations utilityCore::safeGetline(fp_in, line); - while (!line.empty() && fp_in.good()) { + int lineCount = 0; + while (!line.empty() && fp_in.good() && lineCount++ < 3) { vector tokens = utilityCore::tokenizeString(line); //load tranformations @@ -84,6 +92,27 @@ int Scene::loadGeom(string objectid) { newGeom.inverseTransform = glm::inverse(newGeom.transform); newGeom.invTranspose = glm::inverseTranspose(newGeom.transform); + if (newGeom.type == MESH) { + if (!line.empty() && fp_in.good()) { + vector tokens = utilityCore::tokenizeString(line); + if (strcmp(tokens[0].c_str(), "FILE") == 0) { + std::string fileName = "../scenes/" + tokens[1]; + loadObjFile(fileName, newGeom); + } + } + } + else { + if (!line.empty() && fp_in.good()) { + vector tokens = utilityCore::tokenizeString(line); + if (strcmp(tokens[0].c_str(), "MOTH") == 0) { + std::cout << "motion" << std::endl; + newGeom.hasMotion = 1; + newGeom.motion = glm::vec3(atof(tokens[1].c_str()), atof(tokens[2].c_str()), atof(tokens[3].c_str())); + //std::cout << newGeom.motion[0] << newGeom.motion[1] << newGeom.motion[2] << std::endl; + } + } + } + geoms.push_back(newGeom); return 1; } @@ -111,7 +140,7 @@ int Scene::loadCamera() { state.traceDepth = atoi(tokens[1].c_str()); } else if (strcmp(tokens[0].c_str(), "FILE") == 0) { state.imageName = tokens[1]; - } + } } string line; @@ -125,6 +154,12 @@ int Scene::loadCamera() { } else if (strcmp(tokens[0].c_str(), "UP") == 0) { camera.up = glm::vec3(atof(tokens[1].c_str()), atof(tokens[2].c_str()), atof(tokens[3].c_str())); } + else if (strcmp(tokens[0].c_str(), "FOCALDIST") == 0) { + camera.focalLength = atof(tokens[1].c_str()); + } + else if (strcmp(tokens[0].c_str(), "LENSRAD") == 0) { + camera.lensRadius = atof(tokens[1].c_str()); + } utilityCore::safeGetline(fp_in, line); } @@ -186,3 +221,54 @@ int Scene::loadMaterial(string materialid) { return 1; } } + +int Scene::loadObjFile(std::string fileName, Geom& geom) { + tinyobj::attrib_t attrib; + std::vector shapes; + std::vector materials; + std::string err; + std::cout << fileName << std::endl; + bool ret = tinyobj::LoadObj(&attrib, &shapes, &materials, &err, fileName.c_str()); + + if (!err.empty()) { // `err` may contain warning message. + std::cerr << err << std::endl; + } + + if (!ret) { + exit(1); + } + geom.triStartIndex = tris.size(); + for (size_t s = 0; s < shapes.size(); s++) { + // Loop over faces(polygon) + size_t index_offset = 0; + for (size_t f = 0; f < shapes[s].mesh.num_face_vertices.size(); f++) { + int fv = shapes[s].mesh.num_face_vertices[f]; + Triangle t; + // Loop over vertices in the face. + for (size_t v = 0; v < fv; v++) { + // access to vertex + tinyobj::index_t idx = shapes[s].mesh.indices[index_offset + v]; + tinyobj::real_t vx = attrib.vertices[3 * idx.vertex_index + 0]; + tinyobj::real_t vy = attrib.vertices[3 * idx.vertex_index + 1]; + tinyobj::real_t vz = attrib.vertices[3 * idx.vertex_index + 2]; + tinyobj::real_t nx = attrib.normals[3 * idx.normal_index + 0]; + tinyobj::real_t ny = attrib.normals[3 * idx.normal_index + 1]; + tinyobj::real_t nz = attrib.normals[3 * idx.normal_index + 2]; + tinyobj::real_t tx = attrib.texcoords[2 * idx.texcoord_index + 0]; + tinyobj::real_t ty = attrib.texcoords[2 * idx.texcoord_index + 1]; + // Optional: vertex colors + // tinyobj::real_t red = attrib.colors[3*idx.vertex_index+0]; + // tinyobj::real_t green = attrib.colors[3*idx.vertex_index+1]; + // tinyobj::real_t blue = attrib.colors[3*idx.vertex_index+2]; + t.vertices[v] = glm::vec3(vx, vy, vz); + t.normals[v] = glm::vec3(nx, ny, nz); + t.uvs[v] = glm::vec2(tx, ty); + } + index_offset += fv; + tris.push_back(t); + } + } + + geom.triEndIndex = tris.size() - 1; + return 1; +} diff --git a/src/scene.h b/src/scene.h index f29a917..2b5d249 100644 --- a/src/scene.h +++ b/src/scene.h @@ -16,6 +16,7 @@ class Scene { int loadMaterial(string materialid); int loadGeom(string objectid); int loadCamera(); + int loadObjFile(std::string fileName, Geom& geom); public: Scene(string filename); ~Scene(); @@ -23,4 +24,6 @@ class Scene { std::vector geoms; std::vector materials; RenderState state; + + std::vector tris; }; diff --git a/src/sceneStructs.h b/src/sceneStructs.h index b38b820..0a907c5 100644 --- a/src/sceneStructs.h +++ b/src/sceneStructs.h @@ -10,6 +10,7 @@ enum GeomType { SPHERE, CUBE, + MESH, }; struct Ray { @@ -26,6 +27,14 @@ struct Geom { glm::mat4 transform; glm::mat4 inverseTransform; glm::mat4 invTranspose; + + // mesh + int triStartIndex; + int triEndIndex; + + // motion + int hasMotion = 0; + glm::vec3 motion; }; struct Material { @@ -49,6 +58,8 @@ struct Camera { glm::vec3 right; glm::vec2 fov; glm::vec2 pixelLength; + float focalLength; + float lensRadius; }; struct RenderState { @@ -74,3 +85,9 @@ struct ShadeableIntersection { glm::vec3 surfaceNormal; int materialId; }; + +struct Triangle { + glm::vec3 vertices[3]; + glm::vec3 normals[3]; + glm::vec2 uvs[3]; +}; \ No newline at end of file diff --git a/src/tiny_obj_loader.cc b/src/tiny_obj_loader.cc new file mode 100644 index 0000000..e57d044 --- /dev/null +++ b/src/tiny_obj_loader.cc @@ -0,0 +1,2 @@ +#define TINYOBJLOADER_IMPLEMENTATION +#include "tiny_obj_loader.h" diff --git a/src/tiny_obj_loader.h b/src/tiny_obj_loader.h new file mode 100644 index 0000000..f627e94 --- /dev/null +++ b/src/tiny_obj_loader.h @@ -0,0 +1,2518 @@ +/* +The MIT License (MIT) + +Copyright (c) 2012-2018 Syoyo Fujita and many contributors. + +Permission is hereby granted, free of charge, to any person obtaining a copy +of this software and associated documentation files (the "Software"), to deal +in the Software without restriction, including without limitation the rights +to use, copy, modify, merge, publish, distribute, sublicense, and/or sell +copies of the Software, and to permit persons to whom the Software is +furnished to do so, subject to the following conditions: + +The above copyright notice and this permission notice shall be included in +all copies or substantial portions of the Software. + +THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR +IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, +FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE +AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER +LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, +OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN +THE SOFTWARE. +*/ + +// +// version 1.2.3 : Added color space extension('-colorspace') to tex opts. +// version 1.2.2 : Parse multiple group names. +// version 1.2.1 : Added initial support for line('l') primitive(PR #178) +// version 1.2.0 : Hardened implementation(#175) +// version 1.1.1 : Support smoothing groups(#162) +// version 1.1.0 : Support parsing vertex color(#144) +// version 1.0.8 : Fix parsing `g` tag just after `usemtl`(#138) +// version 1.0.7 : Support multiple tex options(#126) +// version 1.0.6 : Add TINYOBJLOADER_USE_DOUBLE option(#124) +// version 1.0.5 : Ignore `Tr` when `d` exists in MTL(#43) +// version 1.0.4 : Support multiple filenames for 'mtllib'(#112) +// version 1.0.3 : Support parsing texture options(#85) +// version 1.0.2 : Improve parsing speed by about a factor of 2 for large +// files(#105) +// version 1.0.1 : Fixes a shape is lost if obj ends with a 'usemtl'(#104) +// version 1.0.0 : Change data structure. Change license from BSD to MIT. +// + +// +// Use this in *one* .cc +// #define TINYOBJLOADER_IMPLEMENTATION +// #include "tiny_obj_loader.h" +// + +#ifndef TINY_OBJ_LOADER_H_ +#define TINY_OBJ_LOADER_H_ + +#include +#include +#include + +namespace tinyobj { + +#ifdef __clang__ +#pragma clang diagnostic push +#if __has_warning("-Wzero-as-null-pointer-constant") +#pragma clang diagnostic ignored "-Wzero-as-null-pointer-constant" +#endif + +#pragma clang diagnostic ignored "-Wpadded" + +#endif + +// https://en.wikipedia.org/wiki/Wavefront_.obj_file says ... +// +// -blendu on | off # set horizontal texture blending +// (default on) +// -blendv on | off # set vertical texture blending +// (default on) +// -boost real_value # boost mip-map sharpness +// -mm base_value gain_value # modify texture map values (default +// 0 1) +// # base_value = brightness, +// gain_value = contrast +// -o u [v [w]] # Origin offset (default +// 0 0 0) +// -s u [v [w]] # Scale (default +// 1 1 1) +// -t u [v [w]] # Turbulence (default +// 0 0 0) +// -texres resolution # texture resolution to create +// -clamp on | off # only render texels in the clamped +// 0-1 range (default off) +// # When unclamped, textures are +// repeated across a surface, +// # when clamped, only texels which +// fall within the 0-1 +// # range are rendered. +// -bm mult_value # bump multiplier (for bump maps +// only) +// +// -imfchan r | g | b | m | l | z # specifies which channel of the file +// is used to +// # create a scalar or bump texture. +// r:red, g:green, +// # b:blue, m:matte, l:luminance, +// z:z-depth.. +// # (the default for bump is 'l' and +// for decal is 'm') +// bump -imfchan r bumpmap.tga # says to use the red channel of +// bumpmap.tga as the bumpmap +// +// For reflection maps... +// +// -type sphere # specifies a sphere for a "refl" +// reflection map +// -type cube_top | cube_bottom | # when using a cube map, the texture +// file for each +// cube_front | cube_back | # side of the cube is specified +// separately +// cube_left | cube_right +// +// TinyObjLoader extension. +// +// -colorspace SPACE # Color space of the texture. e.g. 'sRGB` or 'linear' +// + +#ifdef TINYOBJLOADER_USE_DOUBLE +//#pragma message "using double" +typedef double real_t; +#else +//#pragma message "using float" +typedef float real_t; +#endif + +typedef enum { + TEXTURE_TYPE_NONE, // default + TEXTURE_TYPE_SPHERE, + TEXTURE_TYPE_CUBE_TOP, + TEXTURE_TYPE_CUBE_BOTTOM, + TEXTURE_TYPE_CUBE_FRONT, + TEXTURE_TYPE_CUBE_BACK, + TEXTURE_TYPE_CUBE_LEFT, + TEXTURE_TYPE_CUBE_RIGHT +} texture_type_t; + +typedef struct { + texture_type_t type; // -type (default TEXTURE_TYPE_NONE) + real_t sharpness; // -boost (default 1.0?) + real_t brightness; // base_value in -mm option (default 0) + real_t contrast; // gain_value in -mm option (default 1) + real_t origin_offset[3]; // -o u [v [w]] (default 0 0 0) + real_t scale[3]; // -s u [v [w]] (default 1 1 1) + real_t turbulence[3]; // -t u [v [w]] (default 0 0 0) + // int texture_resolution; // -texres resolution (default = ?) TODO + bool clamp; // -clamp (default false) + char imfchan; // -imfchan (the default for bump is 'l' and for decal is 'm') + bool blendu; // -blendu (default on) + bool blendv; // -blendv (default on) + real_t bump_multiplier; // -bm (for bump maps only, default 1.0) + + // extension + std::string colorspace; // Explicitly specify color space of stored value. Usually `sRGB` or `linear` (default empty). +} texture_option_t; + +typedef struct { + std::string name; + + real_t ambient[3]; + real_t diffuse[3]; + real_t specular[3]; + real_t transmittance[3]; + real_t emission[3]; + real_t shininess; + real_t ior; // index of refraction + real_t dissolve; // 1 == opaque; 0 == fully transparent + // illumination model (see http://www.fileformat.info/format/material/) + int illum; + + int dummy; // Suppress padding warning. + + std::string ambient_texname; // map_Ka + std::string diffuse_texname; // map_Kd + std::string specular_texname; // map_Ks + std::string specular_highlight_texname; // map_Ns + std::string bump_texname; // map_bump, map_Bump, bump + std::string displacement_texname; // disp + std::string alpha_texname; // map_d + std::string reflection_texname; // refl + + texture_option_t ambient_texopt; + texture_option_t diffuse_texopt; + texture_option_t specular_texopt; + texture_option_t specular_highlight_texopt; + texture_option_t bump_texopt; + texture_option_t displacement_texopt; + texture_option_t alpha_texopt; + texture_option_t reflection_texopt; + + // PBR extension + // http://exocortex.com/blog/extending_wavefront_mtl_to_support_pbr + real_t roughness; // [0, 1] default 0 + real_t metallic; // [0, 1] default 0 + real_t sheen; // [0, 1] default 0 + real_t clearcoat_thickness; // [0, 1] default 0 + real_t clearcoat_roughness; // [0, 1] default 0 + real_t anisotropy; // aniso. [0, 1] default 0 + real_t anisotropy_rotation; // anisor. [0, 1] default 0 + real_t pad0; + std::string roughness_texname; // map_Pr + std::string metallic_texname; // map_Pm + std::string sheen_texname; // map_Ps + std::string emissive_texname; // map_Ke + std::string normal_texname; // norm. For normal mapping. + + texture_option_t roughness_texopt; + texture_option_t metallic_texopt; + texture_option_t sheen_texopt; + texture_option_t emissive_texopt; + texture_option_t normal_texopt; + + int pad2; + + std::map unknown_parameter; +} material_t; + +typedef struct { + std::string name; + + std::vector intValues; + std::vector floatValues; + std::vector stringValues; +} tag_t; + +// Index struct to support different indices for vtx/normal/texcoord. +// -1 means not used. +typedef struct { + int vertex_index; + int normal_index; + int texcoord_index; +} index_t; + +typedef struct { + std::vector indices; + std::vector num_face_vertices; // The number of vertices per + // face. 3 = polygon, 4 = quad, + // ... Up to 255. + std::vector material_ids; // per-face material ID + std::vector smoothing_group_ids; // per-face smoothing group + // ID(0 = off. positive value + // = group id) + std::vector tags; // SubD tag +} mesh_t; + +typedef struct { + std::vector indices; // pairs of indices for lines +} path_t; + +typedef struct { + std::string name; + mesh_t mesh; + path_t path; +} shape_t; + +// Vertex attributes +typedef struct { + std::vector vertices; // 'v' + std::vector normals; // 'vn' + std::vector texcoords; // 'vt' + std::vector colors; // extension: vertex colors +} attrib_t; + +typedef struct callback_t_ { + // W is optional and set to 1 if there is no `w` item in `v` line + void (*vertex_cb)(void *user_data, real_t x, real_t y, real_t z, real_t w); + void (*normal_cb)(void *user_data, real_t x, real_t y, real_t z); + + // y and z are optional and set to 0 if there is no `y` and/or `z` item(s) in + // `vt` line. + void (*texcoord_cb)(void *user_data, real_t x, real_t y, real_t z); + + // called per 'f' line. num_indices is the number of face indices(e.g. 3 for + // triangle, 4 for quad) + // 0 will be passed for undefined index in index_t members. + void (*index_cb)(void *user_data, index_t *indices, int num_indices); + // `name` material name, `material_id` = the array index of material_t[]. -1 + // if + // a material not found in .mtl + void (*usemtl_cb)(void *user_data, const char *name, int material_id); + // `materials` = parsed material data. + void (*mtllib_cb)(void *user_data, const material_t *materials, + int num_materials); + // There may be multiple group names + void (*group_cb)(void *user_data, const char **names, int num_names); + void (*object_cb)(void *user_data, const char *name); + + callback_t_() + : vertex_cb(NULL), + normal_cb(NULL), + texcoord_cb(NULL), + index_cb(NULL), + usemtl_cb(NULL), + mtllib_cb(NULL), + group_cb(NULL), + object_cb(NULL) {} +} callback_t; + +class MaterialReader { + public: + MaterialReader() {} + virtual ~MaterialReader(); + + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) = 0; +}; + +class MaterialFileReader : public MaterialReader { + public: + explicit MaterialFileReader(const std::string &mtl_basedir) + : m_mtlBaseDir(mtl_basedir) {} + virtual ~MaterialFileReader() {} + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, std::string *err); + + private: + std::string m_mtlBaseDir; +}; + +class MaterialStreamReader : public MaterialReader { + public: + explicit MaterialStreamReader(std::istream &inStream) + : m_inStream(inStream) {} + virtual ~MaterialStreamReader() {} + virtual bool operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, std::string *err); + + private: + std::istream &m_inStream; +}; + +/// Loads .obj from a file. +/// 'attrib', 'shapes' and 'materials' will be filled with parsed shape data +/// 'shapes' will be filled with parsed shape data +/// Returns true when loading .obj become success. +/// Returns warning and error message into `err` +/// 'mtl_basedir' is optional, and used for base directory for .mtl file. +/// In default(`NULL'), .mtl file is searched from an application's working +/// directory. +/// 'triangulate' is optional, and used whether triangulate polygon face in .obj +/// or not. +/// Option 'default_vcols_fallback' specifies whether vertex colors should +/// always be defined, even if no colors are given (fallback to white). +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + const char *filename, const char *mtl_basedir = NULL, + bool triangulate = true, bool default_vcols_fallback = true); + +/// Loads .obj from a file with custom user callback. +/// .mtl is loaded as usual and parsed material_t data will be passed to +/// `callback.mtllib_cb`. +/// Returns true when loading .obj/.mtl become success. +/// Returns warning and error message into `err` +/// See `examples/callback_api/` for how to use this function. +bool LoadObjWithCallback(std::istream &inStream, const callback_t &callback, + void *user_data = NULL, + MaterialReader *readMatFn = NULL, + std::string *err = NULL); + +/// Loads object from a std::istream, uses GetMtlIStreamFn to retrieve +/// std::istream for materials. +/// Returns true when loading .obj become success. +/// Returns warning and error message into `err` +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + std::istream *inStream, MaterialReader *readMatFn = NULL, + bool triangulate = true, bool default_vcols_fallback = true); + +/// Loads materials into std::map +void LoadMtl(std::map *material_map, + std::vector *materials, std::istream *inStream, + std::string *warning); + +} // namespace tinyobj + +#endif // TINY_OBJ_LOADER_H_ + +#ifdef TINYOBJLOADER_IMPLEMENTATION +#include +#include +#include +#include +#include +#include +#include +#include + +#include +#include + +namespace tinyobj { + +MaterialReader::~MaterialReader() {} + +struct vertex_index_t { + int v_idx, vt_idx, vn_idx; + vertex_index_t() : v_idx(-1), vt_idx(-1), vn_idx(-1) {} + explicit vertex_index_t(int idx) : v_idx(idx), vt_idx(idx), vn_idx(idx) {} + vertex_index_t(int vidx, int vtidx, int vnidx) + : v_idx(vidx), vt_idx(vtidx), vn_idx(vnidx) {} +}; + +// Internal data structure for face representation +// index + smoothing group. +struct face_t { + unsigned int + smoothing_group_id; // smoothing group id. 0 = smoothing groupd is off. + int pad_; + std::vector vertex_indices; // face vertex indices. + + face_t() : smoothing_group_id(0) {} +}; + +struct line_t { + int idx0; + int idx1; +}; + +struct tag_sizes { + tag_sizes() : num_ints(0), num_reals(0), num_strings(0) {} + int num_ints; + int num_reals; + int num_strings; +}; + +struct obj_shape { + std::vector v; + std::vector vn; + std::vector vt; +}; + +// See +// http://stackoverflow.com/questions/6089231/getting-std-ifstream-to-handle-lf-cr-and-crlf +static std::istream &safeGetline(std::istream &is, std::string &t) { + t.clear(); + + // The characters in the stream are read one-by-one using a std::streambuf. + // That is faster than reading them one-by-one using the std::istream. + // Code that uses streambuf this way must be guarded by a sentry object. + // The sentry object performs various tasks, + // such as thread synchronization and updating the stream state. + + std::istream::sentry se(is, true); + std::streambuf *sb = is.rdbuf(); + + if (se) { + for (;;) { + int c = sb->sbumpc(); + switch (c) { + case '\n': + return is; + case '\r': + if (sb->sgetc() == '\n') sb->sbumpc(); + return is; + case EOF: + // Also handle the case when the last line has no line ending + if (t.empty()) is.setstate(std::ios::eofbit); + return is; + default: + t += static_cast(c); + } + } + } + + return is; +} + +#define IS_SPACE(x) (((x) == ' ') || ((x) == '\t')) +#define IS_DIGIT(x) \ + (static_cast((x) - '0') < static_cast(10)) +#define IS_NEW_LINE(x) (((x) == '\r') || ((x) == '\n') || ((x) == '\0')) + +// Make index zero-base, and also support relative index. +static inline bool fixIndex(int idx, int n, int *ret) { + if (!ret) { + return false; + } + + if (idx > 0) { + (*ret) = idx - 1; + return true; + } + + if (idx == 0) { + // zero is not allowed according to the spec. + return false; + } + + if (idx < 0) { + (*ret) = n + idx; // negative value = relative + return true; + } + + return false; // never reach here. +} + +static inline std::string parseString(const char **token) { + std::string s; + (*token) += strspn((*token), " \t"); + size_t e = strcspn((*token), " \t\r"); + s = std::string((*token), &(*token)[e]); + (*token) += e; + return s; +} + +static inline int parseInt(const char **token) { + (*token) += strspn((*token), " \t"); + int i = atoi((*token)); + (*token) += strcspn((*token), " \t\r"); + return i; +} + +// Tries to parse a floating point number located at s. +// +// s_end should be a location in the string where reading should absolutely +// stop. For example at the end of the string, to prevent buffer overflows. +// +// Parses the following EBNF grammar: +// sign = "+" | "-" ; +// END = ? anything not in digit ? +// digit = "0" | "1" | "2" | "3" | "4" | "5" | "6" | "7" | "8" | "9" ; +// integer = [sign] , digit , {digit} ; +// decimal = integer , ["." , integer] ; +// float = ( decimal , END ) | ( decimal , ("E" | "e") , integer , END ) ; +// +// Valid strings are for example: +// -0 +3.1417e+2 -0.0E-3 1.0324 -1.41 11e2 +// +// If the parsing is a success, result is set to the parsed value and true +// is returned. +// +// The function is greedy and will parse until any of the following happens: +// - a non-conforming character is encountered. +// - s_end is reached. +// +// The following situations triggers a failure: +// - s >= s_end. +// - parse failure. +// +static bool tryParseDouble(const char *s, const char *s_end, double *result) { + if (s >= s_end) { + return false; + } + + double mantissa = 0.0; + // This exponent is base 2 rather than 10. + // However the exponent we parse is supposed to be one of ten, + // thus we must take care to convert the exponent/and or the + // mantissa to a * 2^E, where a is the mantissa and E is the + // exponent. + // To get the final double we will use ldexp, it requires the + // exponent to be in base 2. + int exponent = 0; + + // NOTE: THESE MUST BE DECLARED HERE SINCE WE ARE NOT ALLOWED + // TO JUMP OVER DEFINITIONS. + char sign = '+'; + char exp_sign = '+'; + char const *curr = s; + + // How many characters were read in a loop. + int read = 0; + // Tells whether a loop terminated due to reaching s_end. + bool end_not_reached = false; + + /* + BEGIN PARSING. + */ + + // Find out what sign we've got. + if (*curr == '+' || *curr == '-') { + sign = *curr; + curr++; + } else if (IS_DIGIT(*curr)) { /* Pass through. */ + } else { + goto fail; + } + + // Read the integer part. + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + mantissa *= 10; + mantissa += static_cast(*curr - 0x30); + curr++; + read++; + end_not_reached = (curr != s_end); + } + + // We must make sure we actually got something. + if (read == 0) goto fail; + // We allow numbers of form "#", "###" etc. + if (!end_not_reached) goto assemble; + + // Read the decimal part. + if (*curr == '.') { + curr++; + read = 1; + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + static const double pow_lut[] = { + 1.0, 0.1, 0.01, 0.001, 0.0001, 0.00001, 0.000001, 0.0000001, + }; + const int lut_entries = sizeof pow_lut / sizeof pow_lut[0]; + + // NOTE: Don't use powf here, it will absolutely murder precision. + mantissa += static_cast(*curr - 0x30) * + (read < lut_entries ? pow_lut[read] : std::pow(10.0, -read)); + read++; + curr++; + end_not_reached = (curr != s_end); + } + } else if (*curr == 'e' || *curr == 'E') { + } else { + goto assemble; + } + + if (!end_not_reached) goto assemble; + + // Read the exponent part. + if (*curr == 'e' || *curr == 'E') { + curr++; + // Figure out if a sign is present and if it is. + end_not_reached = (curr != s_end); + if (end_not_reached && (*curr == '+' || *curr == '-')) { + exp_sign = *curr; + curr++; + } else if (IS_DIGIT(*curr)) { /* Pass through. */ + } else { + // Empty E is not allowed. + goto fail; + } + + read = 0; + end_not_reached = (curr != s_end); + while (end_not_reached && IS_DIGIT(*curr)) { + exponent *= 10; + exponent += static_cast(*curr - 0x30); + curr++; + read++; + end_not_reached = (curr != s_end); + } + exponent *= (exp_sign == '+' ? 1 : -1); + if (read == 0) goto fail; + } + +assemble: + *result = (sign == '+' ? 1 : -1) * + (exponent ? std::ldexp(mantissa * std::pow(5.0, exponent), exponent) + : mantissa); + return true; +fail: + return false; +} + +static inline real_t parseReal(const char **token, double default_value = 0.0) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + double val = default_value; + tryParseDouble((*token), end, &val); + real_t f = static_cast(val); + (*token) = end; + return f; +} + +static inline bool parseReal(const char **token, real_t *out) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + double val; + bool ret = tryParseDouble((*token), end, &val); + if (ret) { + real_t f = static_cast(val); + (*out) = f; + } + (*token) = end; + return ret; +} + +static inline void parseReal2(real_t *x, real_t *y, const char **token, + const double default_x = 0.0, + const double default_y = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); +} + +static inline void parseReal3(real_t *x, real_t *y, real_t *z, + const char **token, const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); +} + +static inline void parseV(real_t *x, real_t *y, real_t *z, real_t *w, + const char **token, const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0, + const double default_w = 1.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); + (*w) = parseReal(token, default_w); +} + +// Extension: parse vertex with colors(6 items) +static inline bool parseVertexWithColor(real_t *x, real_t *y, real_t *z, + real_t *r, real_t *g, real_t *b, + const char **token, + const double default_x = 0.0, + const double default_y = 0.0, + const double default_z = 0.0) { + (*x) = parseReal(token, default_x); + (*y) = parseReal(token, default_y); + (*z) = parseReal(token, default_z); + + const bool found_color = parseReal(token, r) && parseReal(token, g) && parseReal(token, b); + + if (!found_color) { + (*r) = (*g) = (*b) = 1.0; + } + + return found_color; +} + +static inline bool parseOnOff(const char **token, bool default_value = true) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + + bool ret = default_value; + if ((0 == strncmp((*token), "on", 2))) { + ret = true; + } else if ((0 == strncmp((*token), "off", 3))) { + ret = false; + } + + (*token) = end; + return ret; +} + +static inline texture_type_t parseTextureType( + const char **token, texture_type_t default_value = TEXTURE_TYPE_NONE) { + (*token) += strspn((*token), " \t"); + const char *end = (*token) + strcspn((*token), " \t\r"); + texture_type_t ty = default_value; + + if ((0 == strncmp((*token), "cube_top", strlen("cube_top")))) { + ty = TEXTURE_TYPE_CUBE_TOP; + } else if ((0 == strncmp((*token), "cube_bottom", strlen("cube_bottom")))) { + ty = TEXTURE_TYPE_CUBE_BOTTOM; + } else if ((0 == strncmp((*token), "cube_left", strlen("cube_left")))) { + ty = TEXTURE_TYPE_CUBE_LEFT; + } else if ((0 == strncmp((*token), "cube_right", strlen("cube_right")))) { + ty = TEXTURE_TYPE_CUBE_RIGHT; + } else if ((0 == strncmp((*token), "cube_front", strlen("cube_front")))) { + ty = TEXTURE_TYPE_CUBE_FRONT; + } else if ((0 == strncmp((*token), "cube_back", strlen("cube_back")))) { + ty = TEXTURE_TYPE_CUBE_BACK; + } else if ((0 == strncmp((*token), "sphere", strlen("sphere")))) { + ty = TEXTURE_TYPE_SPHERE; + } + + (*token) = end; + return ty; +} + +static tag_sizes parseTagTriple(const char **token) { + tag_sizes ts; + + (*token) += strspn((*token), " \t"); + ts.num_ints = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return ts; + } + + (*token)++; // Skip '/' + + (*token) += strspn((*token), " \t"); + ts.num_reals = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return ts; + } + (*token)++; // Skip '/' + + ts.num_strings = parseInt(token); + + return ts; +} + +// Parse triples with index offsets: i, i/j/k, i//k, i/j +static bool parseTriple(const char **token, int vsize, int vnsize, int vtsize, + vertex_index_t *ret) { + if (!ret) { + return false; + } + + vertex_index_t vi(-1); + + if (!fixIndex(atoi((*token)), vsize, &(vi.v_idx))) { + return false; + } + + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + (*ret) = vi; + return true; + } + (*token)++; + + // i//k + if ((*token)[0] == '/') { + (*token)++; + if (!fixIndex(atoi((*token)), vnsize, &(vi.vn_idx))) { + return false; + } + (*token) += strcspn((*token), "/ \t\r"); + (*ret) = vi; + return true; + } + + // i/j/k or i/j + if (!fixIndex(atoi((*token)), vtsize, &(vi.vt_idx))) { + return false; + } + + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + (*ret) = vi; + return true; + } + + // i/j/k + (*token)++; // skip '/' + if (!fixIndex(atoi((*token)), vnsize, &(vi.vn_idx))) { + return false; + } + (*token) += strcspn((*token), "/ \t\r"); + + (*ret) = vi; + + return true; +} + +// Parse raw triples: i, i/j/k, i//k, i/j +static vertex_index_t parseRawTriple(const char **token) { + vertex_index_t vi(static_cast(0)); // 0 is an invalid index in OBJ + + vi.v_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return vi; + } + (*token)++; + + // i//k + if ((*token)[0] == '/') { + (*token)++; + vi.vn_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + return vi; + } + + // i/j/k or i/j + vi.vt_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + if ((*token)[0] != '/') { + return vi; + } + + // i/j/k + (*token)++; // skip '/' + vi.vn_idx = atoi((*token)); + (*token) += strcspn((*token), "/ \t\r"); + return vi; +} + +static bool ParseTextureNameAndOption(std::string *texname, + texture_option_t *texopt, + const char *linebuf, const bool is_bump) { + // @todo { write more robust lexer and parser. } + bool found_texname = false; + std::string texture_name; + + // Fill with default value for texopt. + if (is_bump) { + texopt->imfchan = 'l'; + } else { + texopt->imfchan = 'm'; + } + texopt->bump_multiplier = static_cast(1.0); + texopt->clamp = false; + texopt->blendu = true; + texopt->blendv = true; + texopt->sharpness = static_cast(1.0); + texopt->brightness = static_cast(0.0); + texopt->contrast = static_cast(1.0); + texopt->origin_offset[0] = static_cast(0.0); + texopt->origin_offset[1] = static_cast(0.0); + texopt->origin_offset[2] = static_cast(0.0); + texopt->scale[0] = static_cast(1.0); + texopt->scale[1] = static_cast(1.0); + texopt->scale[2] = static_cast(1.0); + texopt->turbulence[0] = static_cast(0.0); + texopt->turbulence[1] = static_cast(0.0); + texopt->turbulence[2] = static_cast(0.0); + texopt->type = TEXTURE_TYPE_NONE; + + const char *token = linebuf; // Assume line ends with NULL + + while (!IS_NEW_LINE((*token))) { + token += strspn(token, " \t"); // skip space + if ((0 == strncmp(token, "-blendu", 7)) && IS_SPACE((token[7]))) { + token += 8; + texopt->blendu = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-blendv", 7)) && IS_SPACE((token[7]))) { + token += 8; + texopt->blendv = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-clamp", 6)) && IS_SPACE((token[6]))) { + token += 7; + texopt->clamp = parseOnOff(&token, /* default */ true); + } else if ((0 == strncmp(token, "-boost", 6)) && IS_SPACE((token[6]))) { + token += 7; + texopt->sharpness = parseReal(&token, 1.0); + } else if ((0 == strncmp(token, "-bm", 3)) && IS_SPACE((token[3]))) { + token += 4; + texopt->bump_multiplier = parseReal(&token, 1.0); + } else if ((0 == strncmp(token, "-o", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->origin_offset[0]), &(texopt->origin_offset[1]), + &(texopt->origin_offset[2]), &token); + } else if ((0 == strncmp(token, "-s", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->scale[0]), &(texopt->scale[1]), &(texopt->scale[2]), + &token, 1.0, 1.0, 1.0); + } else if ((0 == strncmp(token, "-t", 2)) && IS_SPACE((token[2]))) { + token += 3; + parseReal3(&(texopt->turbulence[0]), &(texopt->turbulence[1]), + &(texopt->turbulence[2]), &token); + } else if ((0 == strncmp(token, "-type", 5)) && IS_SPACE((token[5]))) { + token += 5; + texopt->type = parseTextureType((&token), TEXTURE_TYPE_NONE); + } else if ((0 == strncmp(token, "-imfchan", 8)) && IS_SPACE((token[8]))) { + token += 9; + token += strspn(token, " \t"); + const char *end = token + strcspn(token, " \t\r"); + if ((end - token) == 1) { // Assume one char for -imfchan + texopt->imfchan = (*token); + } + token = end; + } else if ((0 == strncmp(token, "-mm", 3)) && IS_SPACE((token[3]))) { + token += 4; + parseReal2(&(texopt->brightness), &(texopt->contrast), &token, 0.0, 1.0); + } else if ((0 == strncmp(token, "-colorspace", 11)) && IS_SPACE((token[11]))) { + token += 12; + texopt->colorspace = parseString(&token); + } else { + // Assume texture filename +#if 0 + size_t len = strcspn(token, " \t\r"); // untile next space + texture_name = std::string(token, token + len); + token += len; + + token += strspn(token, " \t"); // skip space +#else + // Read filename until line end to parse filename containing whitespace + // TODO(syoyo): Support parsing texture option flag after the filename. + texture_name = std::string(token); + token += texture_name.length(); +#endif + + found_texname = true; + } + } + + if (found_texname) { + (*texname) = texture_name; + return true; + } else { + return false; + } +} + +static void InitMaterial(material_t *material) { + material->name = ""; + material->ambient_texname = ""; + material->diffuse_texname = ""; + material->specular_texname = ""; + material->specular_highlight_texname = ""; + material->bump_texname = ""; + material->displacement_texname = ""; + material->reflection_texname = ""; + material->alpha_texname = ""; + for (int i = 0; i < 3; i++) { + material->ambient[i] = static_cast(0.0); + material->diffuse[i] = static_cast(0.0); + material->specular[i] = static_cast(0.0); + material->transmittance[i] = static_cast(0.0); + material->emission[i] = static_cast(0.0); + } + material->illum = 0; + material->dissolve = static_cast(1.0); + material->shininess = static_cast(1.0); + material->ior = static_cast(1.0); + + material->roughness = static_cast(0.0); + material->metallic = static_cast(0.0); + material->sheen = static_cast(0.0); + material->clearcoat_thickness = static_cast(0.0); + material->clearcoat_roughness = static_cast(0.0); + material->anisotropy_rotation = static_cast(0.0); + material->anisotropy = static_cast(0.0); + material->roughness_texname = ""; + material->metallic_texname = ""; + material->sheen_texname = ""; + material->emissive_texname = ""; + material->normal_texname = ""; + + material->unknown_parameter.clear(); +} + +// code from https://wrf.ecse.rpi.edu//Research/Short_Notes/pnpoly.html +template +static int pnpoly(int nvert, T *vertx, T *verty, T testx, T testy) { + int i, j, c = 0; + for (i = 0, j = nvert - 1; i < nvert; j = i++) { + if (((verty[i] > testy) != (verty[j] > testy)) && + (testx < + (vertx[j] - vertx[i]) * (testy - verty[i]) / (verty[j] - verty[i]) + + vertx[i])) + c = !c; + } + return c; +} + +// TODO(syoyo): refactor function. +static bool exportGroupsToShape(shape_t *shape, + const std::vector &faceGroup, + std::vector &lineGroup, + const std::vector &tags, + const int material_id, const std::string &name, + bool triangulate, + const std::vector &v) { + if (faceGroup.empty() && lineGroup.empty()) { + return false; + } + + if (!faceGroup.empty()) { + // Flatten vertices and indices + for (size_t i = 0; i < faceGroup.size(); i++) { + const face_t &face = faceGroup[i]; + + size_t npolys = face.vertex_indices.size(); + + if (npolys < 3) { + // Face must have 3+ vertices. + continue; + } + + vertex_index_t i0 = face.vertex_indices[0]; + vertex_index_t i1(-1); + vertex_index_t i2 = face.vertex_indices[1]; + + if (triangulate) { + // find the two axes to work in + size_t axes[2] = {1, 2}; + for (size_t k = 0; k < npolys; ++k) { + i0 = face.vertex_indices[(k + 0) % npolys]; + i1 = face.vertex_indices[(k + 1) % npolys]; + i2 = face.vertex_indices[(k + 2) % npolys]; + size_t vi0 = size_t(i0.v_idx); + size_t vi1 = size_t(i1.v_idx); + size_t vi2 = size_t(i2.v_idx); + + if (((3 * vi0 + 2) >= v.size()) || ((3 * vi1 + 2) >= v.size()) || + ((3 * vi2 + 2) >= v.size())) { + // Invalid triangle. + // FIXME(syoyo): Is it ok to simply skip this invalid triangle? + continue; + } + real_t v0x = v[vi0 * 3 + 0]; + real_t v0y = v[vi0 * 3 + 1]; + real_t v0z = v[vi0 * 3 + 2]; + real_t v1x = v[vi1 * 3 + 0]; + real_t v1y = v[vi1 * 3 + 1]; + real_t v1z = v[vi1 * 3 + 2]; + real_t v2x = v[vi2 * 3 + 0]; + real_t v2y = v[vi2 * 3 + 1]; + real_t v2z = v[vi2 * 3 + 2]; + real_t e0x = v1x - v0x; + real_t e0y = v1y - v0y; + real_t e0z = v1z - v0z; + real_t e1x = v2x - v1x; + real_t e1y = v2y - v1y; + real_t e1z = v2z - v1z; + real_t cx = std::fabs(e0y * e1z - e0z * e1y); + real_t cy = std::fabs(e0z * e1x - e0x * e1z); + real_t cz = std::fabs(e0x * e1y - e0y * e1x); + const real_t epsilon = std::numeric_limits::epsilon(); + if (cx > epsilon || cy > epsilon || cz > epsilon) { + // found a corner + if (cx > cy && cx > cz) { + } else { + axes[0] = 0; + if (cz > cx && cz > cy) axes[1] = 1; + } + break; + } + } + + real_t area = 0; + for (size_t k = 0; k < npolys; ++k) { + i0 = face.vertex_indices[(k + 0) % npolys]; + i1 = face.vertex_indices[(k + 1) % npolys]; + size_t vi0 = size_t(i0.v_idx); + size_t vi1 = size_t(i1.v_idx); + if (((vi0 * 3 + axes[0]) >= v.size()) || + ((vi0 * 3 + axes[1]) >= v.size()) || + ((vi1 * 3 + axes[0]) >= v.size()) || + ((vi1 * 3 + axes[1]) >= v.size())) { + // Invalid index. + continue; + } + real_t v0x = v[vi0 * 3 + axes[0]]; + real_t v0y = v[vi0 * 3 + axes[1]]; + real_t v1x = v[vi1 * 3 + axes[0]]; + real_t v1y = v[vi1 * 3 + axes[1]]; + area += (v0x * v1y - v0y * v1x) * static_cast(0.5); + } + + int maxRounds = 10; // arbitrary max loop count to protect against + // unexpected errors + + face_t remainingFace = face; // copy + size_t guess_vert = 0; + vertex_index_t ind[3]; + real_t vx[3]; + real_t vy[3]; + while (remainingFace.vertex_indices.size() > 3 && maxRounds > 0) { + npolys = remainingFace.vertex_indices.size(); + if (guess_vert >= npolys) { + maxRounds -= 1; + guess_vert -= npolys; + } + for (size_t k = 0; k < 3; k++) { + ind[k] = remainingFace.vertex_indices[(guess_vert + k) % npolys]; + size_t vi = size_t(ind[k].v_idx); + if (((vi * 3 + axes[0]) >= v.size()) || + ((vi * 3 + axes[1]) >= v.size())) { + // ??? + vx[k] = static_cast(0.0); + vy[k] = static_cast(0.0); + } else { + vx[k] = v[vi * 3 + axes[0]]; + vy[k] = v[vi * 3 + axes[1]]; + } + } + real_t e0x = vx[1] - vx[0]; + real_t e0y = vy[1] - vy[0]; + real_t e1x = vx[2] - vx[1]; + real_t e1y = vy[2] - vy[1]; + real_t cross = e0x * e1y - e0y * e1x; + // if an internal angle + if (cross * area < static_cast(0.0)) { + guess_vert += 1; + continue; + } + + // check all other verts in case they are inside this triangle + bool overlap = false; + for (size_t otherVert = 3; otherVert < npolys; ++otherVert) { + size_t idx = (guess_vert + otherVert) % npolys; + + if (idx >= remainingFace.vertex_indices.size()) { + // ??? + continue; + } + + size_t ovi = size_t(remainingFace.vertex_indices[idx].v_idx); + + if (((ovi * 3 + axes[0]) >= v.size()) || + ((ovi * 3 + axes[1]) >= v.size())) { + // ??? + continue; + } + real_t tx = v[ovi * 3 + axes[0]]; + real_t ty = v[ovi * 3 + axes[1]]; + if (pnpoly(3, vx, vy, tx, ty)) { + overlap = true; + break; + } + } + + if (overlap) { + guess_vert += 1; + continue; + } + + // this triangle is an ear + { + index_t idx0, idx1, idx2; + idx0.vertex_index = ind[0].v_idx; + idx0.normal_index = ind[0].vn_idx; + idx0.texcoord_index = ind[0].vt_idx; + idx1.vertex_index = ind[1].v_idx; + idx1.normal_index = ind[1].vn_idx; + idx1.texcoord_index = ind[1].vt_idx; + idx2.vertex_index = ind[2].v_idx; + idx2.normal_index = ind[2].vn_idx; + idx2.texcoord_index = ind[2].vt_idx; + + shape->mesh.indices.push_back(idx0); + shape->mesh.indices.push_back(idx1); + shape->mesh.indices.push_back(idx2); + + shape->mesh.num_face_vertices.push_back(3); + shape->mesh.material_ids.push_back(material_id); + shape->mesh.smoothing_group_ids.push_back(face.smoothing_group_id); + } + + // remove v1 from the list + size_t removed_vert_index = (guess_vert + 1) % npolys; + while (removed_vert_index + 1 < npolys) { + remainingFace.vertex_indices[removed_vert_index] = + remainingFace.vertex_indices[removed_vert_index + 1]; + removed_vert_index += 1; + } + remainingFace.vertex_indices.pop_back(); + } + + if (remainingFace.vertex_indices.size() == 3) { + i0 = remainingFace.vertex_indices[0]; + i1 = remainingFace.vertex_indices[1]; + i2 = remainingFace.vertex_indices[2]; + { + index_t idx0, idx1, idx2; + idx0.vertex_index = i0.v_idx; + idx0.normal_index = i0.vn_idx; + idx0.texcoord_index = i0.vt_idx; + idx1.vertex_index = i1.v_idx; + idx1.normal_index = i1.vn_idx; + idx1.texcoord_index = i1.vt_idx; + idx2.vertex_index = i2.v_idx; + idx2.normal_index = i2.vn_idx; + idx2.texcoord_index = i2.vt_idx; + + shape->mesh.indices.push_back(idx0); + shape->mesh.indices.push_back(idx1); + shape->mesh.indices.push_back(idx2); + + shape->mesh.num_face_vertices.push_back(3); + shape->mesh.material_ids.push_back(material_id); + shape->mesh.smoothing_group_ids.push_back(face.smoothing_group_id); + } + } + } else { + for (size_t k = 0; k < npolys; k++) { + index_t idx; + idx.vertex_index = face.vertex_indices[k].v_idx; + idx.normal_index = face.vertex_indices[k].vn_idx; + idx.texcoord_index = face.vertex_indices[k].vt_idx; + shape->mesh.indices.push_back(idx); + } + + shape->mesh.num_face_vertices.push_back( + static_cast(npolys)); + shape->mesh.material_ids.push_back(material_id); // per face + shape->mesh.smoothing_group_ids.push_back( + face.smoothing_group_id); // per face + } + } + + shape->name = name; + shape->mesh.tags = tags; + } + + if (!lineGroup.empty()) { + shape->path.indices.swap(lineGroup); + } + + return true; +} + +// Split a string with specified delimiter character. +// http://stackoverflow.com/questions/236129/split-a-string-in-c +static void SplitString(const std::string &s, char delim, + std::vector &elems) { + std::stringstream ss; + ss.str(s); + std::string item; + while (std::getline(ss, item, delim)) { + elems.push_back(item); + } +} + +void LoadMtl(std::map *material_map, + std::vector *materials, std::istream *inStream, + std::string *warning) { + // Create a default material anyway. + material_t material; + InitMaterial(&material); + + // Issue 43. `d` wins against `Tr` since `Tr` is not in the MTL specification. + bool has_d = false; + bool has_tr = false; + + std::stringstream ss; + + std::string linebuf; + while (inStream->peek() != -1) { + safeGetline(*inStream, linebuf); + + // Trim trailing whitespace. + if (linebuf.size() > 0) { + linebuf = linebuf.substr(0, linebuf.find_last_not_of(" \t") + 1); + } + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // new mtl + if ((0 == strncmp(token, "newmtl", 6)) && IS_SPACE((token[6]))) { + // flush previous material. + if (!material.name.empty()) { + material_map->insert(std::pair( + material.name, static_cast(materials->size()))); + materials->push_back(material); + } + + // initial temporary material + InitMaterial(&material); + + has_d = false; + has_tr = false; + + // set new mtl name + token += 7; + { + std::stringstream sstr; + sstr << token; + material.name = sstr.str(); + } + continue; + } + + // ambient + if (token[0] == 'K' && token[1] == 'a' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.ambient[0] = r; + material.ambient[1] = g; + material.ambient[2] = b; + continue; + } + + // diffuse + if (token[0] == 'K' && token[1] == 'd' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.diffuse[0] = r; + material.diffuse[1] = g; + material.diffuse[2] = b; + continue; + } + + // specular + if (token[0] == 'K' && token[1] == 's' && IS_SPACE((token[2]))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.specular[0] = r; + material.specular[1] = g; + material.specular[2] = b; + continue; + } + + // transmittance + if ((token[0] == 'K' && token[1] == 't' && IS_SPACE((token[2]))) || + (token[0] == 'T' && token[1] == 'f' && IS_SPACE((token[2])))) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.transmittance[0] = r; + material.transmittance[1] = g; + material.transmittance[2] = b; + continue; + } + + // ior(index of refraction) + if (token[0] == 'N' && token[1] == 'i' && IS_SPACE((token[2]))) { + token += 2; + material.ior = parseReal(&token); + continue; + } + + // emission + if (token[0] == 'K' && token[1] == 'e' && IS_SPACE(token[2])) { + token += 2; + real_t r, g, b; + parseReal3(&r, &g, &b, &token); + material.emission[0] = r; + material.emission[1] = g; + material.emission[2] = b; + continue; + } + + // shininess + if (token[0] == 'N' && token[1] == 's' && IS_SPACE(token[2])) { + token += 2; + material.shininess = parseReal(&token); + continue; + } + + // illum model + if (0 == strncmp(token, "illum", 5) && IS_SPACE(token[5])) { + token += 6; + material.illum = parseInt(&token); + continue; + } + + // dissolve + if ((token[0] == 'd' && IS_SPACE(token[1]))) { + token += 1; + material.dissolve = parseReal(&token); + + if (has_tr) { + ss << "WARN: Both `d` and `Tr` parameters defined for \"" + << material.name << "\". Use the value of `d` for dissolve." + << std::endl; + } + has_d = true; + continue; + } + if (token[0] == 'T' && token[1] == 'r' && IS_SPACE(token[2])) { + token += 2; + if (has_d) { + // `d` wins. Ignore `Tr` value. + ss << "WARN: Both `d` and `Tr` parameters defined for \"" + << material.name << "\". Use the value of `d` for dissolve." + << std::endl; + } else { + // We invert value of Tr(assume Tr is in range [0, 1]) + // NOTE: Interpretation of Tr is application(exporter) dependent. For + // some application(e.g. 3ds max obj exporter), Tr = d(Issue 43) + material.dissolve = static_cast(1.0) - parseReal(&token); + } + has_tr = true; + continue; + } + + // PBR: roughness + if (token[0] == 'P' && token[1] == 'r' && IS_SPACE(token[2])) { + token += 2; + material.roughness = parseReal(&token); + continue; + } + + // PBR: metallic + if (token[0] == 'P' && token[1] == 'm' && IS_SPACE(token[2])) { + token += 2; + material.metallic = parseReal(&token); + continue; + } + + // PBR: sheen + if (token[0] == 'P' && token[1] == 's' && IS_SPACE(token[2])) { + token += 2; + material.sheen = parseReal(&token); + continue; + } + + // PBR: clearcoat thickness + if (token[0] == 'P' && token[1] == 'c' && IS_SPACE(token[2])) { + token += 2; + material.clearcoat_thickness = parseReal(&token); + continue; + } + + // PBR: clearcoat roughness + if ((0 == strncmp(token, "Pcr", 3)) && IS_SPACE(token[3])) { + token += 4; + material.clearcoat_roughness = parseReal(&token); + continue; + } + + // PBR: anisotropy + if ((0 == strncmp(token, "aniso", 5)) && IS_SPACE(token[5])) { + token += 6; + material.anisotropy = parseReal(&token); + continue; + } + + // PBR: anisotropy rotation + if ((0 == strncmp(token, "anisor", 6)) && IS_SPACE(token[6])) { + token += 7; + material.anisotropy_rotation = parseReal(&token); + continue; + } + + // ambient texture + if ((0 == strncmp(token, "map_Ka", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.ambient_texname), + &(material.ambient_texopt), token, + /* is_bump */ false); + continue; + } + + // diffuse texture + if ((0 == strncmp(token, "map_Kd", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.diffuse_texname), + &(material.diffuse_texopt), token, + /* is_bump */ false); + continue; + } + + // specular texture + if ((0 == strncmp(token, "map_Ks", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.specular_texname), + &(material.specular_texopt), token, + /* is_bump */ false); + continue; + } + + // specular highlight texture + if ((0 == strncmp(token, "map_Ns", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.specular_highlight_texname), + &(material.specular_highlight_texopt), token, + /* is_bump */ false); + continue; + } + + // bump texture + if ((0 == strncmp(token, "map_bump", 8)) && IS_SPACE(token[8])) { + token += 9; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // bump texture + if ((0 == strncmp(token, "map_Bump", 8)) && IS_SPACE(token[8])) { + token += 9; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // bump texture + if ((0 == strncmp(token, "bump", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.bump_texname), + &(material.bump_texopt), token, + /* is_bump */ true); + continue; + } + + // alpha texture + if ((0 == strncmp(token, "map_d", 5)) && IS_SPACE(token[5])) { + token += 6; + material.alpha_texname = token; + ParseTextureNameAndOption(&(material.alpha_texname), + &(material.alpha_texopt), token, + /* is_bump */ false); + continue; + } + + // displacement texture + if ((0 == strncmp(token, "disp", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.displacement_texname), + &(material.displacement_texopt), token, + /* is_bump */ false); + continue; + } + + // reflection map + if ((0 == strncmp(token, "refl", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption(&(material.reflection_texname), + &(material.reflection_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: roughness texture + if ((0 == strncmp(token, "map_Pr", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.roughness_texname), + &(material.roughness_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: metallic texture + if ((0 == strncmp(token, "map_Pm", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.metallic_texname), + &(material.metallic_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: sheen texture + if ((0 == strncmp(token, "map_Ps", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.sheen_texname), + &(material.sheen_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: emissive texture + if ((0 == strncmp(token, "map_Ke", 6)) && IS_SPACE(token[6])) { + token += 7; + ParseTextureNameAndOption(&(material.emissive_texname), + &(material.emissive_texopt), token, + /* is_bump */ false); + continue; + } + + // PBR: normal map texture + if ((0 == strncmp(token, "norm", 4)) && IS_SPACE(token[4])) { + token += 5; + ParseTextureNameAndOption( + &(material.normal_texname), &(material.normal_texopt), token, + /* is_bump */ false); // @fixme { is_bump will be true? } + continue; + } + + // unknown parameter + const char *_space = strchr(token, ' '); + if (!_space) { + _space = strchr(token, '\t'); + } + if (_space) { + std::ptrdiff_t len = _space - token; + std::string key(token, static_cast(len)); + std::string value = _space + 1; + material.unknown_parameter.insert( + std::pair(key, value)); + } + } + // flush last material. + material_map->insert(std::pair( + material.name, static_cast(materials->size()))); + materials->push_back(material); + + if (warning) { + (*warning) = ss.str(); + } +} + +bool MaterialFileReader::operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) { + std::string filepath; + + if (!m_mtlBaseDir.empty()) { + filepath = std::string(m_mtlBaseDir) + matId; + } else { + filepath = matId; + } + + std::ifstream matIStream(filepath.c_str()); + if (!matIStream) { + std::stringstream ss; + ss << "WARN: Material file [ " << filepath << " ] not found." << std::endl; + if (err) { + (*err) += ss.str(); + } + return false; + } + + std::string warning; + LoadMtl(matMap, materials, &matIStream, &warning); + + if (!warning.empty()) { + if (err) { + (*err) += warning; + } + } + + return true; +} + +bool MaterialStreamReader::operator()(const std::string &matId, + std::vector *materials, + std::map *matMap, + std::string *err) { + (void)matId; + if (!m_inStream) { + std::stringstream ss; + ss << "WARN: Material stream in error state. " << std::endl; + if (err) { + (*err) += ss.str(); + } + return false; + } + + std::string warning; + LoadMtl(matMap, materials, &m_inStream, &warning); + + if (!warning.empty()) { + if (err) { + (*err) += warning; + } + } + + return true; +} + +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + const char *filename, const char *mtl_basedir, + bool trianglulate, bool default_vcols_fallback) { + attrib->vertices.clear(); + attrib->normals.clear(); + attrib->texcoords.clear(); + attrib->colors.clear(); + shapes->clear(); + + std::stringstream errss; + + std::ifstream ifs(filename); + if (!ifs) { + errss << "Cannot open file [" << filename << "]" << std::endl; + if (err) { + (*err) = errss.str(); + } + return false; + } + + std::string baseDir = mtl_basedir ? mtl_basedir : ""; + if (!baseDir.empty()) { +#ifndef _WIN32 + const char dirsep = '/'; +#else + const char dirsep = '\\'; +#endif + if (baseDir[baseDir.length() - 1] != dirsep) baseDir += dirsep; + } + MaterialFileReader matFileReader(baseDir); + + return LoadObj(attrib, shapes, materials, err, &ifs, &matFileReader, + trianglulate, default_vcols_fallback); +} + +bool LoadObj(attrib_t *attrib, std::vector *shapes, + std::vector *materials, std::string *err, + std::istream *inStream, MaterialReader *readMatFn /*= NULL*/, + bool triangulate, bool default_vcols_fallback) { + std::stringstream errss; + + std::vector v; + std::vector vn; + std::vector vt; + std::vector vc; + std::vector tags; + std::vector faceGroup; + std::vector lineGroup; + std::string name; + + // material + std::map material_map; + int material = -1; + + // smoothing group id + unsigned int current_smoothing_id = + 0; // Initial value. 0 means no smoothing. + + int greatest_v_idx = -1; + int greatest_vn_idx = -1; + int greatest_vt_idx = -1; + + shape_t shape; + + bool found_all_colors = true; + + size_t line_num = 0; + std::string linebuf; + while (inStream->peek() != -1) { + safeGetline(*inStream, linebuf); + + line_num++; + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // vertex + if (token[0] == 'v' && IS_SPACE((token[1]))) { + token += 2; + real_t x, y, z; + real_t r, g, b; + + found_all_colors &= parseVertexWithColor(&x, &y, &z, &r, &g, &b, &token); + + v.push_back(x); + v.push_back(y); + v.push_back(z); + + if (found_all_colors || default_vcols_fallback) { + vc.push_back(r); + vc.push_back(g); + vc.push_back(b); + } + + continue; + } + + // normal + if (token[0] == 'v' && token[1] == 'n' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; + parseReal3(&x, &y, &z, &token); + vn.push_back(x); + vn.push_back(y); + vn.push_back(z); + continue; + } + + // texcoord + if (token[0] == 'v' && token[1] == 't' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y; + parseReal2(&x, &y, &token); + vt.push_back(x); + vt.push_back(y); + continue; + } + + // line + if (token[0] == 'l' && IS_SPACE((token[1]))) { + token += 2; + + line_t line_cache; + bool end_line_bit = 0; + while (!IS_NEW_LINE(token[0])) { + // get index from string + int idx; + fixIndex(parseInt(&token), 0, &idx); + + size_t n = strspn(token, " \t\r"); + token += n; + + if (!end_line_bit) { + line_cache.idx0 = idx; + } else { + line_cache.idx1 = idx; + lineGroup.push_back(line_cache.idx0); + lineGroup.push_back(line_cache.idx1); + line_cache = line_t(); + } + end_line_bit = !end_line_bit; + } + + continue; + } + // face + if (token[0] == 'f' && IS_SPACE((token[1]))) { + token += 2; + token += strspn(token, " \t"); + + face_t face; + + face.smoothing_group_id = current_smoothing_id; + face.vertex_indices.reserve(3); + + while (!IS_NEW_LINE(token[0])) { + vertex_index_t vi; + if (!parseTriple(&token, static_cast(v.size() / 3), + static_cast(vn.size() / 3), + static_cast(vt.size() / 2), &vi)) { + if (err) { + (*err) = "Failed parse `f' line(e.g. zero value for face index).\n"; + } + return false; + } + + greatest_v_idx = greatest_v_idx > vi.v_idx ? greatest_v_idx : vi.v_idx; + greatest_vn_idx = greatest_vn_idx > vi.vn_idx ? greatest_vn_idx : vi.vn_idx; + greatest_vt_idx = greatest_vt_idx > vi.vt_idx ? greatest_vt_idx : vi.vt_idx; + + face.vertex_indices.push_back(vi); + size_t n = strspn(token, " \t\r"); + token += n; + } + + // replace with emplace_back + std::move on C++11 + faceGroup.push_back(face); + + continue; + } + + // use mtl + if ((0 == strncmp(token, "usemtl", 6)) && IS_SPACE((token[6]))) { + token += 7; + std::stringstream ss; + ss << token; + std::string namebuf = ss.str(); + + int newMaterialId = -1; + if (material_map.find(namebuf) != material_map.end()) { + newMaterialId = material_map[namebuf]; + } else { + // { error!! material not found } + } + + if (newMaterialId != material) { + // Create per-face material. Thus we don't add `shape` to `shapes` at + // this time. + // just clear `faceGroup` after `exportGroupsToShape()` call. + exportGroupsToShape(&shape, faceGroup, lineGroup, tags, material, name, + triangulate, v); + faceGroup.clear(); + material = newMaterialId; + } + + continue; + } + + // load mtl + if ((0 == strncmp(token, "mtllib", 6)) && IS_SPACE((token[6]))) { + if (readMatFn) { + token += 7; + + std::vector filenames; + SplitString(std::string(token), ' ', filenames); + + if (filenames.empty()) { + if (err) { + (*err) += + "WARN: Looks like empty filename for mtllib. Use default " + "material. \n"; + } + } else { + bool found = false; + for (size_t s = 0; s < filenames.size(); s++) { + std::string err_mtl; + bool ok = (*readMatFn)(filenames[s].c_str(), materials, + &material_map, &err_mtl); + if (err && (!err_mtl.empty())) { + (*err) += err_mtl; // This should be warn message. + } + + if (ok) { + found = true; + break; + } + } + + if (!found) { + if (err) { + (*err) += + "WARN: Failed to load material file(s). Use default " + "material.\n"; + } + } + } + } + + continue; + } + + // group name + if (token[0] == 'g' && IS_SPACE((token[1]))) { + // flush previous face group. + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, + material, name, triangulate, v); + (void)ret; // return value not used. + + if (shape.mesh.indices.size() > 0) { + shapes->push_back(shape); + } + + shape = shape_t(); + + // material = -1; + faceGroup.clear(); + + std::vector names; + + while (!IS_NEW_LINE(token[0])) { + std::string str = parseString(&token); + names.push_back(str); + token += strspn(token, " \t\r"); // skip tag + } + + // names[0] must be 'g' + + if (names.size() < 2) { + // 'g' with empty names + if (err) { + std::stringstream ss; + ss << "WARN: Empty group name. line: " << line_num << "\n"; + (*err) += ss.str(); + name = ""; + } + } else { + + std::stringstream ss; + ss << names[1]; + + // tinyobjloader does not support multiple groups for a primitive. + // Currently we concatinate multiple group names with a space to get + // single group name. + + for (size_t i = 2; i < names.size(); i++) { + ss << " " << names[i]; + } + + name = ss.str(); + + } + + continue; + } + + // object name + if (token[0] == 'o' && IS_SPACE((token[1]))) { + // flush previous face group. + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, + material, name, triangulate, v); + if (ret) { + shapes->push_back(shape); + } + + // material = -1; + faceGroup.clear(); + shape = shape_t(); + + // @todo { multiple object name? } + token += 2; + std::stringstream ss; + ss << token; + name = ss.str(); + + continue; + } + + if (token[0] == 't' && IS_SPACE(token[1])) { + const int max_tag_nums = 8192; // FIXME(syoyo): Parameterize. + tag_t tag; + + token += 2; + + tag.name = parseString(&token); + + tag_sizes ts = parseTagTriple(&token); + + if (ts.num_ints < 0) { + ts.num_ints = 0; + } + if (ts.num_ints > max_tag_nums) { + ts.num_ints = max_tag_nums; + } + + if (ts.num_reals < 0) { + ts.num_reals = 0; + } + if (ts.num_reals > max_tag_nums) { + ts.num_reals = max_tag_nums; + } + + if (ts.num_strings < 0) { + ts.num_strings = 0; + } + if (ts.num_strings > max_tag_nums) { + ts.num_strings = max_tag_nums; + } + + tag.intValues.resize(static_cast(ts.num_ints)); + + for (size_t i = 0; i < static_cast(ts.num_ints); ++i) { + tag.intValues[i] = parseInt(&token); + } + + tag.floatValues.resize(static_cast(ts.num_reals)); + for (size_t i = 0; i < static_cast(ts.num_reals); ++i) { + tag.floatValues[i] = parseReal(&token); + } + + tag.stringValues.resize(static_cast(ts.num_strings)); + for (size_t i = 0; i < static_cast(ts.num_strings); ++i) { + tag.stringValues[i] = parseString(&token); + } + + tags.push_back(tag); + + continue; + } + + if (token[0] == 's' && IS_SPACE(token[1])) { + // smoothing group id + token += 2; + + // skip space. + token += strspn(token, " \t"); // skip space + + if (token[0] == '\0') { + continue; + } + + if (token[0] == '\r' || token[1] == '\n') { + continue; + } + + if (strlen(token) >= 3) { + if (token[0] == 'o' && token[1] == 'f' && token[2] == 'f') { + current_smoothing_id = 0; + } + } else { + // assume number + int smGroupId = parseInt(&token); + if (smGroupId < 0) { + // parse error. force set to 0. + // FIXME(syoyo): Report warning. + current_smoothing_id = 0; + } else { + current_smoothing_id = static_cast(smGroupId); + } + } + + continue; + } // smoothing group id + + // Ignore unknown command. + } + + // not all vertices have colors, no default colors desired? -> clear colors + if (!found_all_colors && !default_vcols_fallback) { + vc.clear(); + } + + if (greatest_v_idx >= static_cast(v.size() / 3)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + if (greatest_vn_idx >= static_cast(vn.size() / 3)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex normal indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + if (greatest_vt_idx >= static_cast(vt.size() / 2)) + { + if (err) { + std::stringstream ss; + ss << "WARN: Vertex texcoord indices out of bounds.\n" << std::endl; + (*err) += ss.str(); + } + } + + bool ret = exportGroupsToShape(&shape, faceGroup, lineGroup, tags, material, + name, triangulate, v); + // exportGroupsToShape return false when `usemtl` is called in the last + // line. + // we also add `shape` to `shapes` when `shape.mesh` has already some + // faces(indices) + if (ret || shape.mesh.indices.size()) { + shapes->push_back(shape); + } + faceGroup.clear(); // for safety + + if (err) { + (*err) += errss.str(); + } + + attrib->vertices.swap(v); + attrib->normals.swap(vn); + attrib->texcoords.swap(vt); + attrib->colors.swap(vc); + + return true; +} + +bool LoadObjWithCallback(std::istream &inStream, const callback_t &callback, + void *user_data /*= NULL*/, + MaterialReader *readMatFn /*= NULL*/, + std::string *err /*= NULL*/) { + std::stringstream errss; + + // material + std::map material_map; + int material_id = -1; // -1 = invalid + + std::vector indices; + std::vector materials; + std::vector names; + names.reserve(2); + std::vector names_out; + + std::string linebuf; + while (inStream.peek() != -1) { + safeGetline(inStream, linebuf); + + // Trim newline '\r\n' or '\n' + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\n') + linebuf.erase(linebuf.size() - 1); + } + if (linebuf.size() > 0) { + if (linebuf[linebuf.size() - 1] == '\r') + linebuf.erase(linebuf.size() - 1); + } + + // Skip if empty line. + if (linebuf.empty()) { + continue; + } + + // Skip leading space. + const char *token = linebuf.c_str(); + token += strspn(token, " \t"); + + assert(token); + if (token[0] == '\0') continue; // empty line + + if (token[0] == '#') continue; // comment line + + // vertex + if (token[0] == 'v' && IS_SPACE((token[1]))) { + token += 2; + // TODO(syoyo): Support parsing vertex color extension. + real_t x, y, z, w; // w is optional. default = 1.0 + parseV(&x, &y, &z, &w, &token); + if (callback.vertex_cb) { + callback.vertex_cb(user_data, x, y, z, w); + } + continue; + } + + // normal + if (token[0] == 'v' && token[1] == 'n' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; + parseReal3(&x, &y, &z, &token); + if (callback.normal_cb) { + callback.normal_cb(user_data, x, y, z); + } + continue; + } + + // texcoord + if (token[0] == 'v' && token[1] == 't' && IS_SPACE((token[2]))) { + token += 3; + real_t x, y, z; // y and z are optional. default = 0.0 + parseReal3(&x, &y, &z, &token); + if (callback.texcoord_cb) { + callback.texcoord_cb(user_data, x, y, z); + } + continue; + } + + // face + if (token[0] == 'f' && IS_SPACE((token[1]))) { + token += 2; + token += strspn(token, " \t"); + + indices.clear(); + while (!IS_NEW_LINE(token[0])) { + vertex_index_t vi = parseRawTriple(&token); + + index_t idx; + idx.vertex_index = vi.v_idx; + idx.normal_index = vi.vn_idx; + idx.texcoord_index = vi.vt_idx; + + indices.push_back(idx); + size_t n = strspn(token, " \t\r"); + token += n; + } + + if (callback.index_cb && indices.size() > 0) { + callback.index_cb(user_data, &indices.at(0), + static_cast(indices.size())); + } + + continue; + } + + // use mtl + if ((0 == strncmp(token, "usemtl", 6)) && IS_SPACE((token[6]))) { + token += 7; + std::stringstream ss; + ss << token; + std::string namebuf = ss.str(); + + int newMaterialId = -1; + if (material_map.find(namebuf) != material_map.end()) { + newMaterialId = material_map[namebuf]; + } else { + // { error!! material not found } + } + + if (newMaterialId != material_id) { + material_id = newMaterialId; + } + + if (callback.usemtl_cb) { + callback.usemtl_cb(user_data, namebuf.c_str(), material_id); + } + + continue; + } + + // load mtl + if ((0 == strncmp(token, "mtllib", 6)) && IS_SPACE((token[6]))) { + if (readMatFn) { + token += 7; + + std::vector filenames; + SplitString(std::string(token), ' ', filenames); + + if (filenames.empty()) { + if (err) { + (*err) += + "WARN: Looks like empty filename for mtllib. Use default " + "material. \n"; + } + } else { + bool found = false; + for (size_t s = 0; s < filenames.size(); s++) { + std::string err_mtl; + bool ok = (*readMatFn)(filenames[s].c_str(), &materials, + &material_map, &err_mtl); + if (err && (!err_mtl.empty())) { + (*err) += err_mtl; // This should be warn message. + } + + if (ok) { + found = true; + break; + } + } + + if (!found) { + if (err) { + (*err) += + "WARN: Failed to load material file(s). Use default " + "material.\n"; + } + } else { + if (callback.mtllib_cb) { + callback.mtllib_cb(user_data, &materials.at(0), + static_cast(materials.size())); + } + } + } + } + + continue; + } + + // group name + if (token[0] == 'g' && IS_SPACE((token[1]))) { + names.clear(); + + while (!IS_NEW_LINE(token[0])) { + std::string str = parseString(&token); + names.push_back(str); + token += strspn(token, " \t\r"); // skip tag + } + + assert(names.size() > 0); + + if (callback.group_cb) { + if (names.size() > 1) { + // create const char* array. + names_out.resize(names.size() - 1); + for (size_t j = 0; j < names_out.size(); j++) { + names_out[j] = names[j + 1].c_str(); + } + callback.group_cb(user_data, &names_out.at(0), + static_cast(names_out.size())); + + } else { + callback.group_cb(user_data, NULL, 0); + } + } + + continue; + } + + // object name + if (token[0] == 'o' && IS_SPACE((token[1]))) { + // @todo { multiple object name? } + token += 2; + + std::stringstream ss; + ss << token; + std::string object_name = ss.str(); + + if (callback.object_cb) { + callback.object_cb(user_data, object_name.c_str()); + } + + continue; + } + +#if 0 // @todo + if (token[0] == 't' && IS_SPACE(token[1])) { + tag_t tag; + + token += 2; + std::stringstream ss; + ss << token; + tag.name = ss.str(); + + token += tag.name.size() + 1; + + tag_sizes ts = parseTagTriple(&token); + + tag.intValues.resize(static_cast(ts.num_ints)); + + for (size_t i = 0; i < static_cast(ts.num_ints); ++i) { + tag.intValues[i] = atoi(token); + token += strcspn(token, "/ \t\r") + 1; + } + + tag.floatValues.resize(static_cast(ts.num_reals)); + for (size_t i = 0; i < static_cast(ts.num_reals); ++i) { + tag.floatValues[i] = parseReal(&token); + token += strcspn(token, "/ \t\r") + 1; + } + + tag.stringValues.resize(static_cast(ts.num_strings)); + for (size_t i = 0; i < static_cast(ts.num_strings); ++i) { + std::stringstream ss; + ss << token; + tag.stringValues[i] = ss.str(); + token += tag.stringValues[i].size() + 1; + } + + tags.push_back(tag); + } +#endif + + // Ignore unknown command. + } + + if (err) { + (*err) += errss.str(); + } + + return true; +} + +#ifdef __clang__ +#pragma clang diagnostic pop +#endif +} // namespace tinyobj + +#endif diff --git a/stream_compaction/CMakeLists.txt b/stream_compaction/CMakeLists.txt index ac358c9..e31ca3c 100644 --- a/stream_compaction/CMakeLists.txt +++ b/stream_compaction/CMakeLists.txt @@ -1,7 +1,17 @@ set(SOURCE_FILES + "common.h" + "common.cu" + "cpu.h" + "cpu.cu" + "naive.h" + "naive.cu" + "efficient.h" + "efficient.cu" + "thrust.h" + "thrust.cu" ) cuda_add_library(stream_compaction ${SOURCE_FILES} - OPTIONS -arch=sm_20 + OPTIONS -arch=sm_30 ) diff --git a/stream_compaction/common.cu b/stream_compaction/common.cu new file mode 100644 index 0000000..6e54f5c --- /dev/null +++ b/stream_compaction/common.cu @@ -0,0 +1,51 @@ +#include "common.h" + +//void checkCUDAErrorFn(const char *msg, const char *file, int line) { +// cudaError_t err = cudaGetLastError(); +// if (cudaSuccess == err) { +// return; +// } +// +// fprintf(stderr, "CUDA error"); +// if (file) { +// fprintf(stderr, " (%s:%d)", file, line); +// } +// fprintf(stderr, ": %s: %s\n", msg, cudaGetErrorString(err)); +// exit(EXIT_FAILURE); +//} + + +namespace StreamCompaction { + namespace Common { + + /** + * Maps an array to an array of 0s and 1s for stream compaction. Elements + * which map to 0 will be removed, and elements which map to 1 will be kept. + */ + __global__ void kernMapToBoolean(int n, int *bools, const int *idata) { + // TODO + int index = blockDim.x * blockIdx.x + threadIdx.x; + if (index >= n) { + return; + } + bools[index] = idata[index] == 0 ? 0 : 1; + } + + /** + * Performs scatter on an array. That is, for each element in idata, + * if bools[idx] == 1, it copies idata[idx] to odata[indices[idx]]. + */ + __global__ void kernScatter(int n, int *odata, + const int *idata, const int *bools, const int *indices) { + // TODO + int index = blockDim.x * blockIdx.x + threadIdx.x; + if (index >= n) { + return; + } + if (bools[index] == 1) { + odata[indices[index]] = idata[index]; + } + } + + } +} diff --git a/stream_compaction/common.h b/stream_compaction/common.h new file mode 100644 index 0000000..996997e --- /dev/null +++ b/stream_compaction/common.h @@ -0,0 +1,132 @@ +#pragma once + +#include +#include + +#include +#include +#include +#include +#include +#include + +#define FILENAME (strrchr(__FILE__, '/') ? strrchr(__FILE__, '/') + 1 : __FILE__) +#define checkCUDAError(msg) checkCUDAErrorFn(msg, FILENAME, __LINE__) + +/** + * Check for CUDA errors; print and exit if there was a problem. + */ +void checkCUDAErrorFn(const char *msg, const char *file = NULL, int line = -1); + +inline int ilog2(int x) { + int lg = 0; + while (x >>= 1) { + ++lg; + } + return lg; +} + +inline int ilog2ceil(int x) { + return x == 1 ? 0 : ilog2(x - 1) + 1; +} + +namespace StreamCompaction { + namespace Common { + __global__ void kernMapToBoolean(int n, int *bools, const int *idata); + + __global__ void kernScatter(int n, int *odata, + const int *idata, const int *bools, const int *indices); + + /** + * This class is used for timing the performance + * Uncopyable and unmovable + * + * Adapted from WindyDarian(https://github.com/WindyDarian) + */ + class PerformanceTimer + { + public: + PerformanceTimer() + { + cudaEventCreate(&event_start); + cudaEventCreate(&event_end); + } + + ~PerformanceTimer() + { + cudaEventDestroy(event_start); + cudaEventDestroy(event_end); + } + + void startCpuTimer() + { + if (cpu_timer_started) { throw std::runtime_error("CPU timer already started"); } + cpu_timer_started = true; + + time_start_cpu = std::chrono::high_resolution_clock::now(); + } + + void endCpuTimer() + { + time_end_cpu = std::chrono::high_resolution_clock::now(); + + if (!cpu_timer_started) { throw std::runtime_error("CPU timer not started"); } + + std::chrono::duration duro = time_end_cpu - time_start_cpu; + prev_elapsed_time_cpu_milliseconds = + static_cast(duro.count()); + + cpu_timer_started = false; + } + + void startGpuTimer() + { + if (gpu_timer_started) { throw std::runtime_error("GPU timer already started"); } + gpu_timer_started = true; + + cudaEventRecord(event_start); + } + + void endGpuTimer() + { + cudaEventRecord(event_end); + cudaEventSynchronize(event_end); + + if (!gpu_timer_started) { throw std::runtime_error("GPU timer not started"); } + + cudaEventElapsedTime(&prev_elapsed_time_gpu_milliseconds, event_start, event_end); + gpu_timer_started = false; + } + + float getCpuElapsedTimeForPreviousOperation() //noexcept //(damn I need VS 2015 + { + return prev_elapsed_time_cpu_milliseconds; + } + + float getGpuElapsedTimeForPreviousOperation() //noexcept + { + return prev_elapsed_time_gpu_milliseconds; + } + + // remove copy and move functions + PerformanceTimer(const PerformanceTimer&) = delete; + PerformanceTimer(PerformanceTimer&&) = delete; + PerformanceTimer& operator=(const PerformanceTimer&) = delete; + PerformanceTimer& operator=(PerformanceTimer&&) = delete; + + private: + cudaEvent_t event_start = nullptr; + cudaEvent_t event_end = nullptr; + + using time_point_t = std::chrono::high_resolution_clock::time_point; + time_point_t time_start_cpu; + time_point_t time_end_cpu; + + bool cpu_timer_started = false; + bool gpu_timer_started = false; + + float prev_elapsed_time_cpu_milliseconds = 0.f; + float prev_elapsed_time_gpu_milliseconds = 0.f; + }; + } +} diff --git a/stream_compaction/cpu.cu b/stream_compaction/cpu.cu new file mode 100644 index 0000000..296eb0b --- /dev/null +++ b/stream_compaction/cpu.cu @@ -0,0 +1,81 @@ +#include +#include "cpu.h" + +#include "common.h" + + +namespace StreamCompaction { + namespace CPU { + using StreamCompaction::Common::PerformanceTimer; + PerformanceTimer& timer() + { + static PerformanceTimer timer; + return timer; + } + + /** + * CPU scan (prefix sum). + * For performance analysis, this is supposed to be a simple for loop. + * (Optional) For better understanding before starting moving to GPU, you can simulate your GPU scan in this function first. + */ + void scan(int n, int *odata, const int *idata) { + //timer().startCpuTimer(); + // TODO + odata[0] = 0; + for (int i = 1; i < n; i++) { + odata[i] = odata[i - 1] + idata[i - 1]; + } + //timer().endCpuTimer(); + } + + /** + * CPU stream compaction without using the scan function. + * + * @returns the number of elements remaining after compaction. + */ + int compactWithoutScan(int n, int *odata, const int *idata) { + timer().startCpuTimer(); + // TODO + int count = 0; + for (int i = 0; i < n; i++) { + if (idata[i] == 0) { + continue; + } + odata[count++] = idata[i]; + } + timer().endCpuTimer(); + return count; + } + + /**s + * CPU stream compaction using scan and scatter, like the parallel version. + * + * @returns the number of elements remaining after compaction. + */ + int compactWithScan(int n, int *odata, const int *idata) { + timer().startCpuTimer(); + // TODO + int *mdata = new int[n]; + int *sdata = new int[n]; + int count = 0; + for (int i = 0; i < n; i++) { + if (idata[i] == 0) { + mdata[i] = 0; + } + else { + mdata[i] = 1; + } + } + scan(n, sdata, mdata); + for (int i = 0; i < n; i++) { + if (mdata[i] != 0) { + odata[sdata[i]] = idata[i]; + count++; + } + } + delete[] mdata, sdata; + timer().endCpuTimer(); + return count; + } + } +} diff --git a/stream_compaction/cpu.h b/stream_compaction/cpu.h new file mode 100644 index 0000000..236ce11 --- /dev/null +++ b/stream_compaction/cpu.h @@ -0,0 +1,15 @@ +#pragma once + +#include "common.h" + +namespace StreamCompaction { + namespace CPU { + StreamCompaction::Common::PerformanceTimer& timer(); + + void scan(int n, int *odata, const int *idata); + + int compactWithoutScan(int n, int *odata, const int *idata); + + int compactWithScan(int n, int *odata, const int *idata); + } +} diff --git a/stream_compaction/efficient.cu b/stream_compaction/efficient.cu new file mode 100644 index 0000000..bf1bcc7 --- /dev/null +++ b/stream_compaction/efficient.cu @@ -0,0 +1,354 @@ +#include +#include +#include "common.h" +#include "efficient.h" +#include "device_launch_parameters.h" +#include +#include +#include +#include "thrust.h" + +#define NUM_BANKS 16 +#define LOG_NUM_BANKS 4 +#define CONFLICT_FREE_OFFSET(n) \ + ((n) >> NUM_BANKS + (n) >> (2 * LOG_NUM_BANKS)) +#define BLOCKSIZE_SHARED 64 + +namespace StreamCompaction { + namespace Efficient { + using StreamCompaction::Common::PerformanceTimer; + PerformanceTimer& timer() + { + static PerformanceTimer timer; + return timer; + } + + __global__ void kernUpSweep(int n, int d, int *odata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + int step = 1 << (d + 1); + if (index % step == 0) { + odata[index + step - 1] += odata[index + (1 << d) - 1]; + } + } + + __global__ void kernDownSweep(int n, int d, int *odata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + int step = 1 << (d + 1); + if (index % step == 0) { + int t = odata[index + (1 << d) - 1]; + odata[index + (1 << d) - 1] = odata[index + step - 1]; + odata[index + step - 1] += t; + } + } + + + + /** + * Performs prefix-sum (aka scan) on idata, storing the result into odata. + */ + void nonoptscan(int n, int *odata, const int *idata) { + // TODO + int upLimit = ilog2ceil(n); + int len = 1 << upLimit; + + int blockSize = 1024; + dim3 threadsPerBlock(blockSize); + dim3 blocksPerGrid((len + blockSize - 1) / blockSize); + + int *dev_data; + cudaMalloc((void**)&dev_data, len * sizeof(int)); + checkCUDAError("cudaMalloc dev_data failed!"); + cudaMemcpy(dev_data, idata, len * sizeof(int), cudaMemcpyHostToDevice); + checkCUDAError("cudaMemcpy dev_data failed!"); + + + + timer().startGpuTimer(); + for (int d = 0; d <= upLimit - 1; d++) { + kernUpSweep << > > (len, d, dev_data); + checkCUDAError("kernUpSweep failed!"); + } + + cudaMemset(&dev_data[len - 1], 0, sizeof(int)); + checkCUDAError("cudaMemcpy set last one to be zero failed!"); + + for (int d = upLimit - 1; d >= 0; d--) { + kernDownSweep << > > (len, d, dev_data); + checkCUDAError("kernDownSweep failed!"); + } + timer().endGpuTimer(); + cudaMemcpy(odata, dev_data, len * sizeof(int), cudaMemcpyDeviceToHost); + checkCUDAError("cudaMemcpy dev_data failed!"); + cudaFree(dev_data); + cudaDeviceSynchronize(); + } + + __global__ void kernOptUpSweep(int n, int d, int *odata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + int step = 1 << (d + 1); + odata[index * step + step - 1] += odata[index * step + (1 << d) - 1]; + } + + __global__ void kernOptDownSweep(int n, int d, int *odata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + int step = 1 << (d + 1); + int t = odata[index*step + (1 << d) - 1]; + odata[index * step + (1 << d) - 1] = odata[index * step + step - 1]; + odata[index * step + step - 1] += t; + } + + void scan(int n, int *odata, const int *idata) { + int upLimit = ilog2ceil(n); + int len = 1 << upLimit; + + int blockSize = 1024; + dim3 threadsPerBlock(blockSize); + dim3 blocksPerGrid((len + blockSize - 1) / blockSize); + + int *dev_data; + cudaMalloc((void**)&dev_data, len * sizeof(int)); + checkCUDAError("cudaMalloc dev_data failed!"); + cudaMemcpy(dev_data, idata, len * sizeof(int), cudaMemcpyHostToDevice); + checkCUDAError("cudaMemcpy dev_data failed!"); + + + + timer().startGpuTimer(); + for (int d = 0; d <= upLimit - 1; d++) { + int step = 1 << (d + 1); + int tempLen = len / step; + blocksPerGrid = dim3((tempLen + blockSize) / blockSize); + kernOptUpSweep << > > (tempLen, d, dev_data); + checkCUDAError("kernUpSweep failed!"); + } + + cudaMemset(&dev_data[len - 1], 0, sizeof(int)); + checkCUDAError("cudaMemcpy set last one to be zero failed!"); + + for (int d = upLimit - 1; d >= 0; d--) { + int step = 1 << (d + 1); + int tempLen = len / step; + blocksPerGrid = dim3((tempLen + blockSize) / blockSize); + kernOptDownSweep << > > (tempLen, d, dev_data); + checkCUDAError("kernDownSweep failed!"); + } + timer().endGpuTimer(); + cudaMemcpy(odata, dev_data, len * sizeof(int), cudaMemcpyDeviceToHost); + checkCUDAError("cudaMemcpy dev_data failed!"); + cudaFree(dev_data); + cudaDeviceSynchronize(); + } + + void gpuScan(int n, int *data) { + int upLimit = ilog2ceil(n); + int blockSize = 1024; + dim3 threadsPerBlock(blockSize); + dim3 blocksPerGrid((n + blockSize - 1) / blockSize); + + for (int d = 0; d <= upLimit - 1; d++) { + kernUpSweep << > > (n, d, data); + checkCUDAError("kernUpSweep failed!"); + } + + cudaMemset(&data[n - 1], 0, sizeof(int)); + checkCUDAError("cudaMemcpy set last one to be zero failed!"); + + for (int d = upLimit - 1; d >= 0; d--) { + kernDownSweep << > > (n, d, data); + checkCUDAError("kernDownSweep failed!"); + } + } + + __global__ void kernSharedMemoryScan(int n, int *odata, const int *idata, int *blockData) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + extern __shared__ int temp[BLOCKSIZE_SHARED + NUM_BANKS]; + int thid = threadIdx.x; + int offset = 1; + int stride = blockIdx.x * blockDim.x; + int ai = thid << 1; + int bi = ai + 1; + int bankOffsetA = CONFLICT_FREE_OFFSET(ai); + int bankOffsetB = CONFLICT_FREE_OFFSET(bi); + temp[ai + bankOffsetA] = idata[ai + stride]; + temp[bi + bankOffsetB] = idata[bi + stride]; + for (int d = BLOCKSIZE_SHARED >> 1; d > 0; d >>= 1) + { + __syncthreads(); + if (thid < d) + { + int ai = offset*(2 * thid + 1) - 1; + int bi = offset*(2 * thid + 2) - 1; + ai += CONFLICT_FREE_OFFSET(ai); + bi += CONFLICT_FREE_OFFSET(bi); + temp[bi] += temp[ai]; + } + offset *= 2; + } + __syncthreads(); + if (thid != 0) { + blockData[blockIdx.x] = temp[BLOCKSIZE_SHARED - 1 + CONFLICT_FREE_OFFSET(BLOCKSIZE_SHARED - 1)]; + temp[BLOCKSIZE_SHARED - 1 + CONFLICT_FREE_OFFSET(BLOCKSIZE_SHARED - 1)] = 0; + } + for (int d = 1; d < BLOCKSIZE_SHARED; d *= 2) // traverse down tree & build scan + { + offset >>= 1; + __syncthreads(); + if (thid < d) + { + int ai = offset*(2 * thid + 1) - 1; + int bi = offset*(2 * thid + 2) - 1; + ai += CONFLICT_FREE_OFFSET(ai); + bi += CONFLICT_FREE_OFFSET(bi); + float t = temp[ai]; + temp[ai] = temp[bi]; + temp[bi] += t; + } + } + ai = thid << 1; + bi = ai + 1; + odata[ai + stride] = temp[ai + bankOffsetA]; + odata[bi + stride] = temp[bi + bankOffsetB]; + } + + __global__ void kernAddBlockData(int n, const int *idata, int *odata, const int *blockData) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + int bid = blockIdx.x; + if (bid > 0) { + odata[index] = idata[index] + blockData[index]; + } + else { + odata[index] = idata[index]; + } + } + + void sharedMemoryScan(int n, int *odata, const int *idata) { + int upLimit = ilog2ceil(n); + int len = 1 << upLimit; + + int blockSize = 64; + dim3 threadsPerBlock(BLOCKSIZE_SHARED / 2); + dim3 blocksPerGrid((n + BLOCKSIZE_SHARED - 1) / blockSize); + + + int alen = ((n + BLOCKSIZE_SHARED - 1) / BLOCKSIZE_SHARED); + dim3 threadsPerBlock_Block(BLOCKSIZE_SHARED / 2); + dim3 blocksPerGrid_Block((alen + BLOCKSIZE_SHARED - 1) / BLOCKSIZE_SHARED); + int *dev_idata; + int *dev_odata; + int *dev_blockData; + int *dev_blockSum; + cudaMalloc((void**)&dev_idata, len * sizeof(int)); + checkCUDAError("cudaMalloc dev_idata failed!"); + cudaMalloc((void**)&dev_odata, len * sizeof(int)); + checkCUDAError("cudaMalloc dev_odata failed!"); + cudaMalloc((void**)&dev_blockData, alen * sizeof(int)); + checkCUDAError("cudaMalloc dev_blockData failed!"); + cudaMalloc((void**)&dev_blockSum, alen * sizeof(int)); + checkCUDAError("cudaMalloc dev_blockSum failed!"); + cudaMemcpy(dev_idata, idata, n * sizeof(int), cudaMemcpyHostToDevice); + checkCUDAError("cudaMemcpy dev_idata failed!"); + + int *host_aux, *host_aux_sum; + cudaMallocHost((void**)&host_aux, alen * sizeof(int)); // Use Pin-Memory to ensure the maximum memory speed + cudaMallocHost((void**)&host_aux_sum, alen * sizeof(int)); + + timer().startGpuTimer(); + kernSharedMemoryScan << > > (len, dev_odata, dev_idata, dev_blockData); + checkCUDAError("kernSharedMemoryScan failed!"); + cudaMemcpy(host_aux, dev_blockData, alen * sizeof(int), cudaMemcpyDeviceToHost); + thrust::inclusive_scan(host_aux, host_aux + alen, host_aux_sum); + cudaMemcpy(dev_blockSum, host_aux_sum, alen * sizeof(int), cudaMemcpyHostToDevice); + kernAddBlockData << > > (n, dev_odata, dev_idata, dev_blockSum); + checkCUDAError("kernSharedMemoryScan failed!"); + timer().endGpuTimer(); + cudaMemcpy(odata, dev_odata, n * sizeof(int), cudaMemcpyDeviceToHost); + checkCUDAError("cudaMemcpy dev_odata failed!"); + + cudaFree(dev_idata); + cudaFree(dev_odata); + cudaFree(dev_blockData); + cudaFree(dev_blockSum); + cudaDeviceSynchronize(); + cudaFreeHost(host_aux); + cudaFreeHost(host_aux_sum); + } + + + + /** + * Performs stream compaction on idata, storing the result into odata. + * All zeroes are discarded. + * + * @param n The number of elements in idata. + * @param odata The array into which to store elements. + * @param idata The array of elements to compact. + * @returns The number of elements remaining after compaction. + */ + int compact(int n, int *odata, const int *idata) { + int upLimit = ilog2ceil(n); + int len = 1 << upLimit; + int blockSize = 1024; + dim3 threadsPerBlock(blockSize); + dim3 blocksPerGrid((len + blockSize - 1) / blockSize); + int *dev_odata; + int *dev_idata; + int *dev_bools; + int *dev_indices; + + cudaMalloc((void**)&dev_odata, sizeof(int) * len); + checkCUDAError("cudaMalloc dev_odata failed!"); + cudaMalloc((void**)&dev_idata, sizeof(int) * len); + checkCUDAError("cudaMalloc dev_idata failed!"); + cudaMalloc((void**)&dev_bools, sizeof(int) * len); + checkCUDAError("cudaMalloc dev_bools failed!"); + cudaMalloc((void**)&dev_indices, sizeof(int) * len); + checkCUDAError("cudaMalloc dev_indices failed!"); + cudaMemcpy(dev_idata, idata, sizeof(int) * n, cudaMemcpyHostToDevice); + checkCUDAError("cudaMemcpy dev_idata failed!"); + + timer().startGpuTimer(); + + StreamCompaction::Common::kernMapToBoolean << > > (len, dev_bools, dev_idata); + checkCUDAError("kernMaptoBoolean failed!"); + + cudaMemcpy(dev_indices, dev_bools, sizeof(int) * len, cudaMemcpyDeviceToDevice); + cudaMemset(&dev_odata[n], 0, sizeof(int) * (len - n)); + gpuScan(len, dev_indices); + + StreamCompaction::Common::kernScatter << > > (len, dev_odata, dev_idata, dev_bools, dev_indices); + checkCUDAError("kernScatter failed!"); + + timer().endGpuTimer(); + cudaMemcpy(odata, dev_odata, sizeof(int) * len, cudaMemcpyDeviceToHost); + checkCUDAError("cudaMemcpy odata failed!"); + + cudaFree(dev_odata); + cudaFree(dev_idata); + cudaFree(dev_indices); + cudaFree(dev_bools); + int count = 0; + while (odata[count] != 0) { + count++; + } + return count; + } + } +} diff --git a/stream_compaction/efficient.h b/stream_compaction/efficient.h new file mode 100644 index 0000000..1ac55d8 --- /dev/null +++ b/stream_compaction/efficient.h @@ -0,0 +1,15 @@ +#pragma once + +#include "common.h" + +namespace StreamCompaction { + namespace Efficient { + StreamCompaction::Common::PerformanceTimer& timer(); + + void scan(int n, int *odata, const int *idata); + + void sharedMemoryScan(int n, int *odata, const int *idata); + + int compact(int n, int *odata, const int *idata); + } +} diff --git a/stream_compaction/naive.cu b/stream_compaction/naive.cu new file mode 100644 index 0000000..afe9a92 --- /dev/null +++ b/stream_compaction/naive.cu @@ -0,0 +1,79 @@ +#include +#include +#include "common.h" +#include "naive.h" + +namespace StreamCompaction { + namespace Naive { + using StreamCompaction::Common::PerformanceTimer; + PerformanceTimer& timer() + { + static PerformanceTimer timer; + return timer; + } + // TODO: __global__ + __global__ void kernNaiveScan(int n, int bound, int *odata, const int *idata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + if (index >= bound) { + odata[index] = idata[index - bound] + idata[index]; + } + else { + odata[index] = idata[index]; + } + } + + __global__ void kernInclusiveToExclusive(int n, int *odata, const int *idata) { + int index = (blockDim.x * blockIdx.x) + threadIdx.x; + if (index >= n) { + return; + } + odata[index] = index == 0 ? 0 : idata[index - 1]; + } + + + /** + * Performs prefix-sum (aka scan) on idata, storing the result into odata. + */ + void scan(int n, int *odata, const int *idata) { + // TODO + int blockSize = 1024; + dim3 threadsPerBlock(blockSize); + dim3 blocksPerGrid((n + blockSize - 1) / blockSize); + + // allocate memory + int *dev_idata; + int *dev_odata; + + cudaMalloc((void**)&dev_idata, n * sizeof(int)); + checkCUDAError("cudaMalloc dev_idata failed!"); + cudaMalloc((void**)&dev_odata, n * sizeof(int)); + checkCUDAError("cudaMalloc dev_odata failed!"); + + + cudaMemcpy(dev_idata, idata, n * sizeof(int), cudaMemcpyHostToDevice); + checkCUDAError("cudaMemcpy dev_idata failed!"); + timer().startGpuTimer(); + int depth = ilog2ceil(n); + int bound = 1; + for (int d = 1; d <= depth; d++) { + kernNaiveScan << > > (n, bound, dev_odata, dev_idata); + checkCUDAError("kernNaiveScan failed!"); + std::swap(dev_odata, dev_idata); + bound *= 2; + } + kernInclusiveToExclusive << > > (n, dev_odata, dev_idata); + checkCUDAError("kernInclusiveToExclusive failed!"); + cudaMemcpy(odata, dev_odata, n * sizeof(int), cudaMemcpyDeviceToHost); + checkCUDAError("cudaMemcpy dev_odata failed!"); + timer().endGpuTimer(); + cudaFree(dev_odata); + cudaFree(dev_idata); + cudaDeviceSynchronize(); + + } + + } +} diff --git a/stream_compaction/naive.h b/stream_compaction/naive.h new file mode 100644 index 0000000..37dcb06 --- /dev/null +++ b/stream_compaction/naive.h @@ -0,0 +1,11 @@ +#pragma once + +#include "common.h" + +namespace StreamCompaction { + namespace Naive { + StreamCompaction::Common::PerformanceTimer& timer(); + + void scan(int n, int *odata, const int *idata); + } +} diff --git a/stream_compaction/thrust.cu b/stream_compaction/thrust.cu new file mode 100644 index 0000000..e49819d --- /dev/null +++ b/stream_compaction/thrust.cu @@ -0,0 +1,36 @@ +#include +#include +#include +#include +#include +#include "common.h" +#include "thrust.h" + +namespace StreamCompaction { + namespace Thrust { + using StreamCompaction::Common::PerformanceTimer; + PerformanceTimer& timer() + { + static PerformanceTimer timer; + return timer; + } + /** + * Performs prefix-sum (aka scan) on idata, storing the result into odata. + */ + void scan(int n, int *odata, const int *idata) { + + // TODO use `thrust::exclusive_scan` + // example: for device_vectors dv_in and dv_out: + // thrust::exclusive_scan(dv_in.begin(), dv_in.end(), dv_out.begin()); + + thrust::device_vector dev_idata(idata, idata + n); + thrust::device_vector dev_odata(odata, odata + n); + + timer().startGpuTimer(); + //thrust::exclusive_scan(idata, idata + n, odata); + thrust::exclusive_scan(dev_idata.begin(), dev_idata.end(), dev_odata.begin()); + timer().endGpuTimer(); + thrust::copy(dev_odata.begin(), dev_odata.end(), odata); + } + } +} diff --git a/stream_compaction/thrust.h b/stream_compaction/thrust.h new file mode 100644 index 0000000..fe98206 --- /dev/null +++ b/stream_compaction/thrust.h @@ -0,0 +1,11 @@ +#pragma once + +#include "common.h" + +namespace StreamCompaction { + namespace Thrust { + StreamCompaction::Common::PerformanceTimer& timer(); + + void scan(int n, int *odata, const int *idata); + } +}