Add document on schema validation feature.

Now schema validation is getting to work.

LICENSE.rapidjson.txt

@@ -0,0 +1,13 @@
+// Tencent is pleased to support the open source community by making RapidJSON available.
+//
+// Copyright (C) 2015 THL A29 Limited, a Tencent company, and Milo Yip. All rights reserved.
+//
+// Licensed under the MIT License (the "License"); you may not use this file except
+// in compliance with the License. You may obtain a copy of the License at
+//
+// http://opensource.org/licenses/MIT
+//
+// Unless required by applicable law or agreed to in writing, software distributed
+// under the License is distributed on an "AS IS" BASIS, WITHOUT WARRANTIES OR
+// CONDITIONS OF ANY KIND, either express or implied. See the License for the
+// specific language governing permissions and limitations under the License.

Makefile

@@ -2,8 +2,12 @@
 # Use this for strict compilation check(will work on clang 3.8+)
 #EXTRA_CXXFLAGS := -fsanitize=address -Wall -Werror -Weverything -Wno-c++11-long-long -Wno-c++98-compat
 
+# With draco
+# EXTRA_CXXFLAGS := -I../draco/src/ -I../draco/build -DTINYGLTF_ENABLE_DRACO -L../draco/build
+# EXTRA_LINKFLAGS := -L../draco/build/ -ldracodec -ldraco
+
 all:
-	clang++  $(EXTRA_CXXFLAGS) -std=c++11 -g -O0 -o loader_example loader_example.cc
+	clang++  $(EXTRA_CXXFLAGS) -std=c++11 -g -O0 -o loader_example loader_example.cc $(EXTRA_LINKFLAGS)
 
 lint:
 	deps/cpplint.py tiny_gltf.h

README.md

@@ -2,12 +2,14 @@
 
 `TinyGLTF` is a header only C++11 glTF 2.0 https://github.com/KhronosGroup/glTF library.
 
-`TinyGLTF` uses Niels Lohmann's json library(https://github.com/nlohmann/json), so now it requires C++11 compiler.
-If you are looking for old, C++03 version, please use `devel-picojson` branch.  
+`TinyGLTF` uses single-header version of RapidJson(https://github.com/Tencent/rapidjson) as a JSON parser.
 
 ## Status
 
-v2.0.0 release(22 Aug, 2018)! 
+ - v2.3.0 release(Support built-in schema validation)
+ - v2.2.0 release(Support loading 16bit PNG. Sparse accessor support)
+ - v2.1.0 release(Draco support)
+ - v2.0.0 release(22 Aug, 2018)!
 
 ## Builds
 
@@ -32,20 +34,24 @@ v2.0.0 release(22 Aug, 2018)!
   * [x] ASCII glTF
   * [x] Binary glTF(GLB)
   * [x] PBR material description
+  * [x] Validate inpit glTF JSON with schema
 * Buffers
   * [x] Parse BASE64 encoded embedded buffer data(DataURI).
   * [x] Load `.bin` file.
 * Image(Using stb_image)
   * [x] Parse BASE64 encoded embedded image data(DataURI).
   * [x] Load external image file.
-  * [x] PNG(8bit only)
-  * [x] JPEG(8bit only)
-  * [x] BMP
-  * [x] GIF
+  * [x] Load PNG(8bit and 16bit)
+  * [x] Load JPEG(8bit only)
+  * [x] Load BMP
+  * [x] Load GIF
   * [x] Custom Image decoder callback(e.g. for decoding OpenEXR image)
-* Load from memory
+* Morph traget
+  * [x] Sparse accessor
+* Load glTF from memory
 * Custom callback handler
   * [x] Image load
+  * [x] Image save
 * Extensions
   * [x] Draco mesh decoding
 
@@ -63,18 +69,23 @@ v2.0.0 release(22 Aug, 2018)!
 * [TinyGltfImporter](http://doc.magnum.graphics/magnum/classMagnum_1_1Trade_1_1TinyGltfImporter.html) plugin for [Magnum](https://github.com/mosra/magnum), a lightweight and modular C++11/C++14 graphics middleware for games and data visualization.
 * Your projects here! (Please send PR)
 
+## For developer
+
+Generate single rapidjson file using this node.js script: https://github.com/Tencent/rapidjson/issues/863
+Add `cursorstreamwrapper.h` and `error/en.h` inclusion in `rapidjson-all.h` before running merge script.
+
 ## TODOs
 
 * [ ] Write C++ code generator which emits C++ code from JSON schema for robust parsing.
 * [ ] Mesh Compression/decompression(Open3DGC, etc)
   * [x] Load Draco compressed mesh
-  * [x] Save Draco compressed mesh
+  * [ ] Save Draco compressed mesh
+  * [ ] Open3DGC?
 * [ ] Support `extensions` and `extras` property
 * [ ] HDR image?
   * [ ] OpenEXR extension through TinyEXR.
-* [ ] Write example and tests for `animation` and `skin` 
-  * [ ] Skinning
-  * [ ] Morph targets
+* [ ] 16bit PNG support in Serialization
+* [ ] Write example and tests for `animation` and `skin`
 
 ## Licenses
 
@@ -90,7 +101,7 @@ TinyGLTF uses the following third party libraries.
 
 ## Build and example
 
-Copy `stb_image.h`, `stb_image_write.h`, `json.hpp` and `tiny_gltf.h` to your project.
+Copy `stb_image.h`, `stb_image_write.h`, `rapidjson-amalgamation.h` and `tiny_gltf.h` to your project.
 
 ### Loading glTF 2.0 model
 
@@ -99,18 +110,23 @@ Copy `stb_image.h`, `stb_image_write.h`, `json.hpp` and `tiny_gltf.h` to your pr
 #define TINYGLTF_IMPLEMENTATION
 #define STB_IMAGE_IMPLEMENTATION
 #define STB_IMAGE_WRITE_IMPLEMENTATION
+// #define TINYGLTF_ENABLE_SCHEMA_VALIDATOR // optional. enable schema validation API.
 // #define TINYGLTF_NOEXCEPTION // optional. disable exception handling.
 #include "tiny_gltf.h"
 
 using namespace tinygltf;
 
-Model model; 
+Model model;
 TinyGLTF loader;
 std::string err;
 std::string warn;
-  
+
 bool ret = loader.LoadASCIIFromFile(&model, &err, &warn, argv[1]);
-//bool ret = loader.LoadBinaryFromFile(&model, &err, &warn, argv[1]); // for binary glTF(.glb) 
+//bool ret = loader.LoadBinaryFromFile(&model, &err, &warn, argv[1]); // for binary glTF(.glb)
+// Validate with glTF Schema
+// #if defined(TINYGLTF_ENABLE_SCHEMA_VALIDATOR)
+// bool ret = loader.LoadASCIIFromFileWithValidation(&model, &err, &warn, argv[1]);
+// #endif
 
 if (!warn.empty()) {
   printf("Warn: %s\n", warn.c_str());
@@ -128,12 +144,17 @@ if (!ret) {
 
 ## Compile options
 
+* `TINYGLTF_ENABLE_SCHEMA_VALIDATOR` : Enable API with schema validation. glTF Schema JSON(`gltf.schema.resolved.inc`) are embeded into application binary, thus no need to read glTF Schema file at a runtime.
 * `TINYGLTF_NOEXCEPTION` : Disable C++ exception in JSON parsing. You can use `-fno-exceptions` or by defining the symbol `JSON_NOEXCEPTION` and `TINYGLTF_NOEXCEPTION`  to fully remove C++ exception codes when compiling TinyGLTF.
 * `TINYGLTF_NO_STB_IMAGE` : Do not load images with stb_image. Instead use `TinyGLTF::SetImageLoader(LoadimageDataFunction LoadImageData, void *user_data)` to set a callback for loading images.
 * `TINYGLTF_NO_STB_IMAGE_WRITE` : Do not write images with stb_image_write. Instead use `TinyGLTF::SetImageWriter(WriteimageDataFunction WriteImageData, void *user_data)` to set a callback for writing images.
-* `TINYGLTF_NO_EXTERNAL_IMAGE` : Do not try to load external image file. This option woulde be helpful if you do not want load image file during glTF parsing.
+* `TINYGLTF_NO_EXTERNAL_IMAGE` : Do not try to load external image file. This option would be helpful if you do not want to load image files during glTF parsing.
 * `TINYGLTF_ANDROID_LOAD_FROM_ASSETS`: Load all files from packaged app assets instead of the regular file system. **Note:** You must pass a valid asset manager from your android app to `tinygltf::asset_manager` beforehand.
 * `TINYGLTF_ENABLE_DRACO`: Enable Draco compression. User must provide include path and link correspnding libraries in your project file.
+* `TINYGLTF_NO_INCLUDE_JSON `: Disable including `json.hpp` from within `tiny_gltf.h` because it has been already included before or you want to include it using custom path before including `tiny_gltf.h`.
+* `TINYGLTF_NO_INCLUDE_STB_IMAGE `: Disable including `stb_image.h` from within `tiny_gltf.h` because it has been already included before or you want to include it using custom path before including `tiny_gltf.h`.
+* `TINYGLTF_NO_INCLUDE_STB_IMAGE_WRITE `: Disable including `stb_image_write.h` from within `tiny_gltf.h` because it has been already included before or you want to include it using custom path before including `tiny_gltf.h`.
+
 
 ### Saving gltTF 2.0 model
 * [ ] Buffers.

examples/basic/main.cpp

@@ -50,8 +50,9 @@ std::map<int, GLuint> bindMesh(std::map<int, GLuint> vbos,
                    From spec2.0 readme:
                    https://github.com/KhronosGroup/glTF/tree/master/specification/2.0
                             ... drawArrays function should be used with a count equal to
-                   the count            property of any of the accessors referenced by the attributes
-                   property            (they are all equal for a given primitive).
+                   the count            property of any of the accessors referenced by the
+                   attributes            property            (they are all equal for a given
+                   primitive).
                  */
     }
 
@@ -112,8 +113,29 @@ std::map<int, GLuint> bindMesh(std::map<int, GLuint> vbos,
     glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_S, GL_REPEAT);
     glTexParameteri(GL_TEXTURE_2D, GL_TEXTURE_WRAP_T, GL_REPEAT);
 
+    GLenum format = GL_RGBA;
+
+    if (image.component == 1) {
+      format = GL_RED;
+    } else if (image.component == 2) {
+      format = GL_RG;
+    } else if (image.component == 3) {
+      format = GL_RGB;
+    } else {
+      // ???
+    }
+
+    GLenum type = GL_UNSIGNED_BYTE;
+    if (image.bits == 8) {
+      // ok
+    } else if (image.bits == 16) {
+      type = GL_UNSIGNED_SHORT;
+    } else {
+      // ???
+    }
+
     glTexImage2D(GL_TEXTURE_2D, 0, GL_RGBA, image.width, image.height, 0,
-                 GL_RGBA, GL_UNSIGNED_BYTE, &image.image.at(0));
+                 format, type, &image.image.at(0));
   }
 
   return vbos;
@@ -255,7 +277,7 @@ void displayLoop(Window &window, const std::string &filename) {
   glm::vec3 model_pos = glm::vec3(-3, 0, -3);
 
   // generate a camera view, based on eye-position and lookAt world-position
-  glm::mat4 view_mat = genView(glm::vec3(2, 2, 2), model_pos);
+  glm::mat4 view_mat = genView(glm::vec3(2, 2, 20), model_pos);
 
   glm::vec3 sun_position = glm::vec3(3.0, 10.0, -5.0);
   glm::vec3 sun_color = glm::vec3(1.0);
@@ -287,6 +309,11 @@ void displayLoop(Window &window, const std::string &filename) {
   }
 }
 
+static void error_callback(int error, const char *description) {
+  (void)error;
+  fprintf(stderr, "Error: %s\n", description);
+}
+
 int main(int argc, char **argv) {
   std::string filename = "../../models/Cube/Cube.gltf";
 
@@ -294,14 +321,18 @@ int main(int argc, char **argv) {
     filename = argv[1];
   }
 
+  glfwSetErrorCallback(error_callback);
+
   if (!glfwInit()) return -1;
 
-  // Force create 3.3 profile.
+  // Force create OpenGL 3.3
+  // NOTE(syoyo): Linux + NVIDIA driver segfaults for some reason? commenting out glfwWindowHint will work.
   glfwWindowHint(GLFW_CONTEXT_VERSION_MAJOR, 3);
   glfwWindowHint(GLFW_CONTEXT_VERSION_MINOR, 3);
   glfwWindowHint(GLFW_OPENGL_PROFILE, GLFW_OPENGL_CORE_PROFILE);
 #ifdef __APPLE__
   glfwWindowHint(GLFW_OPENGL_FORWARD_COMPAT, GL_TRUE);
+#endif
 #endif
 
   Window window = Window(800, 600, "TinyGLTF basic example");

examples/common/LICENSE.lodepng

@@ -0,0 +1,21 @@
+Copyright (c) 2005-2018 Lode Vandevenne
+
+This software is provided 'as-is', without any express or implied
+warranty. In no event will the authors be held liable for any damages
+arising from the use of this software.
+
+Permission is granted to anyone to use this software for any purpose,
+including commercial applications, and to alter it and redistribute it
+freely, subject to the following restrictions:
+
+    1. The origin of this software must not be misrepresented; you must not
+    claim that you wrote the original software. If you use this software
+    in a product, an acknowledgment in the product documentation would be
+    appreciated but is not required.
+
+    2. Altered source versions must be plainly marked as such, and must not be
+    misrepresented as being the original software.
+
+    3. This notice may not be removed or altered from any source
+    distribution.
+    

examples/gltfutil/CMakeLists.txt

@@ -4,9 +4,10 @@ project(gltfutil)
 set(CMAKE_CXX_STANDARD 11)
 
 include_directories(../../)
+include_directories(../common/)
 
 file(GLOB gltfutil_sources *.cc *.h)
-add_executable(gltfutil ${gltfutil_sources})
+add_executable(gltfutil ${gltfutil_sources} ../common/lodepng.cpp)
 
 install ( TARGETS
   gltfutil

examples/gltfutil/gltfuilconfig.h

@@ -49,6 +49,7 @@ struct configuration {
   cli_action action = cli_action::not_set;
   texture_dumper::texture_output_format requested_format =
       texture_dumper::texture_output_format::not_specified;
+  bool use_exr = false;
 
   bool has_output_dir;
   bool is_valid() {

examples/gltfutil/main.cc

@@ -11,6 +11,9 @@
 #define STB_IMAGE_WRITE_IMPLEMENTATION
 #include "stb_image_write.h"
 
+#define TINYEXR_IMPLEMENTATION
+#include "tinyexr.h"
+
 namespace gltfutil {
 int usage(int ret = 0) {
   using std::cout;
@@ -20,8 +23,9 @@ int usage(int ret = 0) {
           "[path to output directory])\n\n"
        //<< "\t\t -i: start in interactive mode\n"
        << "\t\t -d: dump enclosed content (image assets)\n"
-       << "\t\t -f: file format for image output"
-       << "\t\t -o: ouptput directory path"
+       << "\t\t -f: file format for image output\n"
+       << "\t\t -o: ouptput directory path\n"
+       << "\t\t -e: Use OpenEXR format for 16bit image\n"
        << "\t\t -h: print this help\n";
   return ret;
 }
@@ -44,6 +48,9 @@ int parse_args(int argc, char** argv) {
           config.mode = ui_mode::cli;
           config.action = cli_action::dump;
           break;
+        case 'e':
+          config.use_exr = true;
+          break;
         case 'i':
           config.mode = ui_mode::interactive;
           break;
@@ -97,6 +104,11 @@ int parse_args(int argc, char** argv) {
 
         case cli_action::dump: {
           texture_dumper dumper(model);
+
+          if (config.use_exr) {
+            dumper.set_use_exr(true);
+          }
+
           if (config.requested_format !=
               texture_dumper::texture_output_format::not_specified)
             dumper.set_output_format(config.requested_format);

examples/gltfutil/texture_dumper.cc

@@ -1,15 +1,84 @@
-#include <iostream>
 #include <algorithm>
+#include <iostream>
 
 #include "stb_image_write.h"
 #include "texture_dumper.h"
 
+#include "lodepng.h"  // ../common
+
+#include "tinyexr.h"
+
 #include <tiny_gltf.h>
 
 using namespace gltfutil;
 using namespace tinygltf;
 using std::cout;
 
+static LodePNGColorType GetLodePNGColorType(int channels) {
+  if (channels == 1) {
+    return LodePNGColorType::LCT_GREY;
+  } else if (channels == 2) {
+    return LodePNGColorType::LCT_GREY_ALPHA;
+  } else if (channels == 3) {
+    return LodePNGColorType::LCT_RGB;
+  } else if (channels == 4) {
+    return LodePNGColorType::LCT_RGBA;
+  } else {
+    std::cerr << "??? unsupported channels " << channels << "\n";
+    return LodePNGColorType::LCT_RGB;  // FIXME(syoyo): Raise error
+  }
+}
+
+static void ToBigEndian(std::vector<uint8_t>* image) {
+
+  assert(image->size() % 2 == 0);
+
+  union {
+    unsigned int i;
+    char c[4];
+  } bint = {0x01020304};
+
+  bool is_big_endian = (bint.c[0] == 1);
+
+  if (is_big_endian) {
+    return;
+  }
+
+  uint16_t *ptr = reinterpret_cast<uint16_t *>(image->data());
+  size_t n = image->size() / 2;
+
+  for (size_t i = 0; i < n; i++) {
+    ptr[i] = ((0xFF00 & ptr[i]) >> 8) | ((0x00FF & ptr[i]) << 8);
+  }
+}
+
+static bool Save16bitImageAsEXR(const std::string& filename,
+                                const tinygltf::Image& image) {
+  assert(image.bits == 16);
+
+  std::vector<float> buf(image.width * image.height * image.component);
+
+  // widen to float image.
+  // Store as is(i.e, pixel value range is [0.0, 65535.0])
+  const unsigned short* ptr =
+      reinterpret_cast<const unsigned short*>(image.image.data());
+  for (size_t i = 0; i < image.width * image.height * image.component; i++) {
+    buf[i] = float(ptr[i]);
+  }
+
+  const char* err = nullptr;
+  int ret = SaveEXR(buf.data(), image.width, image.height, image.component,
+                    /* save_as_fp16 */ 0, filename.c_str(), &err);
+
+  if (err) {
+    std::cerr << "EXR err: " << err << std::endl;
+    FreeEXRErrorMessage(err);
+    return false;
+  }
+
+  return (ret == TINYEXR_SUCCESS);
+}
+
 texture_dumper::texture_dumper(const Model& input)
     : model(input), configured_format(texture_output_format::png) {
   cout << "Texture dumper\n";
@@ -26,26 +95,58 @@ void texture_dumper::dump_to_folder(const std::string& path) {
     cout << "image name is: \"" << image.name << "\"\n";
     cout << "image size is: " << image.width << 'x' << image.height << '\n';
     cout << "pixel channel count :" << image.component << '\n';
-    std::string name = image.name.empty() ? std::to_string(index) : image.name;
+    cout << "pixel bit depth :" << image.bits << '\n';
+    std::string basename =
+        image.name.empty() ? std::to_string(index) : image.name;
 
+    unsigned char* bytes_to_write =
+        const_cast<unsigned char*>(image.image.data());
+
+    std::string filename;
     switch (configured_format) {
       case texture_output_format::png:
-        name = path + "/" + name + ".png";
-        std::cout << "Image will be written to " << name << '\n';
-        stbi_write_png(name.c_str(), image.width, image.height, image.component,
-                       image.image.data(), 0);
+        filename = path + "/" + basename + ".png";
+
+        if (this->use_exr) {
+          if (image.pixel_type == TINYGLTF_COMPONENT_TYPE_UNSIGNED_SHORT) {
+            filename = path + "/" + basename + ".exr";
+          }
+        }
+
+        std::cout << "Image will be written to " << filename << '\n';
+
+        if (image.pixel_type == TINYGLTF_COMPONENT_TYPE_UNSIGNED_SHORT) {
+          if (this->use_exr) {
+            bool ret = Save16bitImageAsEXR(filename, image);
+            assert(ret);
+          } else {
+            // Use lodepng to save 16bit PNG.
+            // NOTE(syoyo): `loadpng::encode` requires image data must be stored in big endian.
+            std::vector<uint8_t> tmp = image.image; // copy
+            ToBigEndian(&tmp);
+
+            unsigned ret = lodepng::encode(
+                filename, tmp.data(), image.width, image.height,
+                GetLodePNGColorType(image.component), /* bits */ 16);
+            assert(ret == 0);  // 0 = no err.
+          }
+        } else {
+          // TODO(syoyo): check status
+          stbi_write_png(filename.c_str(), image.width, image.height,
+                         image.component, bytes_to_write, 0);
+        }
         break;
       case texture_output_format::bmp:
-        std::cout << "Image will be written to " << name << '\n';
-        name = path + "/" + name + ".bmp";
-        stbi_write_bmp(name.c_str(), image.width, image.height, image.component,
-                       image.image.data());
+        filename = path + "/" + basename + ".bmp";
+        std::cout << "Image will be written to " << filename << '\n';
+        stbi_write_bmp(filename.c_str(), image.width, image.height,
+                       image.component, bytes_to_write);
         break;
       case texture_output_format::tga:
-        std::cout << "Image will be written to " << name << '\n';
-        name = path + "/" + name + ".tga";
-        stbi_write_tga(name.c_str(), image.width, image.height, image.component,
-                       image.image.data());
+        filename = path + "/" + basename + ".tga";
+        std::cout << "Image will be written to " << filename << '\n';
+        stbi_write_tga(filename.c_str(), image.width, image.height,
+                       image.component, bytes_to_write);
         break;
     }
   }

examples/gltfutil/texture_dumper.h

@@ -12,11 +12,15 @@ class texture_dumper {
  private:
   const tinygltf::Model& model;
   texture_output_format configured_format;
+  bool use_exr = false; // Use EXR for 16bit image?
 
  public:
   texture_dumper(const tinygltf::Model& inputModel);
   void dump_to_folder(const std::string& path = "./");
   void set_output_format(texture_output_format format);
+  void set_use_exr(const bool value) {
+    use_exr = value;
+  }
 
   static texture_output_format get_fromat_from_string(const std::string& str);
 };

examples/glview/glview.cc

@@ -28,7 +28,6 @@
 #include "tiny_gltf.h"
 #endif
 
-
 #define BUFFER_OFFSET(i) ((char *)NULL + (i))
 
 #define CheckGLErrors(desc)                                                   \
@@ -55,7 +54,9 @@ float eye[3], lookat[3], up[3];
 
 GLFWwindow *window;
 
-typedef struct { GLuint vb; } GLBufferState;
+typedef struct {
+  GLuint vb;
+} GLBufferState;
 
 typedef struct {
   std::vector<GLuint> diffuseTex;  // for each primitive in mesh
@@ -254,6 +255,26 @@ void motionFunc(GLFWwindow *window, double mouse_x, double mouse_y) {
   prevMouseY = mouse_y;
 }
 
+static size_t ComponentTypeByteSize(int type) {
+  switch (type) {
+    case TINYGLTF_COMPONENT_TYPE_UNSIGNED_BYTE:
+    case TINYGLTF_COMPONENT_TYPE_BYTE:
+      return sizeof(char);
+    case TINYGLTF_COMPONENT_TYPE_UNSIGNED_SHORT:
+    case TINYGLTF_COMPONENT_TYPE_SHORT:
+      return sizeof(short);
+    case TINYGLTF_COMPONENT_TYPE_UNSIGNED_INT:
+    case TINYGLTF_COMPONENT_TYPE_INT:
+      return sizeof(int);
+    case TINYGLTF_COMPONENT_TYPE_FLOAT:
+      return sizeof(float);
+    case TINYGLTF_COMPONENT_TYPE_DOUBLE:
+      return sizeof(double);
+    default:
+      return 0;
+  }
+}
+
 static void SetupMeshState(tinygltf::Model &model, GLuint progId) {
   // Buffer
   {
@@ -264,14 +285,117 @@ static void SetupMeshState(tinygltf::Model &model, GLuint progId) {
         continue;  // Unsupported bufferView.
       }
 
+      int sparse_accessor = -1;
+      for (size_t a_i = 0; a_i < model.accessors.size(); ++a_i) {
+        const auto &accessor = model.accessors[a_i];
+        if (accessor.bufferView == i) {
+          std::cout << i << " is used by accessor " << a_i << std::endl;
+          if (accessor.sparse.isSparse) {
+            std::cout
+                << "WARN: this bufferView has at least one sparse accessor to "
+                   "it. We are going to load the data as patched by this "
+                   "sparse accessor, not the original data"
+                << std::endl;
+            sparse_accessor = a_i;
+            break;
+          }
+        }
+      }
+
       const tinygltf::Buffer &buffer = model.buffers[bufferView.buffer];
       GLBufferState state;
       glGenBuffers(1, &state.vb);
       glBindBuffer(bufferView.target, state.vb);
       std::cout << "buffer.size= " << buffer.data.size()
                 << ", byteOffset = " << bufferView.byteOffset << std::endl;
-      glBufferData(bufferView.target, bufferView.byteLength,
-                   &buffer.data.at(0) + bufferView.byteOffset, GL_STATIC_DRAW);
+
+      if (sparse_accessor < 0)
+        glBufferData(bufferView.target, bufferView.byteLength,
+                     &buffer.data.at(0) + bufferView.byteOffset,
+                     GL_STATIC_DRAW);
+      else {
+        const auto accessor = model.accessors[sparse_accessor];
+        // copy the buffer to a temporary one for sparse patching
+        unsigned char *tmp_buffer = new unsigned char[bufferView.byteLength];
+        memcpy(tmp_buffer, buffer.data.data() + bufferView.byteOffset,
+               bufferView.byteLength);
+
+        const size_t size_of_object_in_buffer =
+            ComponentTypeByteSize(accessor.componentType);
+        const size_t size_of_sparse_indices =
+            ComponentTypeByteSize(accessor.sparse.indices.componentType);
+
+        const auto &indices_buffer_view =
+            model.bufferViews[accessor.sparse.indices.bufferView];
+        const auto &indices_buffer = model.buffers[indices_buffer_view.buffer];
+
+        const auto &values_buffer_view =
+            model.bufferViews[accessor.sparse.values.bufferView];
+        const auto &values_buffer = model.buffers[values_buffer_view.buffer];
+
+        for (size_t sparse_index = 0; sparse_index < accessor.sparse.count;
+             ++sparse_index) {
+          int index = 0;
+          // std::cout << "accessor.sparse.indices.componentType = " <<
+          // accessor.sparse.indices.componentType << std::endl;
+          switch (accessor.sparse.indices.componentType) {
+            case TINYGLTF_COMPONENT_TYPE_BYTE:
+            case TINYGLTF_COMPONENT_TYPE_UNSIGNED_BYTE:
+              index = (int)*(
+                  unsigned char *)(indices_buffer.data.data() +
+                                   indices_buffer_view.byteOffset +
+                                   accessor.sparse.indices.byteOffset +
+                                   (sparse_index * size_of_sparse_indices));
+              break;
+            case TINYGLTF_COMPONENT_TYPE_SHORT:
+            case TINYGLTF_COMPONENT_TYPE_UNSIGNED_SHORT:
+              index = (int)*(
+                  unsigned short *)(indices_buffer.data.data() +
+                                    indices_buffer_view.byteOffset +
+                                    accessor.sparse.indices.byteOffset +
+                                    (sparse_index * size_of_sparse_indices));
+              break;
+            case TINYGLTF_COMPONENT_TYPE_INT:
+            case TINYGLTF_COMPONENT_TYPE_UNSIGNED_INT:
+              index = (int)*(
+                  unsigned int *)(indices_buffer.data.data() +
+                                  indices_buffer_view.byteOffset +
+                                  accessor.sparse.indices.byteOffset +
+                                  (sparse_index * size_of_sparse_indices));
+              break;
+          }
+          std::cout << "updating sparse data at index  : " << index
+                    << std::endl;
+          // index is now the target of the sparse index to patch in
+          const unsigned char *read_from =
+              values_buffer.data.data() +
+              (values_buffer_view.byteOffset +
+               accessor.sparse.values.byteOffset) +
+              (sparse_index * (size_of_object_in_buffer * accessor.type));
+
+          /*
+          std::cout << ((float*)read_from)[0] << "\n";
+          std::cout << ((float*)read_from)[1] << "\n";
+          std::cout << ((float*)read_from)[2] << "\n";
+          */
+
+          unsigned char *write_to =
+              tmp_buffer + index * (size_of_object_in_buffer * accessor.type);
+
+          memcpy(write_to, read_from, size_of_object_in_buffer * accessor.type);
+        }
+
+        // debug:
+        /*for(size_t p = 0; p < bufferView.byteLength/sizeof(float); p++)
+        {
+          float* b = (float*)tmp_buffer;
+          std::cout << "modified_buffer [" << p << "] = " << b[p] << '\n';
+        }*/
+
+        glBufferData(bufferView.target, bufferView.byteLength, tmp_buffer,
+                     GL_STATIC_DRAW);
+        delete[] tmp_buffer;
+      }
       glBindBuffer(bufferView.target, 0);
 
       gBufferState[i] = state;
@@ -279,55 +403,55 @@ static void SetupMeshState(tinygltf::Model &model, GLuint progId) {
   }
 
 #if 0  // TODO(syoyo): Implement
-  // Texture
-  {
-    for (size_t i = 0; i < model.meshes.size(); i++) {
-      const tinygltf::Mesh &mesh = model.meshes[i];
+	// Texture
+	{
+		for (size_t i = 0; i < model.meshes.size(); i++) {
+			const tinygltf::Mesh &mesh = model.meshes[i];
 
-      gMeshState[mesh.name].diffuseTex.resize(mesh.primitives.size());
-      for (size_t primId = 0; primId < mesh.primitives.size(); primId++) {
-        const tinygltf::Primitive &primitive = mesh.primitives[primId];
+			gMeshState[mesh.name].diffuseTex.resize(mesh.primitives.size());
+			for (size_t primId = 0; primId < mesh.primitives.size(); primId++) {
+				const tinygltf::Primitive &primitive = mesh.primitives[primId];
 
-        gMeshState[mesh.name].diffuseTex[primId] = 0;
+				gMeshState[mesh.name].diffuseTex[primId] = 0;
 
-        if (primitive.material < 0) {
-          continue;
-        }
-        tinygltf::Material &mat = model.materials[primitive.material];
-        // printf("material.name = %s\n", mat.name.c_str());
-        if (mat.values.find("diffuse") != mat.values.end()) {
-          std::string diffuseTexName = mat.values["diffuse"].string_value;
-          if (model.textures.find(diffuseTexName) != model.textures.end()) {
-            tinygltf::Texture &tex = model.textures[diffuseTexName];
-            if (scene.images.find(tex.source) != model.images.end()) {
-              tinygltf::Image &image = model.images[tex.source];
-              GLuint texId;
-              glGenTextures(1, &texId);
-              glBindTexture(tex.target, texId);
-              glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
-              glTexParameterf(tex.target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-              glTexParameterf(tex.target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+				if (primitive.material < 0) {
+					continue;
+				}
+				tinygltf::Material &mat = model.materials[primitive.material];
+				// printf("material.name = %s\n", mat.name.c_str());
+				if (mat.values.find("diffuse") != mat.values.end()) {
+					std::string diffuseTexName = mat.values["diffuse"].string_value;
+					if (model.textures.find(diffuseTexName) != model.textures.end()) {
+						tinygltf::Texture &tex = model.textures[diffuseTexName];
+						if (scene.images.find(tex.source) != model.images.end()) {
+							tinygltf::Image &image = model.images[tex.source];
+							GLuint texId;
+							glGenTextures(1, &texId);
+							glBindTexture(tex.target, texId);
+							glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
+							glTexParameterf(tex.target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+							glTexParameterf(tex.target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
 
-              // Ignore Texture.fomat.
-              GLenum format = GL_RGBA;
-              if (image.component == 3) {
-                format = GL_RGB;
-              }
-              glTexImage2D(tex.target, 0, tex.internalFormat, image.width,
-                           image.height, 0, format, tex.type,
-                           &image.image.at(0));
+							// Ignore Texture.fomat.
+							GLenum format = GL_RGBA;
+							if (image.component == 3) {
+								format = GL_RGB;
+							}
+							glTexImage2D(tex.target, 0, tex.internalFormat, image.width,
+									image.height, 0, format, tex.type,
+									&image.image.at(0));
 
-              CheckErrors("texImage2D");
-              glBindTexture(tex.target, 0);
+							CheckErrors("texImage2D");
+							glBindTexture(tex.target, 0);
 
-              printf("TexId = %d\n", texId);
-              gMeshState[mesh.name].diffuseTex[primId] = texId;
-            }
-          }
-        }
-      }
-    }
-  }
+							printf("TexId = %d\n", texId);
+							gMeshState[mesh.name].diffuseTex[primId] = texId;
+						}
+					}
+				}
+			}
+		}
+	}
 #endif
 
   glUseProgram(progId);
@@ -348,164 +472,164 @@ static void SetupMeshState(tinygltf::Model &model, GLuint progId) {
 #if 0  // TODO(syoyo): Implement
 // Setup curves geometry extension
 static void SetupCurvesState(tinygltf::Scene &scene, GLuint progId) {
-  // Find curves primitive.
-  {
-    std::map<std::string, tinygltf::Mesh>::const_iterator it(
-        scene.meshes.begin());
-    std::map<std::string, tinygltf::Mesh>::const_iterator itEnd(
-        scene.meshes.end());
+	// Find curves primitive.
+	{
+		std::map<std::string, tinygltf::Mesh>::const_iterator it(
+				scene.meshes.begin());
+		std::map<std::string, tinygltf::Mesh>::const_iterator itEnd(
+				scene.meshes.end());
 
-    for (; it != itEnd; it++) {
-      const tinygltf::Mesh &mesh = it->second;
+		for (; it != itEnd; it++) {
+			const tinygltf::Mesh &mesh = it->second;
 
-      // Currently we only support one primitive per mesh.
-      if (mesh.primitives.size() > 1) {
-        continue;
-      }
+			// Currently we only support one primitive per mesh.
+			if (mesh.primitives.size() > 1) {
+				continue;
+			}
 
-      for (size_t primId = 0; primId < mesh.primitives.size(); primId++) {
-        const tinygltf::Primitive &primitive = mesh.primitives[primId];
+			for (size_t primId = 0; primId < mesh.primitives.size(); primId++) {
+				const tinygltf::Primitive &primitive = mesh.primitives[primId];
 
-        gMeshState[mesh.name].diffuseTex[primId] = 0;
+				gMeshState[mesh.name].diffuseTex[primId] = 0;
 
-        if (primitive.material.empty()) {
-          continue;
-        }
+				if (primitive.material.empty()) {
+					continue;
+				}
 
-        bool has_curves = false;
-        if (primitive.extras.IsObject()) {
-          if (primitive.extras.Has("ext_mode")) {
-            const tinygltf::Value::Object &o =
-                primitive.extras.Get<tinygltf::Value::Object>();
-            const tinygltf::Value &ext_mode = o.find("ext_mode")->second;
+				bool has_curves = false;
+				if (primitive.extras.IsObject()) {
+					if (primitive.extras.Has("ext_mode")) {
+						const tinygltf::Value::Object &o =
+							primitive.extras.Get<tinygltf::Value::Object>();
+						const tinygltf::Value &ext_mode = o.find("ext_mode")->second;
 
-            if (ext_mode.IsString()) {
-              const std::string &str = ext_mode.Get<std::string>();
-              if (str.compare("curves") == 0) {
-                has_curves = true;
-              }
-            }
-          }
-        }
+						if (ext_mode.IsString()) {
+							const std::string &str = ext_mode.Get<std::string>();
+							if (str.compare("curves") == 0) {
+								has_curves = true;
+							}
+						}
+					}
+				}
 
-        if (!has_curves) {
-          continue;
-        }
+				if (!has_curves) {
+					continue;
+				}
 
-        // Construct curves buffer
-        const tinygltf::Accessor &vtx_accessor =
-            scene.accessors[primitive.attributes.find("POSITION")->second];
-        const tinygltf::Accessor &nverts_accessor =
-            scene.accessors[primitive.attributes.find("NVERTS")->second];
-        const tinygltf::BufferView &vtx_bufferView =
-            scene.bufferViews[vtx_accessor.bufferView];
-        const tinygltf::BufferView &nverts_bufferView =
-            scene.bufferViews[nverts_accessor.bufferView];
-        const tinygltf::Buffer &vtx_buffer =
-            scene.buffers[vtx_bufferView.buffer];
-        const tinygltf::Buffer &nverts_buffer =
-            scene.buffers[nverts_bufferView.buffer];
+				// Construct curves buffer
+				const tinygltf::Accessor &vtx_accessor =
+					scene.accessors[primitive.attributes.find("POSITION")->second];
+				const tinygltf::Accessor &nverts_accessor =
+					scene.accessors[primitive.attributes.find("NVERTS")->second];
+				const tinygltf::BufferView &vtx_bufferView =
+					scene.bufferViews[vtx_accessor.bufferView];
+				const tinygltf::BufferView &nverts_bufferView =
+					scene.bufferViews[nverts_accessor.bufferView];
+				const tinygltf::Buffer &vtx_buffer =
+					scene.buffers[vtx_bufferView.buffer];
+				const tinygltf::Buffer &nverts_buffer =
+					scene.buffers[nverts_bufferView.buffer];
 
-        // std::cout << "vtx_bufferView = " << vtx_accessor.bufferView <<
-        // std::endl;
-        // std::cout << "nverts_bufferView = " << nverts_accessor.bufferView <<
-        // std::endl;
-        // std::cout << "vtx_buffer.size = " << vtx_buffer.data.size() <<
-        // std::endl;
-        // std::cout << "nverts_buffer.size = " << nverts_buffer.data.size() <<
-        // std::endl;
+				// std::cout << "vtx_bufferView = " << vtx_accessor.bufferView <<
+				// std::endl;
+				// std::cout << "nverts_bufferView = " << nverts_accessor.bufferView <<
+				// std::endl;
+				// std::cout << "vtx_buffer.size = " << vtx_buffer.data.size() <<
+				// std::endl;
+				// std::cout << "nverts_buffer.size = " << nverts_buffer.data.size() <<
+				// std::endl;
 
-        const int *nverts =
-            reinterpret_cast<const int *>(nverts_buffer.data.data());
-        const float *vtx =
-            reinterpret_cast<const float *>(vtx_buffer.data.data());
+				const int *nverts =
+					reinterpret_cast<const int *>(nverts_buffer.data.data());
+				const float *vtx =
+					reinterpret_cast<const float *>(vtx_buffer.data.data());
 
-        // Convert to GL_LINES data.
-        std::vector<float> line_pts;
-        size_t vtx_offset = 0;
-        for (int k = 0; k < static_cast<int>(nverts_accessor.count); k++) {
-          for (int n = 0; n < nverts[k] - 1; n++) {
+				// Convert to GL_LINES data.
+				std::vector<float> line_pts;
+				size_t vtx_offset = 0;
+				for (int k = 0; k < static_cast<int>(nverts_accessor.count); k++) {
+					for (int n = 0; n < nverts[k] - 1; n++) {
 
-            line_pts.push_back(vtx[3 * (vtx_offset + n) + 0]);
-            line_pts.push_back(vtx[3 * (vtx_offset + n) + 1]);
-            line_pts.push_back(vtx[3 * (vtx_offset + n) + 2]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n) + 0]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n) + 1]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n) + 2]);
 
-            line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 0]);
-            line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 1]);
-            line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 2]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 0]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 1]);
+						line_pts.push_back(vtx[3 * (vtx_offset + n + 1) + 2]);
 
-            // std::cout << "p0 " << vtx[3 * (vtx_offset + n) + 0] << ", "
-            //                  << vtx[3 * (vtx_offset + n) + 1] << ", "
-            //                  << vtx[3 * (vtx_offset + n) + 2] << std::endl;
+						// std::cout << "p0 " << vtx[3 * (vtx_offset + n) + 0] << ", "
+						//                  << vtx[3 * (vtx_offset + n) + 1] << ", "
+						//                  << vtx[3 * (vtx_offset + n) + 2] << std::endl;
 
-            // std::cout << "p1 " << vtx[3 * (vtx_offset + n+1) + 0] << ", "
-            //                  << vtx[3 * (vtx_offset + n+1) + 1] << ", "
-            //                  << vtx[3 * (vtx_offset + n+1) + 2] << std::endl;
-          }
+						// std::cout << "p1 " << vtx[3 * (vtx_offset + n+1) + 0] << ", "
+						//                  << vtx[3 * (vtx_offset + n+1) + 1] << ", "
+						//                  << vtx[3 * (vtx_offset + n+1) + 2] << std::endl;
+					}
 
-          vtx_offset += nverts[k];
-        }
+					vtx_offset += nverts[k];
+				}
 
-        GLCurvesState state;
-        glGenBuffers(1, &state.vb);
-        glBindBuffer(GL_ARRAY_BUFFER, state.vb);
-        glBufferData(GL_ARRAY_BUFFER, line_pts.size() * sizeof(float),
-                     line_pts.data(), GL_STATIC_DRAW);
-        glBindBuffer(GL_ARRAY_BUFFER, 0);
+				GLCurvesState state;
+				glGenBuffers(1, &state.vb);
+				glBindBuffer(GL_ARRAY_BUFFER, state.vb);
+				glBufferData(GL_ARRAY_BUFFER, line_pts.size() * sizeof(float),
+						line_pts.data(), GL_STATIC_DRAW);
+				glBindBuffer(GL_ARRAY_BUFFER, 0);
 
-        state.count = line_pts.size() / 3;
-        gCurvesMesh[mesh.name] = state;
+				state.count = line_pts.size() / 3;
+				gCurvesMesh[mesh.name] = state;
 
-        // Material
-        tinygltf::Material &mat = scene.materials[primitive.material];
-        // printf("material.name = %s\n", mat.name.c_str());
-        if (mat.values.find("diffuse") != mat.values.end()) {
-          std::string diffuseTexName = mat.values["diffuse"].string_value;
-          if (scene.textures.find(diffuseTexName) != scene.textures.end()) {
-            tinygltf::Texture &tex = scene.textures[diffuseTexName];
-            if (scene.images.find(tex.source) != scene.images.end()) {
-              tinygltf::Image &image = scene.images[tex.source];
-              GLuint texId;
-              glGenTextures(1, &texId);
-              glBindTexture(tex.target, texId);
-              glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
-              glTexParameterf(tex.target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
-              glTexParameterf(tex.target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
+				// Material
+				tinygltf::Material &mat = scene.materials[primitive.material];
+				// printf("material.name = %s\n", mat.name.c_str());
+				if (mat.values.find("diffuse") != mat.values.end()) {
+					std::string diffuseTexName = mat.values["diffuse"].string_value;
+					if (scene.textures.find(diffuseTexName) != scene.textures.end()) {
+						tinygltf::Texture &tex = scene.textures[diffuseTexName];
+						if (scene.images.find(tex.source) != scene.images.end()) {
+							tinygltf::Image &image = scene.images[tex.source];
+							GLuint texId;
+							glGenTextures(1, &texId);
+							glBindTexture(tex.target, texId);
+							glPixelStorei(GL_UNPACK_ALIGNMENT, 1);
+							glTexParameterf(tex.target, GL_TEXTURE_MIN_FILTER, GL_LINEAR);
+							glTexParameterf(tex.target, GL_TEXTURE_MAG_FILTER, GL_LINEAR);
 
-              // Ignore Texture.fomat.
-              GLenum format = GL_RGBA;
-              if (image.component == 3) {
-                format = GL_RGB;
-              }
-              glTexImage2D(tex.target, 0, tex.internalFormat, image.width,
-                           image.height, 0, format, tex.type,
-                           &image.image.at(0));
+							// Ignore Texture.fomat.
+							GLenum format = GL_RGBA;
+							if (image.component == 3) {
+								format = GL_RGB;
+							}
+							glTexImage2D(tex.target, 0, tex.internalFormat, image.width,
+									image.height, 0, format, tex.type,
+									&image.image.at(0));
 
-              CheckErrors("texImage2D");
-              glBindTexture(tex.target, 0);
+							CheckErrors("texImage2D");
+							glBindTexture(tex.target, 0);
 
-              printf("TexId = %d\n", texId);
-              gMeshState[mesh.name].diffuseTex[primId] = texId;
-            }
-          }
-        }
-      }
-    }
-  }
+							printf("TexId = %d\n", texId);
+							gMeshState[mesh.name].diffuseTex[primId] = texId;
+						}
+					}
+				}
+			}
+		}
+	}
 
-  glUseProgram(progId);
-  GLint vtloc = glGetAttribLocation(progId, "in_vertex");
-  GLint nrmloc = glGetAttribLocation(progId, "in_normal");
-  GLint uvloc = glGetAttribLocation(progId, "in_texcoord");
+	glUseProgram(progId);
+	GLint vtloc = glGetAttribLocation(progId, "in_vertex");
+	GLint nrmloc = glGetAttribLocation(progId, "in_normal");
+	GLint uvloc = glGetAttribLocation(progId, "in_texcoord");
 
-  GLint diffuseTexLoc = glGetUniformLocation(progId, "diffuseTex");
-  GLint isCurvesLoc = glGetUniformLocation(progId, "uIsCurves");
+	GLint diffuseTexLoc = glGetUniformLocation(progId, "diffuseTex");
+	GLint isCurvesLoc = glGetUniformLocation(progId, "uIsCurves");
 
-  gGLProgramState.attribs["POSITION"] = vtloc;
-  gGLProgramState.attribs["NORMAL"] = nrmloc;
-  gGLProgramState.attribs["TEXCOORD_0"] = uvloc;
-  gGLProgramState.uniforms["diffuseTex"] = diffuseTexLoc;
-  gGLProgramState.uniforms["uIsCurves"] = isCurvesLoc;
+	gGLProgramState.attribs["POSITION"] = vtloc;
+	gGLProgramState.attribs["NORMAL"] = nrmloc;
+	gGLProgramState.attribs["TEXCOORD_0"] = uvloc;
+	gGLProgramState.uniforms["diffuseTex"] = diffuseTexLoc;
+	gGLProgramState.uniforms["uIsCurves"] = isCurvesLoc;
 };
 #endif
 
@@ -558,12 +682,13 @@ static void DrawMesh(tinygltf::Model &model, const tinygltf::Mesh &mesh) {
           (it->first.compare("TEXCOORD_0") == 0)) {
         if (gGLProgramState.attribs[it->first] >= 0) {
           // Compute byteStride from Accessor + BufferView combination.
-          int byteStride = accessor.ByteStride(model.bufferViews[accessor.bufferView]);
+          int byteStride =
+              accessor.ByteStride(model.bufferViews[accessor.bufferView]);
           assert(byteStride != -1);
           glVertexAttribPointer(gGLProgramState.attribs[it->first], size,
-                                accessor.componentType, accessor.normalized ? GL_TRUE : GL_FALSE,
-                                byteStride,
-                                BUFFER_OFFSET(accessor.byteOffset));
+                                accessor.componentType,
+                                accessor.normalized ? GL_TRUE : GL_FALSE,
+                                byteStride, BUFFER_OFFSET(accessor.byteOffset));
           CheckErrors("vertex attrib pointer");
           glEnableVertexAttribArray(gGLProgramState.attribs[it->first]);
           CheckErrors("enable vertex attrib array");
@@ -617,32 +742,32 @@ static void DrawMesh(tinygltf::Model &model, const tinygltf::Mesh &mesh) {
 
 #if 0  // TODO(syoyo): Implement
 static void DrawCurves(tinygltf::Scene &scene, const tinygltf::Mesh &mesh) {
-  (void)scene;
+	(void)scene;
 
-  if (gCurvesMesh.find(mesh.name) == gCurvesMesh.end()) {
-    return;
-  }
+	if (gCurvesMesh.find(mesh.name) == gCurvesMesh.end()) {
+		return;
+	}
 
-  if (gGLProgramState.uniforms["isCurvesLoc"] >= 0) {
-    glUniform1i(gGLProgramState.uniforms["isCurvesLoc"], 1);
-  }
+	if (gGLProgramState.uniforms["isCurvesLoc"] >= 0) {
+		glUniform1i(gGLProgramState.uniforms["isCurvesLoc"], 1);
+	}
 
-  GLCurvesState &state = gCurvesMesh[mesh.name];
+	GLCurvesState &state = gCurvesMesh[mesh.name];
 
-  if (gGLProgramState.attribs["POSITION"] >= 0) {
-    glBindBuffer(GL_ARRAY_BUFFER, state.vb);
-    glVertexAttribPointer(gGLProgramState.attribs["POSITION"], 3, GL_FLOAT,
-                          GL_FALSE, /* stride */ 0, BUFFER_OFFSET(0));
-    CheckErrors("curve: vertex attrib pointer");
-    glEnableVertexAttribArray(gGLProgramState.attribs["POSITION"]);
-    CheckErrors("curve: enable vertex attrib array");
-  }
+	if (gGLProgramState.attribs["POSITION"] >= 0) {
+		glBindBuffer(GL_ARRAY_BUFFER, state.vb);
+		glVertexAttribPointer(gGLProgramState.attribs["POSITION"], 3, GL_FLOAT,
+				GL_FALSE, /* stride */ 0, BUFFER_OFFSET(0));
+		CheckErrors("curve: vertex attrib pointer");
+		glEnableVertexAttribArray(gGLProgramState.attribs["POSITION"]);
+		CheckErrors("curve: enable vertex attrib array");
+	}
 
-  glDrawArrays(GL_LINES, 0, state.count);
+	glDrawArrays(GL_LINES, 0, state.count);
 
-  if (gGLProgramState.attribs["POSITION"] >= 0) {
-    glDisableVertexAttribArray(gGLProgramState.attribs["POSITION"]);
-  }
+	if (gGLProgramState.attribs["POSITION"] >= 0) {
+		glDisableVertexAttribArray(gGLProgramState.attribs["POSITION"]);
+	}
 }
 #endif
 
@@ -693,16 +818,16 @@ static void DrawNode(tinygltf::Model &model, const tinygltf::Node &node) {
 
 static void DrawModel(tinygltf::Model &model) {
 #if 0
-  std::map<std::string, tinygltf::Mesh>::const_iterator it(scene.meshes.begin());
-  std::map<std::string, tinygltf::Mesh>::const_iterator itEnd(scene.meshes.end());
+	std::map<std::string, tinygltf::Mesh>::const_iterator it(scene.meshes.begin());
+	std::map<std::string, tinygltf::Mesh>::const_iterator itEnd(scene.meshes.end());
 
-  for (; it != itEnd; it++) {
-    DrawMesh(scene, it->second);
-    DrawCurves(scene, it->second);
-  }
+	for (; it != itEnd; it++) {
+		DrawMesh(scene, it->second);
+		DrawCurves(scene, it->second);
+	}
 #else
-  //If the glTF asset has at least one scene, and doesn't define a default one
-  //just show the first one we can find
+  // If the glTF asset has at least one scene, and doesn't define a default one
+  // just show the first one we can find
   assert(model.scenes.size() > 0);
   int scene_to_display = model.defaultScene > -1 ? model.defaultScene : 0;
   const tinygltf::Scene &scene = model.scenes[scene_to_display];
@@ -752,7 +877,8 @@ int main(int argc, char **argv) {
 
 #ifdef _WIN32
 #ifdef _DEBUG
-    std::string input_filename(argv[1] ? argv[1] : "../../../models/Cube/Cube.gltf");
+  std::string input_filename(argv[1] ? argv[1]
+                                     : "../../../models/Cube/Cube.gltf");
 #endif
 #else
   std::string input_filename(argv[1] ? argv[1] : "../../models/Cube/Cube.gltf");
@@ -763,7 +889,8 @@ int main(int argc, char **argv) {
   bool ret = false;
   if (ext.compare("glb") == 0) {
     // assume binary glTF.
-    ret = loader.LoadBinaryFromFile(&model, &err, &warn, input_filename.c_str());
+    ret =
+        loader.LoadBinaryFromFile(&model, &err, &warn, input_filename.c_str());
   } else {
     // assume ascii glTF.
     ret = loader.LoadASCIIFromFile(&model, &err, &warn, input_filename.c_str());
@@ -825,15 +952,14 @@ int main(int argc, char **argv) {
 
 #ifdef _WIN32
 #ifdef _DEBUG
-    const char *shader_frag_filename = "../shader.frag";
-    const char *shader_vert_filename = "../shader.vert";
+  const char *shader_frag_filename = "../shader.frag";
+  const char *shader_vert_filename = "../shader.vert";
 #endif
 #else
   const char *shader_frag_filename = "shader.frag";
   const char *shader_vert_filename = "shader.vert";
 #endif
 
-
   if (false == LoadShader(GL_VERTEX_SHADER, vertId, shader_vert_filename)) {
     return -1;
   }

loader_example.cc

@@ -2,6 +2,7 @@
 // TODO(syoyo): Print extensions and extras for each glTF object.
 //
 #define TINYGLTF_IMPLEMENTATION
+#define TINYGLTF_ENABLE_SCHEMA_VALIDATOR
 #define STB_IMAGE_IMPLEMENTATION
 #define STB_IMAGE_WRITE_IMPLEMENTATION
 #include "tiny_gltf.h"
@@ -228,7 +229,8 @@ static std::string PrintParameterMap(const tinygltf::ParameterMap &pmap) {
 #endif
 
 static std::string PrintValue(const std::string &name,
-                              const tinygltf::Value &value, const int indent, const bool tag = true) {
+                              const tinygltf::Value &value, const int indent,
+                              const bool tag = true) {
   std::stringstream ss;
 
   if (value.IsObject()) {
@@ -265,11 +267,10 @@ static std::string PrintValue(const std::string &name,
   } else if (value.IsArray()) {
     ss << Indent(indent) << name << " [ ";
     for (size_t i = 0; i < value.Size(); i++) {
-      ss << PrintValue("", value.Get(int(i)), indent + 1, /* tag */false);
-      if (i != (value.ArrayLen()-1)) {
+      ss << PrintValue("", value.Get(int(i)), indent + 1, /* tag */ false);
+      if (i != (value.ArrayLen() - 1)) {
         ss << ", ";
       }
-
     }
     ss << Indent(indent) << "] ";
   }
@@ -330,13 +331,13 @@ static void DumpPrimitive(const tinygltf::Primitive &primitive, int indent) {
             << PrintValue("extras", primitive.extras, indent + 1) << std::endl;
 }
 
-static void DumpExtensions(const tinygltf::ExtensionMap &extension, const int indent)
-{
+static void DumpExtensions(const tinygltf::ExtensionMap &extension,
+                           const int indent) {
   // TODO(syoyo): pritty print Value
   for (auto &e : extension) {
     std::cout << Indent(indent) << e.first << std::endl;
-    std::cout << PrintValue("extensions", e.second, indent+1) << std::endl;
-  }  
+    std::cout << PrintValue("extensions", e.second, indent + 1) << std::endl;
+  }
 }
 
 static void Dump(const tinygltf::Model &model) {
@@ -409,6 +410,30 @@ static void Dump(const tinygltf::Model &model) {
         }
         std::cout << "]" << std::endl;
       }
+
+      if (accessor.sparse.isSparse) {
+        std::cout << Indent(2) << "sparse:" << std::endl;
+        std::cout << Indent(3) << "count  : " << accessor.sparse.count
+                  << std::endl;
+        std::cout << Indent(3) << "indices: " << std::endl;
+        std::cout << Indent(4)
+                  << "bufferView   : " << accessor.sparse.indices.bufferView
+                  << std::endl;
+        std::cout << Indent(4)
+                  << "byteOffset   : " << accessor.sparse.indices.byteOffset
+                  << std::endl;
+        std::cout << Indent(4) << "componentType: "
+                  << PrintComponentType(accessor.sparse.indices.componentType)
+                  << "(" << accessor.sparse.indices.componentType << ")"
+                  << std::endl;
+        std::cout << Indent(3) << "values : " << std::endl;
+        std::cout << Indent(4)
+                  << "bufferView   : " << accessor.sparse.values.bufferView
+                  << std::endl;
+        std::cout << Indent(4)
+                  << "byteOffset   : " << accessor.sparse.values.byteOffset
+                  << std::endl;
+      }
     }
   }
 
@@ -585,10 +610,11 @@ static void Dump(const tinygltf::Model &model) {
       }
     }
   }
-  
+
   // toplevel extensions
   {
-    std::cout << "extensions(items=" << model.extensions.size() << ")" << std::endl;
+    std::cout << "extensions(items=" << model.extensions.size() << ")"
+              << std::endl;
     DumpExtensions(model.extensions, 1);
   }
 }
@@ -602,7 +628,7 @@ int main(int argc, char **argv) {
   tinygltf::Model model;
   tinygltf::TinyGLTF gltf_ctx;
   std::string err;
-  std::string warn; 
+  std::string warn;
   std::string input_filename(argv[1]);
   std::string ext = GetFilePathExtension(input_filename);
 
@@ -610,18 +636,24 @@ int main(int argc, char **argv) {
   if (ext.compare("glb") == 0) {
     std::cout << "Reading binary glTF" << std::endl;
     // assume binary glTF.
-    ret = gltf_ctx.LoadBinaryFromFile(&model, &err, &warn, input_filename.c_str());
+    ret = gltf_ctx.LoadBinaryFromFile(&model, &err, &warn,
+                                      input_filename.c_str());
   } else {
     std::cout << "Reading ASCII glTF" << std::endl;
     // assume ascii glTF.
-    ret = gltf_ctx.LoadASCIIFromFile(&model, &err, &warn, input_filename.c_str());
+#if defined(TINYGLTF_ENABLE_SCHEMA_VALIDATOR)
+    ret =
+        gltf_ctx.LoadASCIIFromFileWithValidation(&model, &err, &warn, input_filename.c_str());
+#else
+    ret =
+        gltf_ctx.LoadASCIIFromFile(&model, &err, &warn, input_filename.c_str());
+#endif
   }
 
   if (!warn.empty()) {
     printf("Warn: %s\n", warn.c_str());
   }
 
-
   if (!err.empty()) {
     printf("Err: %s\n", err.c_str());
   }

test_runner.py

@@ -14,6 +14,7 @@ import subprocess
 sample_model_dir = "/home/syoyo/work/glTF-Sample-Models"
 base_model_dir = os.path.join(sample_model_dir, "2.0")
 
+# Include `glTF-Draco` when you build `loader_example` with draco support.
 kinds = [ "glTF", "glTF-Binary", "glTF-Embedded", "glTF-MaterialsCommon"]
 # ---------------------------------
 

tools/schema_ref_resolver/README.md

@@ -0,0 +1,20 @@
+Create single schema string by deferencing $ref in glTF schema JSON for embedding JSON scheme string in C++.
+
+## Setup
+
+Use python2
+
+```
+$ pip2 install jsonpath-rw
+$ pip2 install simplejson
+```
+
+Put `ref_resolver.py` and `generate_single_schema_doc.py` to glTF schema direcotry(i.e. `$glTF/specification/2.0/schema`)
+
+## Generate
+
+Run `generate_single_schema_doc.py`
+
+## TODO
+
+* [ ] Print date and git commit id?

tools/schema_ref_resolver/generate_single_schema_doc.py

@@ -0,0 +1,37 @@
+# for print with `end` parameter
+from __future__ import print_function
+
+import json
+from ref_resolver import RefResolver
+import base64
+
+f = open("glTF.schema.json")
+j = json.loads(f.read())
+
+# call to API resolve method
+RefResolver("glTF.schema.json").resolve(j)
+
+j_str = json.dumps(j, indent=2)
+
+
+# Run json.dumps twice to get escaped string
+escaped_str = json.dumps(j_str)
+
+# MSVC does not accept string larger than 16K.
+# https://docs.microsoft.com/en-us/cpp/error-messages/compiler-errors-1/compiler-error-c2026?view=vs-2019
+# Also, it has a hard limit of 65,535 bytes even splitting a string with double quotation.
+# So, we write string as an array of string(then application must concatenate it)
+
+# https://stackoverflow.com/questions/9475241/split-string-every-nth-character
+n = 8000 # Conservative number
+
+splitted_string = [escaped_str[i:i+n] for i in range(0, len(escaped_str), n)]
+#print(len(splitted_string))
+
+print("const char *kglTFSchemaStrings[] = {", end='')
+for (i, s) in enumerate(splitted_string):
+    print(s, end='')
+    if i != (len(splitted_string) - 1):
+        print('",\n"', end='')
+
+print("};")

tools/schema_ref_resolver/ref_resolver.py

@@ -0,0 +1,82 @@
+# This Source Code Form is subject to the terms of the Mozilla Public
+# License, v. 2.0. If a copy of the MPL was not distributed with this
+# file, You can obtain one at http://mozilla.org/MPL/2.0/.
+
+from urlparse import urlparse, urljoin
+import simplejson as json
+from os.path import isfile
+import jsonpath_rw
+import requests
+
+
+cache = {}
+
+class RefResolver:
+
+    def __init__(self):
+        self.url_fragments = None
+
+    def __init__(self, id):
+        self.id = id
+        if id is not None:
+            self.url_fragments = urlparse(id)
+        else:
+            self.url_fragments = None
+
+
+
+    def resolve(self, json_obj):
+        if isinstance(json_obj, dict):
+            for key, value in json_obj.items():
+                if key == "$ref":
+                    ref_frag = urlparse(value)
+                    ref_file = ref_frag.netloc + ref_frag.path
+                    json_dump = {}
+                    if ref_file in cache:
+                        json_dump = cache[ref_file]
+                    else:
+                        if self.url_fragments.scheme in ['http', 'https']:
+                            ref_url = urljoin(self.id, ref_file)
+                            if callable(requests.Response.json):
+                                json_dump = requests.get(ref_url).json()
+                            else:
+                                json_dump = requests.get(ref_url).json
+                            ref_id = None
+                            if 'tilte' in json_dump:
+                                ref_id = json_dump['title']
+                            cache[ref_file] = json_dump
+                            RefResolver(ref_id).resolve(json_dump)
+                            cache[ref_file] = json_dump
+                        #elif self.url_fragments.scheme == 'file':
+                        else:
+                            if isfile(ref_file):
+                                # if the ref is another file -> go there and get it
+                                json_dump = json.load(open(ref_file))
+                                ref_id = None
+                                if 'title' in json_dump:
+                                    ref_id = json_dump['title']
+                                cache[ref_file] = json_dump
+                                RefResolver(ref_id).resolve(json_dump)
+                                cache[ref_file] = json_dump
+                            else:
+                                # if the ref is in the same file grab it from the same file
+                                json_dump = json.load(open(self.url_fragments.netloc+self.url_fragments.path))
+                                cache[ref_file] = json_dump
+
+                    ref_path_expr = "$" + ".".join(ref_frag.fragment.split("/"))
+                    path_expression = jsonpath_rw.parse(ref_path_expr)
+                    list_of_values = [match.value for match in path_expression.find(json_dump)]
+
+                    if len(list_of_values) > 0:
+                        resolution = list_of_values[0]
+                        return resolution
+
+                resolved = self.resolve(value)
+                if resolved is not None:
+                    json_obj[key] = resolved
+        elif isinstance(json_obj, list):
+            for (key, value) in enumerate(json_obj):
+                resolved = self.resolve(value)
+                if resolved is not None:
+                    json_obj[key] = resolved
+        return None