Exterior Mapping

This application provides a implementation to an algorithm proposed in Master's thesis with the name of Exterior Mapping. The Exterior Mapping algorithm aims to render large three-dimensional computer scenes using the Image Based Rendering and theory of light fields. For further information please refer to the thesis as it describes the implementation of the proposed algorithm in depth.

The application provides a renderer written using C++ and Vulkan API and was loosely implemented based on the Vulkan tutorial from Alexander Overvoorde, simple renderer by Torbjørn Årdal and some functionality was implemented with the usage of Sascha Willems' repository of Vulkan C++ examples and demos.

Contents of this repository

• eval/ - contains Python script, that generates data presented in the chapter concerning evaluation
• external/ - contains non-downloaded external libraries
• include/ - contains *.h files
• res/ - contains resources needed to run the application, namely models, shaders and configuration files
• src - contains *.cpp files

Building the application

The application uses CMake build system, so please make sure, that CMake is installed on the machine. As mentioned previously, the application also uses the Vulkan API, so latest version of Vulkan needs to be available to the application as well. Once both of these have been successfully installed. The following commands can be ran in the root of this project:

mkdir build && cd build
cmake -DCMAKE_BUILD_TYPE=Release ..
make

(Build the code in Release, as in Debug, the model parsing done by assimp takes a significant amount of time.)

As the application uses the CMake ExternalProject module, all of the libraries needed by the application are downloaded and built into the build/downloaded/ folder. The shader files are also compiled during the build, these are saved into build/compiled_shaders/ folder.

Running the application

Before running the application, please make sure that the necessary models are downloaded from this link and placed into the res/models/ folder. Alternatively, for the thesis submission, they are already placed inside the correct folder. Once that has been done, the application can be run as:

./ExteriorMapping { --recover | --config CONFIG_FILE } [ -w W H ] [ -n W H ] [ -v W H ]

where:

• --recover - runs the application with the configuration saved at the end of the last execution of the application
• --config CONFIG_FILE - runs the application using the configuration specified in the CONFIG_FILE, where CONFIG_FILE is a path to a config file WITHIN the res/configs/ folder (e.g. ./ExteriorMapping --config by_step/config.json)
• -w, -n, -v - represent the window, novel view and view grid image resolution, respectively
• W, H - represent the width and height in pixels

As mentioned, there are also scripts, that run evaluation presented in the last chapter of the thesis. These are located in the eval/ folder. The needed packages can be downloaded by running the following commands in the eval/ folder:

python -m pip install -r requirements.txt

Once that is done, the evaluation can be ran as:

python graphs.py

The graphs will be generated into the eval/graphs/ folder.

Documentation

The documentation to this application has been created using Doxygen. The documentaion can be generated using the BUILD_DOC flag passed to CMake when building.

The models

The downloadable models have not been created by the author of this repository and thesis. The famous Crytek Sponza model by Marco Dabrovic has been downloaded from this page. The model of the Porsche Cayman GTS has been downloaded from a cgtrader website and was created by the user under this link.