thinlens-pathtracer

"thinlens-pathtracer" is a simple path tracer that uses a perspective camera following the thin-lens approximation. The camera is more sophisticated compared to naive pinhole-based implementations, and can achieve effects such as depth of field (due to the lens-based camera system) and anti-aliasing (due to slight perturbations for each sampled camera ray of a certain pixel).

Running the Renderer

To run the renderer, first construct makefiles for the renderer by using cmake in thinlens-pathtracer/src/:

cmake CMakeLists.txt

This will create a makefile for the renderer. To compile the renderer, use make:

make

To run the renderer, execute the executable as follows:

./ThinlensRender

The camera has a default orientation, but it can be altered interactively using a debug mode (see below for more information).

Contents

There will be two applications: A debug mode and a render mode.

The debug mode uses "classic" ray tracing (not recursive) to display the scene and change orientation and lens configuration of the camera interactively. This reaches interactive speed due to the use of basic, non-recursive ray tracing. Exiting the debug mode will write the camera orientation to standard output, which could later be used in the render mode.

The render mode uses a naive path tracing algorithm that only handles lambertian surfaces, and assumes the model is surrounded by daylight. Providing camera orientation and lens configuration via the debug mode is done by piping a camera configuration file from the debug mode to the executable. Alternatively, you could simply pipe the debug mode to the render mode ./ThinlensDebug | ./ThinlensRender.

Requirements

Debug Mode:

SDL is used to facilitate realtime rendering and camera interaction through mouse and keyboard inputs.

• SDL 1.2 (later versions may or may not be compatible)
• c++11
• CMake

Render Mode:

The render mode deliberately avoids SDL in order to reduce the number of dependencies needed to run the application.

• c++11
• CMake

Possible Extensions

• Reading a model from an input file;
• More robust material data structures;
• More material properties:
  1. Specular material;
  2. Translucent material;
  3. Refraction through material.

Resources

• PBRT:
  • Information projective camera systems and how the structure of an implementation could look like;
  • Theory behind path tracing.
• BMP: Very helpful library by Arash Partow that converts in-memory structure to a BMP file.
• Model: From course DH2323 at KTH.
• Wikipedia: Naive implementation of path tracing algorithm.
• Khan Academy (+Pixar): Intuitive explanations of projective cameras.