#CUDA-based GPU raytracer This project implements a GPU raytracer using nVidia CUDA.

The following featues are implemented:

• Parallel Ray casting
• Ray-sphere and Ray-Triangle intersections
• Phong reflection model
• Area lights and soft shadows
  1. Area lights are sampled in 4x4 grid resolutin; each sample ray is jitterd with randomness to decrease blocking effects.
  2. cuRAND is used to generate the random values
• mirror reflection
  1. An iterative approach is used to gather reflectance energy without the use of costly recursive approach is avoided.
  2. The default depth of reflection is 4.
• Wavefront obj model rendering

#Rendering Snapshot

• Recording of execution: http://www.youtube.com/watch?v=46AOCXCAYd8
• 30-sec. rendered animation: http://www.youtube.com/watch?v=ddbjBcigC7Q&feature=youtu.be
• Snapshot of execution: https://raw.github.com/otaku690/Project1-RayTracer/master/execute_snapshot.png
• Snapshot of rendered image: https://raw.github.com/otaku690/Project1-RayTracer/master/render_results.png

#Performance evaluation setting: 1024x768 res., 4x4 shadow & lighting samples, 4 bounces of reflection

Use of shared memory

Given that each ray test the geometries in the same sequence, we can load the geometries into shared memory before testing them, one at a time.

This approach increase the performance by about 4~5%. However, further investigation is needed because some blocks would display anormally.

Block size

• without using shared memory
• 8x8: ~26 sec. per frame
• 16x16: 25 sec. per frame
• 32*32: 25 sec. per frame
• using shared memory
• 8x8: ~25 sec. per frame
• 16x16: 24 sec. per frame
• 32*32: 24 sec. per frame
• There are no significant difference in 8x8, 16x16 and 32x32 blcok sizes

#Third-party code

• GLM object loader from Nate Robins

#Third-party libraries

• GLEW
• Freeglut
• FreeImage
• GLM

#Development Environment

• Visual Studio 2012 on Windows 7
• How to build
  • Make sure the project has correct INCLUDE and LIBRARY Pathes of the above libraries.
  • Make sure the CUDA 5.5 is selected in the [Build Customization] Setting.
  • Make sure the compute_10/sm_10 compute version is remvoed from the Code Generation setting under the [0CUDA C/C++] setting
  • Place the needed DLL inside the execution folder.
  • Place testScene.scene and model venusv.obj & box.obj in the execution folder.
  • You are good to go.