miniRT

A simple ray tracer implemented in C using MiniLibX. This project simulates light rays to render 3D scenes with basic geometric objects, featuring Phong reflection model and camera controls.

Portfolio walkthrough (Korean)

Code entry points

• Scene.c: camera rays, closest-hit selection and recursive reflection
• Ray.c: object intersection and blending local shading with reflected color
• Phong.c: per-light diffuse, specular and shadow calculations
• Matrix.c: SIMD matrix operations used for coordinate transforms
• Parse.c: loading the .rt scene description

Features

• Supported Objects:
  • Sphere
  • Plane
  • Cylinder
  • Circle
  • Triangle
• Lighting:
  • Ambient lighting
  • Diffuse and Specular reflection (Phong model)
  • Shadows
• Camera:
  • Real-time movement and rotation
• File Parsing:
  • Custom .rt scene file format

Getting Started

Prerequisites

• x86-64 macOS with AVX2 support: the current Makefile uses -mavx2 and links the OpenGL and AppKit frameworks.
• GCC or Clang
• MiniLibX dependencies (OpenGL, AppKit on macOS)

Installation

1. Clone the repository:

   git clone https://github.com/haelime/miniRT.git
   cd miniRT

2. Build the project:

   make

Usage

Run the program with a scene file:

./miniRT scenes/sample.rt

Controls

Key	Action
W, S	Move Forward / Backward
A, D	Move Left / Right
Space, Shift	Move Up / Down
↑, ↓, ←, →	Rotate Camera (X, Y axis)
Q, E	Rotate Camera (Z axis)
R	Toggle Debug Mode
ESC	Exit

Scene File Format

The .rt files define the scene using the following elements:

• A (Ambient): ratio [0.0,1.0], color [R,G,B]
• C (Camera): pos [x,y,z], orientation [x,y,z], FOV [0,180]
• L (Light): pos [x,y,z], brightness [0.0,1.0], color [R,G,B]
• sp (Sphere): center [x,y,z], diameter, color [R,G,B]
• pl (Plane): point [x,y,z], normal [x,y,z], color [R,G,B]
• cy (Cylinder): center [x,y,z], axis [x,y,z], diameter, height, color [R,G,B]

Preview

Sphere.RayTracing.mp4

SIMD profiling

The standalone benchmark compares scalar math with the SSE/AVX2 functions used by miniRT. It checks result equivalence before measuring dot products, matrix-vector transforms, and matrix concatenation.

make profile PROFILE_ITERATIONS=5000000

The CSV output reports elapsed time, nanoseconds per call, throughput, and speedup. math_flops_per_call is identical between scalar and SIMD rows because SIMD packs the same arithmetic into wider instructions rather than changing the algorithmic operation count.