3D Ray Tracing Engine in Python

A 3D ray tracing engine in pure Python for STL files.

Description

This is a fun project I made at the end of year 11. It's not meant to be especially fast or high-tech, I just wanted it to render an STL with shadows and reflections using ray tracing. There are more efficiency improvements that I can make to it once I finish other projects - but it works fine for the moment.

The code can render perfect spheres or any STL file (the mesh doen't even need to be closed).

The code uses Matplotlib to display the final result and NumPy for vector maths.

I used OOP and split the classes into different files to improve readability.

The stl_reader.py file handles reading the STL and converting it to a format that uses Triangle classes that are used by main.py for rendering. I used a python generator to find to coordinates of each pixel in the 3D space and convert that into 2D coordinates on the display screen. This is used in the main loop in main.py.

Requirements

If you want to render a custom STL file, make sure it is in ASCII format (binary STLs won't work) and using the millimeter scale. Then save it to the meshes directory.

Running

Run main.py to begin ray tracing. Then enter the file name of the STL file you want to render (you don't need to type the .stl file extension). Alternatively, you can type /s to only render spheres. The coodinates of the centre of the spheres must be in python list format (eg: [.1, 3, -.1]).

Changing coords of camera and light

The coordinate system uses x/y/z where z is up (this is unlike most other rendering programs).

The STLs are centered at the origin, so to change the viewing angle, you must set the coodinate of the camera and the light source. To do that, translate or rotate the camera in main.py.

"""DEFINE COORDS OF CAM AND LIGHT"""
light_pos = [1.5, -4, 4]
light = Light(light_pos, np.array([1, 1, 1]), np.array([1, 1, 1]), np.array([1, 1, 1]))
translate([0, -1.5, 2])   # set camera position
rotate(np.array([1, 0, 0]), -np.pi/4)     # set camera rotation

The code in the repo is set up to render the utah.stl file (Utah teapot) so that is why the camera and light are at odd positions.

For example, to move the light to coord (0, -4, 0), simply edit line 62 to: light_pos = [0, -4, 0]

To move the camera to coord (0, -2, 0), simple edit line 64 to translate([0, -2, 0])

To rotate the camera, use this syntax: rotate(AXIS-OF-ROTATION, ROTATION-ANGLE-IN-RADIANS) The angle of rotation is in radians and a positive angle is an anti-clockwise rotation.

Changing resolution

To improve resolution, make the resolution variable higher. This will slow rendering times.

Changing colours of objects

To assign random colours to each triangle in an STL, set the random_colours variable to True.

Otherwise, to edit the colour of the STL, edit the default_col. The colours use RGB and use a scale from 0 to 1. You can also change other variables, like spec_col, shine and refl_coef. See https://en.wikipedia.org/wiki/Blinn%E2%80%93Phong_reflection_model for more info on how the colouring system works.

Make sure that the light is illuminating a side of the object that is visible to the camera. Otherwise, the image will just appear black becasue the camera is looking at an object that is in shadow.