Add Translate Method to Vector - Demo 05

Objective

Restructure the code towards the model view projection pipeline.

Transforming vertices, such as translating, is one of the core concept of computer graphics.

Demo 05

Demo 05

How to Execute

Load src/modelviewprojection/demo05.py in Spyder and hit the play button.

Move the Paddles using the Keyboard

Keyboard Input

Action

w

Move Left Paddle Up

s

Move Left Paddle Down

k

Move Right Paddle Down

i

Move Right Paddle Up

Translation

Dealing with the two Paddles the way we did before is not ideal. Both Paddles have the same size, although they are placed in different spots of the screen. We should be able to a set of vertices for the Paddle, relative to the paddle’s center, that is independent of its placement in NDC.

Rather than using values for each vector relative to NDC, in the Paddle data structure, each vector will be an offset from the center of the Paddle. The center of the paddle will be considered x=0, y=0. Before rendering, each Paddle’s vertices will need to be translated to its center relative to NDC.

Paddle space

All methods on vertices will be returning new vertices, rather than mutating the instance variables. The author does this on purpose to enable method-chaining the Python methods, which will be useful later on.

Method-chaining is the equivalent of function composition in math.

Code

Data Structures

src/modelviewprojection/demo05.py
32@dataclass
33class Vector2D:
34    x: float  #: The x-component of the 2D Vector
35    y: float  #: The y-component of the 2D Vector

We added a translate method to the Vector class. Given a translation amount, the vector will be shifted by that amount. This is a primitive that we will be using to transform from one space to another.

If the reader wishes to use the data structures to test them out, import them and try the methods

>>> from modelviewprojection.mathutils2d import Vector
>>> a = demo.Vector(x=1,y=2)
>>> a.translate(demo.Vector(x=3,y=4))
Vector(x=4, y=6)

Note the use of “keyword arguments”. Without using keyword arguments, the code might look like this:

>>> from modelviewprojection.mathutils2d import Vector
>>> a = demo.Vector(1,2)
>>> a.translate(demo.Vector(x=3,y=4))
Vector(x=4, y=6)

Keyword arguments allow the reader to understand the purpose of the parameters are, at the call-site of the function.

src/modelviewprojection/demo05.py
106@dataclass
107class Paddle:
108    vertices: list[Vector2D]
109    color: Color3
110    position: Vector2D

Add a position instance variable to the Paddle class. This position is the center of the paddle, defined relative to NDC. The vertices of the paddle will be defined relative to the center of the paddle.

Instantiation of the Paddles

src/modelviewprojection/demo05.py
115paddle1: Paddle = Paddle(
116    vertices=[
117        Vector2D(x=-0.1, y=-0.3),
118        Vector2D(x=0.1, y=-0.3),
119        Vector2D(x=0.1, y=0.3),
120        Vector2D(x=-0.1, y=0.3),
121    ],
122    color=Color3(r=0.578123, g=0.0, b=1.0),
123    position=Vector2D(-0.9, 0.0),
124)
125
126paddle2: Paddle = Paddle(
127    vertices=[
128        Vector2D(-0.1, -0.3),
129        Vector2D(0.1, -0.3),
130        Vector2D(0.1, 0.3),
131        Vector2D(-0.1, 0.3),
132    ],
133    color=Color3(r=1.0, g=1.0, b=0.0),
134    position=Vector2D(0.9, 0.0),
135)
  • The vertices are now defined as relative distances from the center of the paddle. The centers of each paddle are placed in positions relative to NDC that preserve the positions of the paddles, as they were in the previous demo.

Handling User Input

src/modelviewprojection/demo05.py
140def handle_movement_of_paddles() -> None:
141    global paddle1, paddle2
142
143    if glfw.get_key(window, glfw.KEY_S) == glfw.PRESS:
144        paddle1.position.y -= 0.1
145    if glfw.get_key(window, glfw.KEY_W) == glfw.PRESS:
146        paddle1.position.y += 0.1
147    if glfw.get_key(window, glfw.KEY_K) == glfw.PRESS:
148        paddle2.position.y -= 0.1
149    if glfw.get_key(window, glfw.KEY_I) == glfw.PRESS:
150        paddle2.position.y += 0.1
  • We put the transformation on the center of the paddle, instead of directly on each vector. This is because the vertices are defined relative to the center of the paddle.

The Event Loop

src/modelviewprojection/demo05.py
159while not glfw.window_should_close(window):
160    while (
161        glfw.get_time()
162        < time_at_beginning_of_previous_frame + 1.0 / TARGET_FRAMERATE
163    ):
164        pass
165
166    time_at_beginning_of_previous_frame = glfw.get_time()
167
168    glfw.poll_events()
169
170    width, height = glfw.get_framebuffer_size(window)
171    glViewport(0, 0, width, height)
172    glClear(GL_COLOR_BUFFER_BIT | GL_DEPTH_BUFFER_BIT)
173
174    draw_in_square_viewport()
175    handle_movement_of_paddles()
src/modelviewprojection/demo05.py
179    glColor3f(*astuple(paddle1.color))
180
181    glBegin(GL_QUADS)
182    for p1_v_ms in paddle1.vertices:
183        paddle1_vector_ndc: Vector2D = translate(paddle1.position)(p1_v_ms)
184        glVertex2f(paddle1_vector_ndc.x, paddle1_vector_ndc.y)
185    glEnd()

Here each of paddle 1’s vertices, which are in their Modelspace, are converted to NDC by calling the translate method on the vector. This function corresponds to the Cayley graph below, the function from Paddle 1 space to NDC.

src/modelviewprojection/demo05.py
189    glColor3f(*astuple(paddle2.color))
190
191    glBegin(GL_QUADS)
192    for p2_v_ms in paddle2.vertices:
193        paddle2_vector_ndc: Vector2D = translate(paddle2.position)(p2_v_ms)
194        glVertex2f(paddle2_vector_ndc.x, paddle2_vector_ndc.y)
195    glEnd()

Paddle space

The only part of the diagram that we need to think about right now is the function that converts from paddle1’s space to NDC, and from paddle2’s space to NDC.

These functions in the Python code are the translation of the paddle’s center (i.e. paddle1.position) by the vector’s offset from the center.

N.B. In the code, I name the vertices by their space. I.e. “modelSpace” instead of “vector_relative_to_modelspace”. I do this to emphasize that you should view the transformation as happening to the “graph paper”, instead of to each of the points. This will be explained more clearly later.