It was great to get back again with a full year to look forward to and it was also great to see so many people interested in getting involved with CoderDojo Athenry.
In the Bodgers group we got off to a great start, we installed the Mu Python editor, and we wrote our first few programs.
This week we took a look at a technique for generating random dungeons. Although never mentioned on the day, this technique is often called “marching squares”. It looks at the four corners of a space at a time, some of which are open and some of which are closed, and picks a shape that blocks off the closed corners.
There are sixteen possible combinations of corners on and off. All of these can be covered with these five shapes (or a rotation of them) to represent the closed off areas:
Generating the Dungeon
We generated a 2D array (a list of lists) to store our dungeon layout. At each point we used a 2D Perlin noise value (using the noise() function) to calculate a value. The point was deemed to be either open or closed based on whether this value was higher or lower than a threshold value we specified. Varying this threshold value can make the dungeon more open, or more closed in.
Drawing the Dungeon
To draw these shapes we first defined each of them as a list of x, y points defining the shape.
We then used beginShape(), vertex(), and endShape() functions to draw them at the correct size, location and orientation by scale(), transform() and rotate().
Once we were able to draw the shapes, we just needed to loop over our grid, inspecting each set of four adjacent corners in turn and drawing the appropriate shape.
Here’s a screenshot of one random dungeon. Dots (green for open, red for closed) are drawn to show the grid and the lines between the individual shapes are also shown for clarity:
and here is is without these overlays:
Download
The files for this week can be found on our GitHub repository.
This week we creating a game where a tiny plane flies over dynamically generated terrain picking up as many boxes as possible. The player scores a point for every box, but that also makes the plane fly faster, making the game more challenging.
Perlin Noise
Normal random-number generators produce values that jump around all over the place. Perlin noise is different. It is also random, but it varies smoothly. This makes it great for mimicking the sort of randomness we see in nature.
1D Perlin Noise Landscape
Our smooth and changing landscape is generated using a one-dimensional Perlin noise value generated by the P5.js function noise(xoff).
We start with a loop that goes across from 0 -> width across the screen. We’re looking to generate a set of [x, y] points that define our shape.
We use these values:
xstart: A starting value for xoff
xinc: An amount to increment xoff by for every new location
ymin: The smallest y value we want for our landscape – something a little below the top of the screen
ymax: The largest y value we want for our landscape – something a little above the bottom of the screen
Each call to noise() generates a single value in the range 0-1. We use the P5.js function map() to change this value in the range 0-1 into a value in the range ymin-ymax.
Changing the size of xinc controls how choppy or smooth the landscape is. We tune it to a value that gives approximately two peaks and two valleys across the screen, and looks right for our game.
Moving the Landscape
Moving the landscape is achieved by changing the starting value of xoff (aka. xstart) each time we update the screen. By making it a little larger each time, the effect is that the landscape seems to scroll from right to left.
Other Parts of the Game
The other parts of the game are very standard. We define a simple plane shape (drawn using rect() calls) that can move up or down in response to the arrow keys.
We define “cargo” containers that are randomly generated on the surface of the landscape and move right-to-left at the same speed.
The cargo containers have an active property that is false if they move beyond the left-edge of the screen or get sufficiently close to the plane to be “picked up”.
We added a function to the landscape class (Ground.js) that checks for a given [x, y] location to see if that point us under the ground by checking what the height of the landscape is at that x value. If the plane is below the ground we consider it crashed.
We added a simple scoring mechanism that tracks how many boxes were collected and makes the plane move faster (really – the ground scroll faster) every time a box is collected.
Download
The files for this week can be found on our GitHub repository.
We continued working with Pygame Zero this week, we began by getting everyone up to date with where we were with the game. I forgot to mention we were following the tutorial at https://pygame-zero.readthedocs.io/en/stable/introduction.html. We then changed the screen size and added a score to make our game more fun.
Next week we will add some elements to make our game more difficult and even more fun.
This week in the Bodgers group we started working with Pygame Zero. Pygame Zero allows us to write our own games quickly and easily by doing a lot of the work involved in setting up and running a game in the background. This allows us to focus on the more creative aspects of the game.
You will have to install Python and Pygame Zero if you’re going to use Pygame Zero at home, details on how to do this are in my slides. A few of us had installation problems on Saturday, so don’t worry if you have issues we will help sort it out when we return after the break.
CoderDojo Athenry 