top of page
Writer's pictureacrylicode berlin

Generative Art Python Tutorial for Penplotter

Updated: May 23, 2021

Hi there,

I am part of the Acrylicode team and today I want to share with you how I coded the algorithm for creating this artwork.

Colourful variation available as CleanNFT on https://www.hicetnunc.xyz/objkt/83667


The algorithm was written in Python, so you need a basic understanding of coding in Python. I try throughout the tutorial to explain every step thoroughly so it shouldn’t be too complicated. I also used Vsketch, an art toolkit, because it provides a simple API with svg export.

The main idea of the algorithm can be explained in 4 steps:

  1. Create a grid of points and shift them in x and y coordinates by a small random amount.

  2. Group the points by column and store all groups in a list.

  3. Iterate over the grouped points and interpolate the current group with the next one (interpolate groups i and i+1)

  4. Adjust the parameters to the desired output

All steps from installation until the output are listed and explained below


VSketch Installation


Type in a terminal following commands one after the other:


git clone https://github.com/abey79/vsketch

cd vsketch

python3 -m venv venv

source venv/bin/activate

pip install . 

mkdir my_sketches

cd my_sketches

vsk init random_lines

vsk run random_lines
(the pip install step may take a few minutes, be patient and there is a dot after the space)                                                        

Then open your favourite code editor (I use VSCode) and under vsketch/my_sketches/random_lines you should see “sketch_random_lines.py”. This is where the code gets written, we code everything inside the draw() function.


In this tutorial we are going to be using the following functions vsk.random() , vsk.lerp() , vsk.polygon , vsk.point()


1. Step: Create a grid of points


We write a basic nested for-loop and add a vsk.random() to each coordinate point in the draw() function.


    def draw(self, vsk: vsketch.Vsketch) -> None:
        vsk.size("a3", landscape=False)
        vsk.scale("cm")
  
        for row in range(20):
            for col in range(25):
                x = row + vsk.random(1.5)
                y = col + vsk.random(1)
                vsk.point(x,y)

When this gets executed you should see something like this:


2. Step: Grouping our Data


Now we slightly modify the code above by adding a variable that is going to contain all groups of lines “allColumnPoints” before the nested for-loops and a temporary variable that contains the current point group in the outer for-loop.

    def draw(self, vsk: vsketch.Vsketch) -> None:
        vsk.size("a3", landscape=False)
        vsk.scale("cm")

        allColumnsPoints =  []
        for row in range(20):
            columnPoints = []
            for col in range(25):
                x = row + vsk.random(1.5)
                y = col + vsk.random(1)
                columnPoints.append((x,y))
            allColumnsPoints.append(columnPoints)

3. Step: Interpolation


For this next step we need to import numpy as a dependency first because we are going to use vks.lerp() and this accepts a np.array as a parameter. Write this in the first line of the file


import numpy as np    

Once we have done that, we iterate over the “allColumnsPoint”, we grab the current column(current iteration) and the next column(next iteration), we extract the x and y coordinates separately and create a numpy array with these values.


    def draw(self, vsk: vsketch.Vsketch) -> None:
        vsk.size("a3", landscape=False)
        vsk.scale("cm")

        allColumnsPoints =  []
        for row in range(20):
            columnPoints = []
            for col in range(25):
                x = row + vsk.random(1.5)
                y = col + vsk.random(1)
                columnPoints.append((x,y))
            allColumnsPoints.append(columnPoints)

        for index in range(len(allColumnsPoints) -1):
            currentColumnPoints = allColumnsPoints[index]
            nextColumnPoints = allColumnsPoints[index+1]
            
            currentColumnPointsUnzipped = zip(*currentColumnPoints)
            currentColumnPointsUnzipped = list(currentColumnPointsUnzipped)
            xTuples = currentColumnPointsUnzipped[0]
            yTuples = currentColumnPointsUnzipped[1]
            xCoordinatesCurrentColumn = np.array(xTuples)
            yCoordinatesCurrentColumn = np.array(yTuples)

            nextColumnPointsUnzipped = zip(*nextColumnPoints)
            nextColumnPointsUnzipped = list(nextColumnPointsUnzipped)
            xTuples = nextColumnPointsUnzipped[0]
            yTuples = nextColumnPointsUnzipped[1]
            xCoordinatesNextColumn = np.array(xTuples)
            yCoordinatesNextColumn = np.array(yTuples)

Now the “currentColumPoints” is an array of tuples [(x,y) , ….]. In the next line zip(*currentColumPoints) separates the x and y and we get an array of two tuples like this [(x1, x2, x,3,…), (y1,y2,y3,…)], the x-tuple is the first element containing the x- coordinates and the y-tuple contains the y-coordinates. Finally, we just convert the array to a numpy array and we repeat the same process for the “nextColumnPoints”.


We are now ready to interpolate the xCoordinatesCurrentColumn with the xCoordinatesNextColumn and the same for the yCoordinates. Once these points are interpolated individually, we zip them again so we have an array of tuples like this: [(x,y), …] and lastly we call the function vsk.polygon with the array we just created.


    def draw(self, vsk: vsketch.Vsketch) -> None:
        vsk.size("a3", landscape=False)
        vsk.scale("cm")

    
        allColumnsPoints =  []
        for row in range(20):
            columnPoints = []
            for col in range(25):
                x = row + vsk.random(1.5)
                y = col + vsk.random(1)
                columnPoints.append((x,y))
            allColumnsPoints.append(columnPoints)

        for index in range(len(allColumnsPoints) -1):
            currentColumnPoints = allColumnsPoints[index]
            nextColumnPoints = allColumnsPoints[index+1]
            
            currentColumnPointsUnzipped = zip(*currentColumnPoints)
            currentColumnPointsUnzipped = list(currentColumnPointsUnzipped)
            xTuples = currentColumnPointsUnzipped[0]
            yTuples = currentColumnPointsUnzipped[1]
            xCoordinatesCurrentColumn = np.array(xTuples)
            yCoordinatesCurrentColumn = np.array(yTuples)

            nextColumnPointsUnzipped = zip(*nextColumnPoints)
            nextColumnPointsUnzipped = list(nextColumnPointsUnzipped)
            xTuples = nextColumnPointsUnzipped[0]
            yTuples = nextColumnPointsUnzipped[1]
            xCoordinatesNextColumn = np.array(xTuples)
            yCoordinatesNextColumn = np.array(yTuples)

            interpolation_steps = 9
            for interpolation_step in range(interpolation_steps):
                interpolated_x = vsk.lerp(xCoordinatesCurrentColumn, xCoordinatesNextColumn, interpolation_step/interpolation_steps)
                interpolated_y = vsk.lerp(yCoordinatesCurrentColumn, yCoordinatesNextColumn, interpolation_step/interpolation_steps)
                interpolated_coordinates = zip(interpolated_x, interpolated_y)
                vsk.polygon(interpolated_coordinates)

You are done! You should now see something like this:

A little clarification on the line:


interpolated_x = vsk.lerp(xCoordinatesCurrentColumn, xCoordinatesNextColumn, interpolation_step/interpolation_steps)


Here we are interpolating xCoordinatesCurrentColumn, xCoordinatesNextColumn and the interpolation amount ( how far away from each of the two points should the line be interpolated). When the amount is 0 the interpolation is exactly the xCoordinatesCurrentColumn and when the amount is 1 the interpolation is exactly xCoordinatesNextColumn. For any value in between, it will be proportionally placed, which means 0.5 would be in the middle, 0.9 would be closer to the xCoordinatesNextColumn and 0.3 would be closer to xCoordinatesCurrentColumn. I hope it makes sense, you can further check the vsketch documentation for vsk.lerp().


4. Step: Fine Tuning


Now that you have the algorithm up and running, you can modify in the grid creation the range of the nested for-loops, for playing around with the width and the height. You can also change the range of vsk.random() to get different results, and finally, you can play with the interpolations_steps, by randomizing it, changing it gradually or only in the even iterations or be creative and come up with crazy new ideas yourself!


After the tuning, you can save it as svg with the like button in vsketch. An example of a plotted artwork:




I hope you enjoyed this short tutorial. You can always contact us via DM Instagram (@acrylicode.berlin), we appreciate any feedback, if it was too simple or too complex, or what would you like to see in a tutorial, we are open for suggestions. If you want to see a video walk-through, watch our youtube tutorial : https://www.youtube.com/watch?v=9NQVRFnkb1E


Support us:

- by collecting our nft from this tutorial at https://www.hicetnunc.xyz/objkt/83667

- sharing this content and following us in social media @acrylicode.berlin









20,871 views2 comments

Recent Posts

See All

2 Comments


Florian Klier
Florian Klier
Feb 18, 2023

Does this tutorial assume Windows or Linux as the OS? I tried to copy and paste the commands from the beginning but ran into multiple errors. Any help appreciated as I am a total newbie :)

Like

carl
Nov 10, 2022

this downloads and installs a lot of packages, It might be worth listing them and what they do to avoid suspicion. eg why is urllib3 necessary?

Like
bottom of page