Datarapper

2020

/ observable notebook /

In Datarapper, I explored Colombian rap through computational tools and principles. I created 8 posters with images generated by using the lyrics from Gordo Sarkasmus' 'Hallowinners II' album as a dataset.

Each poster was built through the following process:

1. Prepare the dataset: After transcribing the lyrics, Knime was used to remove punctuation, filter the words so that only nouns, adjectives, and verbs remained, and create a JSON file with the resulting bag of words.

2. Get the images: Using the Flickr API, an image search was performed based on the words in the JSON file created in the previous step.

3. Draw the images: From the images obtained for each word, one was selected randomly. Then, using the blend() function of the p5.js library, a canvas was created on which each of the selected images was superimposed.

4. Augment the canvas: After downloading the canvas from p5.js, it was upscaled using Bigjpg (CNN) so the pieces could be printed in a large format.

5. Create the posters: 8 pieces were created by recomposing, remixing, and copying B-movie and R-rated film posters, adding the title of the song and other graphic elements.