Confoederatio Ludens Image Corpus

This is a dataset created for the Confoederatio Ludens research project, which investigates video game design, development and culture in Switzerland from 1968-2000. The images are screenshots from video games relevant to the research project and acquired through MobyGames screenshots and YouTube stills.

Setup

I like working with Miniconda and Jupyter Notebooks. If that speaks to you as well, the setup is easy.

1. Follow the Miniconda installation guide

...then do the follow:

conda deactivate
conda create --your_env_name python=3.9
conda activate your_env_name
pip install notebook

2. Navigate into this repository

jupyter notebook

3. Produce Wikidata query

The scraper scripts work with the resulting csv of a Wikidata SPARQL query. The query file is in the repository and the original query is accessible via here.

Files

• 10_Scrape_Mobygames.ipynb: Download and prepare MobyGames screenshots
• 20_Scrape_Youtube.ipynb: Download, generate and prepare YouTube stills
• 50_Create_Random_Sample.ipynb: Generate a random sample
• 60_Explore.ipynb: Basic setup to explore the dataset through image clustering

Legal

• MobyGames screenshots as well as YouTube stills are (c) by their respective creators
• Dataset created by Adrian Demleitner, 2024, licensed under CC-BY-4.0