Several families of continual learning techniques have been proposed to\nalleviate catastrophic interference in deep neural network training on\nnon-stationary data. However, a comprehensive comparison and analysis of\nlimitations remains largely open due to the inaccessibility to suitable\ndatasets. Empirical examination not only varies immensely between individual\nworks, it further currently relies on contrived composition of benchmarks\nthrough subdivision and concatenation of various prevalent static vision\ndatasets. In this work, our goal is to bridge this gap by introducing a\ncomputer graphics simulation framework that repeatedly renders only upcoming\nurban scene fragments in an endless real-time procedural world generation\nprocess. At its core lies a modular parametric generative model with adaptable\ngenerative factors. The latter can be used to flexibly compose data streams,\nwhich significantly facilitates a detailed analysis and allows for effortless\ninvestigation of various continual learning schemes.\n