Machine learning meets music and Street View imagery
The Never-Ending Journey is a machine learning experiment from the Google Arts & Culture Lab that aims to connect us with the places we know and love, and those we've yet to explore. This series of computer-generated, surrealist-inspired, dreamscapes explores three iconic locations: France, Italy and Spain.
The experiment uses a Generative Adversarial Network, NVIDIA StyleGAN2, trained on thousands of images from famous landmarks and locations in each country.
Using Google Maps and Street View data, we created imaginary round-trips between iconic landmarks in each country, generating thousands of images from a never-ending hallucinated walk.
We combined this data with creative-commons photographs of famous landmarks at each stop of our journey. The end result is a dream-like , never-ending visualisation of three destinations, driven by the music from each country.
In France, we stop by the Eiffel Tower, Mont Saint-Michel, Chamonix via Street View scenes of Plages de Bretagnes et Corse, Dune du Pilat and France's many beautiful villages.
In Spain, we pass the Palacio Real de Madrid, the Arc de Triomf, the city of Cadaques, via the rooftops of Barcelona, the Tabernas Desert, Mount Teide and Concha Beach.
In Italy, we see the Tower of Pisa, La Pelosa, the Trevi Fountain and Mount Etna, via the beaches of Sicily, the canals of Venice and the Colosseum.
Google Arts & Culture