LATENTSCAPE AV

LATENTSCAPE offers a visually immersive journey through digital boundless territories. The project’s visual component transforms data into landscapes reminiscent of geological formations such as cracks, erosions, mountains and unrealistic shapes. Colors and textures within these digital terrains want to evoke a sense of emotional depth. These landscapes, generated using Machine Learning processes like GANs (Generative Adversarial Networks), capture the essence of topographical surfaces. The intentional misleading of training datasets results in unexpected and captivating visual outcomes, portraying imaginary virtual landscapes rich in complexity, textures, and flows. The sound library of LATENTSCAPE is crafted through SampleRNN architecture trained on a diverse datasets of sounds ranging from traditional musical piece, vocal music to radio transmissions and field recordings. Live sounds and music score are produced by blending different sonic scenarios, creating an aural counterpoint that mirrors the diversity and depth of the visual landscapes. The result is an engaging sonic experience that serves as a parallel layer to the visual storytelling. All elevation maps used to crete each different landscape are generated by Machine Learning processes such as GANs, trained with datasets composed of publicly available satellite imagery, mixed and weighted with datasets of different kinds, to mislead the training phase and lead to unexpected results. The landscapes and music, with their topological and timbral diversity, depict the meeting points of human data as the fusion of thoughts, feelings and experiences.