Oscillating Tensors has been inspired by Herbert W. Franke’s “Pendular Oscillogram” series, where a cathode-ray oscillograph converts the electronic signals into images. Here, the trained generative neural network learns to convert floating point numbers inside its tensors into realistic pictures through a series of multiplications and additions. Every frame of this video is an actual output of a custom neural network. Alongside are displayed several (out of 512 total) tensor values used as a number input to the model to create the shown frame. The values change linearly in the multidimensional vector space, allowing the artist to create a smooth animation. Released on the occasion of the Tribute to Herbert W. Franke. https://www.tribute-hwf.com/ AI photography, AI-generated sound Ivona Tau 2022