Time Is Inherently Cruel is a machine learning video art piece that attempts to contain time by registering a stream of data from the organic world and giving it a life of its own in the digital one. Reflections of moments that are no longer there: canvases in constant flux.
The resulting artificial spaces serve as conduits for an inner dialogue –a meditation– about migration and the emotions and feelings that haunt the displaced.
–recommended viewing experience: set the resolution to 4k, watch fullscreen, and wear headphones–
AMBIENT VERSION
MAKING-OF
DATASET
The dataset used in this project was crafted through a durational process. A 6 months-long performance: 1 photo every hour (daylight), every day, for 183 consecutive days. Spanning 2 continents and 4 cities.
MODELS AND SEEDS
After training for multiple days, an evaluation and careful selection of the generated models was made. From a final set of 12 models, a group of 72000 seeds was generated (6000 seeds per model). Out of these seeds, a final selection was made and used for the creation of latent space walks (interpolation videos). In total 14 interpolations were generated, only 5 made it to the final cut.
PROJECTIONS
Finding the matching latent vectors for a set of images.
