Above&Below tells elegant stories about nature with code. “Steering Pixels” was born through a fluid simulation interpreted from a GAN, which was trained on hundreds of images of ever-changing water surfaces—data collected and sculpted by the artists. Taking computer vision to analyse motion of pixels of a neural network outcome, a fluid system initially inherits the velocity and therefore flow of the system. Using machine vision and machine learning techniques to document changing environments, computerised systems allow to not only create 1:1 replicas of environments or copies of real worlds, but allow for multiple, unexpected, and diverse perspectives.
Above&Below用代码优雅地讲述关于自然的故事。“像素舵轮”是产生自一个对抗生成网络的动态模拟，这一网络的训练数据来自上百个关转瞬即逝的水面的图像——这些图像来自艺术家的收集和整理。艺术家通过计算机视觉来分析每个像素的运动，这个液态系统内在地继承了这个系统的速度和流态特征。艺术家通过计算机视觉和机器学习去记录转瞬即逝的环境变化，他们不仅用计算系统1:1地复制既有的环境，或拷贝真实世界，而是在此基础上创造多样的，意料之外的，多样的视角。