While one half of the internet is all crazy about face-swap Snapchat filters, the second half is frightened by the potentially malicious use of deepfakes. Should we worry or should we rather embrace the fakes as regular part of our new visual reality? And how is such reality going to look like?
Let’s have a look in not-so-distant future, where our eyes will be witnessing a constant stream of mixed and remixed visual media, such as virtual or augmented reality, fabricated deepfake videos and remnants of “real” unmodified imagery. I will be publishing here series of visual experiments and thoughts on artificial intelligence becoming the visualisator of our world.
#GANs #syntheticmedia #deepfakes #creativeAI
These images are created by an artificial mind: Generative Adversarial Networks (GANs), who compete between each other to find a sweetspot between what’s real and fake. They are imagining our daily reality, first trying to learn the rules and imitate, later on generating their own idea of our world.
The most intriguing of all examples are human faces. What first looked like a blurry concept of a face is today photorealistic visual representation of – well… people? These are people who never existed. However, they do exist – they are being imagined in an artificial mind.
There are many by-products of this process of “imagination”. To be able to generate realistic faces we – humans – will consider human, the AI has to map much wider space of visual possibilities to get there.
Looking into this space is a trip to dark corners of known universe. You’ll meet mythical creatures and dismorphed ambivalent faces carrying some strong emotions with them. It looks as if transforming through age, race and gender is a painful process.
These faces were generated with StyleGAN – and although this model can actually generate realistic faces such as the examples below, I got more intrigued by the less realistic moments. How do we perceive human face after all? Do we save the photorealistic representation of a person somewhere in our brains? No – we rather break it into a pattern that can be recollected later. These are also abstractions, collection of values that don’t need to be “perfect” in order to function. This is what imagination looks like: before the idea gets bound by real shapes and colours, with full potential and freedom to be still developed further.