Hidden Worlds Technical Information

I constructed Hidden Worlds by chaining together different machine Learning projects and training them in my own specific datasets to get the results I was after. this is an initial experimental round in the development of a greater project around viewing gender through the lens of computer vision.

The concept behind this was to take the idealized figures of Greek and Roman statuary and see how a DCGAN blends them together. this is my first work with a GAN and will not be my last. Generative Adversarial Networks create a piece by learning from source material. a generator creates something and it is immediately critiqued by its adversary. in this way it learns to create works that fulfill the criteria. I have selected these works from a more latent space in the creation process, I wanted something more abstract, open to interpretation. from there, I created 3D models as an interpretation of the GAN’s output.

The DCGAN I used was by Taehoon Kim on the Tensorflow framework with some customizations by me and my own personal dataset – This is an easy GAN to get started with and I enjoyed working with it.

From there I worked with Ryan Kiros’ Skip Thoughts and Neural Storyteller to create captions for the works. using MSCOCO Image recognition and captioning, neural storyteller creates a small story. Again for me the main success was in the abstract thoughts it generated and the way that gender was a completely fluid concept. this writing is inherently nonbinary in its construction and is ideally suited to my work.

The voices were made using apple’s own voice synthesis and the music tracks were made in LangoRhythm – a midi piano music writer that assigns note values to letters and durations to vowels and the way they occur to create interesting musical compositions. Each musical track is based on the generated captions.

This project will expand further to incorporate more generative music and better trained text and captioning for my specific requirements, but for now I enjoy the broadness of the output.

These works have then been united in Unity to create an Augmented Reality Experience