What Was, What Is and What Will Be.
AI generated art has been something on my radar for a while, but the technics were not so accessible. Finally, in the beginning of 2022, I decided to pass the step and create something of my own. While the current trend when I am writing this is to get trained algorithms as Midjourney or Dall-E to create the most realistic outputs, I wanted to make something personnal, poetic and abstract.
My process was to select 226 images of my personnal work and train a model with them, using StyleGan2 computing method. The result is a 1mn video where familiar shapes come to life but never truly trespass the realm of abstraction.
This project was created as part of an overall bigger project studying the evolution of machine consciousness, and I, on my side, really feel like watching a machine dreaming.
I decided to assemble the video piece with a music from Claire Rousay - discrete (the market) - who works on creating musical journey with field recording. This way, she creates new associations and memories, which felt really appropriate in the context of my own experiment.
My process was to select 226 images of my personnal work and train a model with them, using StyleGan2 computing method. The result is a 1mn video where familiar shapes come to life but never truly trespass the realm of abstraction.
This project was created as part of an overall bigger project studying the evolution of machine consciousness, and I, on my side, really feel like watching a machine dreaming.
I decided to assemble the video piece with a music from Claire Rousay - discrete (the market) - who works on creating musical journey with field recording. This way, she creates new associations and memories, which felt really appropriate in the context of my own experiment.
Software used :
Lightroom for editing and preparing all the photos.
RunwayML to generate the video.
Da Vinci Resolve to edit the main art piece and the break down.
Tom Bolles' film emulation for color editing.
Here, I made a short video to break down the process, to allow you to experiment by yourself if you ever wanted to.
Gallery of still extracted from the generated video.
Kodama - 2022 - All Rights Reserved
