Introduction
The following work has been designed for a exam project in Raffles Milano, in collaboration between Giordano Buffa and Noemi Ciaravolo.
The video is meant to be a morphed excerpt from the odyssey.
The video is meant to be a morphed excerpt from the odyssey.
To begin the process we generated a text through an artificial neural network, training the network on the odyssey, it generated an excerpt.
After that we have selected the sentences for coherence, noticing that the net has generated different sentences that reminded prophecies we have adopted that type of storytelling keeping track of those more meaningful.
From the 6 extracted sentences, we started to generate sequences of images from them, we used two other neural networks, the VQGAN.
Once obtained these clips, we started to merge them, and seeing that the generated clips seemed to use a disturbing language, it was opted that language for the video.
Next, audio was generated by a network (Real-Time Voice Cloning) on two voice models, one with Donald Trump's voice, and another with Jennifer Aniston's voice.
In colcusion we edited the clips together with the audio, added subtitles and sound.
Thank you for watching!
