Light and Darkness

StyleGAN is a Generative Adversarial Network which is used to train the model to generate the images. StyleGANs consist of a generator and discriminator.

The generator takes random noise as input and transforms it into images. In the case of StyleGAN, the generator produces images in a step-by-step manner, starting from a low-resolution image and gradually adding details to create a high-resolution image. It does this by learning the underlying patterns and structures present in a dataset of images during training.

The first steps in making this project was curating a dataset of images. I wanted to create monochrome and duochrome visuals based on light and dark places so I began to look for images that fit the theme through my archive of images.

The next step was running python code on a google colab notebook to start training the model. It takes a long time to generate the different iterations of images also known as seeds. I did about ten rounds of generation.

The styleGAN discriminator distinguishes between real images from the dataset and fake images generated by the generator. It learns to classify images as real or fake by comparing them to the images it has seen during training. It can choose an image and generate it based on how it's been trained. I decided to incorporate the images generated through different phases of image model training as the basis for creating my poster designs.

As for generating the textual elements that went on the poster, I sourced quotes from the book "All the Light We Cannot See" by Anthony Doerr. I was watching the show that came out about the book and was really intrigued by it at the time. I used spacy as part of my python code to recognise different parts of speech within the text and then construct custom generated phrases out of them.Further I used the hugging face’ Mistral-7B-v0.1 text generation model to generate text based on certain input prompts.

Text Source -All the Light We Cannot See by Anthony Doerr, All images taken by me