top of page

Utopia

Featured showcase of work

Utopia_0.png
Utopia_3
Utopia_4.png
Utopia_5.png
Utopia_5.jpg
AI Generated Artwork: Portfolio

Dystopia

Featured Showcase of work created using an AI System - Disco Diffusion 

Dystopia IV .png
Dystopia V .png
Dystopia_2.png
Dystopia_0.png
Dystopia_3
AI Generated Artwork: Portfolio

Rage

A series of images created based on the Fear, Rage and Anxiety

Human emotions(0)_2.png
Human emotions(0)_0.png
Human emotions(0)_1.png
humilation_1.png
humilation__2.png
humilation_0.png
Rage_1.png
Rage 3.png
Rage_2.png
Autism 1.png
Autism 2.png
Autism 3.png
AI Generated Artwork: Portfolio

Experiments

How I experimented with AI engines; GTP3, Latent Diffusion, and Disco Diffusion. 

The process starts with words, much like an artist migh use a mind-map to brainstorm ideas.


Using GPT-3 I used an AI-association engine, where a I can input a word or sentence, or a headline and have the AI system come up with 5-10 associated concepts. 


Depending on what the input, these can can be straight forward, or have a more poetic and/or political twist.


The layout can be minimalist and only text based, or have a graphical layout in the style of a force directed graph. 


On the right, is a basic example, where the input prompt “apple” produces a lot of single-word common sense associations.


These prompts are then used by GTP3 to create a sentence. The sentance can be edited or used directly to generate an image in another AI system called Disco Diffusion or Latent, both are image generators.


I used Collab Pro and Colab Pro+

while for the final website I am creating it will be hosted on Colab GCP Marketplace VMs. 


Colab GCP Marketplace allows you to specify the exact runtime resources to use and provide you with a persistent environment you can manage to your liking while still using the Colab UI.

AI Generated Artwork: Text
AI Generated Artwork: Pro Gallery

Using Google Colab

Using AI to generate images

Latent Diffusion: This is relatively straight forward to use and the images are created in a few minutes minutes. However the images are of low resolution and not suitable for detailed work.

latentdiffusion.

Mindseye (UI around VQ-GAN+CLIP or Disco Diffusion 5.0). This is medium difficulty with results in 30 minutes to several hours.

Disco Diffusion 5.2 in Google Colab: Advanced difficulty and results in several hours (depending on resolution and which GPU is use. My preferred choice based on experiments 

https://colab.research.google.com/github/alembics/disco-diffusion/blob/main/Disco_Diffusion.ipynb 

AI Generated Artwork: Text

Setting-Up Google Colab's Disco Diffusion

Using Google Colab's Disco Diffusion a text-to-image `AI engine to create images

AI Generated Artwork: Video

Image Generation Google Colab's Disco Diffusion

Generating an image using Google Colab's Disco Diffusion a text-to-image AI engine.

AI Generated Artwork: Video
AI Generated Artwork: Pro Gallery

Subtitle

This is your Text paragraph. It’s a great place to add a description of your business, your site or what you do. Use this space to add information for your users, write about your philosophy or your journey and define your distinguishing qualities. Consider adding an image for extra engagement.

AI Generated Artwork: Text
  • Facebook
  • Twitter
  • LinkedIn

©2022 by Emma Fossick. Proudly created with Wix.com

bottom of page