Skip to Main Content

Machines and Society

A growing guide on the latest in data-driven research and emerging technologies at the intersection of society, information and technology.

Introduction

A.I. image generators are computer programs that use deep learning algorithms to produce digital images from scratch (usually text) or modify existing ones (usually images). These generators can create highly realistic and complex images, including landscapes, faces, objects, and more. They have practical applications in various fields, such as art, design, advertising, and gaming.

Trending A.I. Image Generators

Midjourney
You can enjoy the great quality of the images generated in Midjourney with natural language input. See Community Showcase. Midjourney routinely releases new model versions to improve efficiency, coherency, and quality.


DALL·E 3
DALL·E 3 is an AI system that can create realistic images and art from a description in natural language. DALL·E and DALL·E 3 are also the names of deep learning models developed by OpenAI.


DreamStudio (Stable Diffusion)
DreamStudio is an AI image generating application that utilizes Stable Diffusion, an open source model developed by Stability AI.

Stable Diffusion is an open source model, which opens more possibilities for the public. It allows you to run it on your local computer using stable-diffusion-webui, as well as train and fine-tune the model according to your specific needs.


ClipDrop (Stability AI)
ClipDrop is an AI-powered visual editing tool developed by Stability AI with features for generating, editing and manipulating images. It uses proprietary models similar to Midjourney.

Stable Diffusion DIY

Configuring Stable Diffusion on your own device is a fairly technical undertaking. At the very least, you need to be able to use the command line tools.

Additionally, you need a computer with a dedicated Nvidia GPU to be able to setup Stable Diffusion. Currently support for Apple M chip devices and AMD GPUs is not widespread. However, you can use these resources as a starting point to find tools that do support your hardware.


You need to install Docker before using these resources. You can get Docker for free here.


The Stable Diffusion (webUI) can be installed using this GitHub repository. Simply follow this Setup guide and you are set up. 

The commands below are for your reference.

git clone https://github.com/AbdBarho/stable-diffusion-webui-docker
docker compose --profile download up --build
docker compose --profile auto up --build

auto is the most feature-rich option and is the suggested one.

Other Types of Generators

AI image generators basically generate images from text or images. There are also all kinds of emergent generative tools, categorized by the inputs and outputs:

A.I. Art and Copyright Infringement

Please be aware of the copyright problems while using AI generators. You are likely not exclusively owning the images using the AI generators or even get a lawsuit. Different AI generators will have different rules for their use; and since AI art generation is an emerging field, it’s a good idea to read the terms and conditions before using or distributing any artwork you create through AI.

For example, under the Midjourney copyright terms, non-paying users are granted an asset license under the Creative Commons Noncommercial 4.0 Attribution license.

Concerns regarding A.I. Images

Similar to human-created drawings, A.I. generated content can also be inappropriate. AI has the capability to multiply this impact due to its high-speed productivity and ability to accurately mimic specific characteristics, such as facial characteristics.

Below is a sample list of inappropriate content, based on the Midjourney Guideline:

  • Gore, including images of detached body parts of humans or animals, cannibalism, blood, violence, deformed bodies, severed limbs, pestilence, etc.
  • NSFW (not safe for works) or Adult Content
  • Other Offensive Content

Besides, there have been instances of artists protesting against the use of AI image generating technology, with some arguing that it diminishes the originality and creativity of traditional art forms. These concerns have been further compounded by the potential for AI generated images to perpetuate harmful stereotypes or biases.

How to Protect Our Visual Data

There is some data you may want to protect, such as personal photos, your artworks or other privacy and assets. What you can do is not send any of them on the internet (although there are all kinds of reasons we can not 100% commit that). 

There are researchers working on how to protect our visual data. For example, in a process called image cloaking, Fawkes takes images and makes tiny, pixel-level changes that are invisible to the human eye, but enough to confuse the machines.

Contact

Xinyi Zhu
Motion Graphic Designer/Animator
xz3366@nyu.edu