Categories
Product Pilots

An introduction to using ChatGPT Plus for image generation

 

An AI generated image of a library. The library is futuristic in its design. It has plants growing around its facade, there are solar panels on the roof, and many of the walls are made of glass.

This post was created as a support resource for participants of the image generation pilot. It may also be helpful for anyone who’s interested in the applications of generative AI and image generation in education.

Overview 

ChatGPT Plus (the paid for version), will create images for users based on their input. To do this, it uses the Dall.E image generation model. 

Interactions with ChatGPT are conversational, meaning users can discuss aspects of the images they’d like to see generated and give feedback on each iteration, enabling the creation of images that fulfil the user’s vision. 

Getting started 

If you decide to use ChatGPT Plus at your oranisation, follow the steps set out below (Skip to step 3 if you already have a free ChatGPT account, skip this Getting Started section entirely if you have a ChatGPT Plus account):

  1. Go to chatgpt.com
  2. You can create an account by clicking ‘Sign Up’ in the top right-hand corner. You then have the option of creating an account using your email, or you can continue with existing Google, Microsoft or Apple accounts.
  3. Click on your profile button in the top right-hand corner (this may be displayed as your initials or an image/avatar of yourself). This will give you the options for upgrading to a paid plan, which is currently necessary for image generation. 

Examples of use 

 I used ChatGPT Plus to create a modern design for a library. Making use of its conversational interface, I began by stating what I wanted to create and asking ChatGPT what factors I should consider. 

 “I would like to create an image of a new library building with a cutting edge design. Before we create this image together, please can you think of some key things I should decide upon or consider beforehand.”

I didn’t give clear guidance on how many factors I wanted to consider, and so I got back a very long list (which was quite hard to digest). I therefore asked it to give me a consolidated list of the key things I should consider about the design of the library, and here’s what I got back: 

“Here are the 5 key things you should consider when designing your cutting-edge library building:

  1. Architectural Style & Aesthetic: Decide on the overall look and feel—futuristic or a blend of modern and traditional. Think about the shape, materials (e.g., glass, steel), and color palette.
  2. Functionality & Space: Plan the interior layout for different user needs (reading areas, tech hubs, flexible spaces) and how technology will be integrated.
  3. Sustainability: Consider eco-friendly features like solar panels, green roofs, natural light, and energy-efficient systems.
  4. User Experience & Accessibility: Focus on user comfort, inclusivity (ramps, elevators), and spaces for community engagement (cafés, event rooms).
  5. Outdoor Integration: Think about outdoor spaces (gardens, plazas) and the way the library interacts with its surroundings, including exterior lighting and landscaping.

These considerations will guide the creation of a modern, innovative library design.”

 

I used these steers to specify how I wanted the library to look. 

“Thank you. Here are my thoughts:
– I’d like the building to blend modern and traditional. I would like there to be a lot of glass, perhaps with a faint hue, and I’d like the building to be more curved than angular in its design
– I want the library’s interior to blend modern and traditional facilities
– I want there to be a living wall, and for non-glass sky-facing surfaces to be covered with solar panels
– the building should be fully accessible
– I would like there to be a courtyard with seating

Okay. Based on this, could you create me a prompt for an image that captures the library design”

And here’s version 1: 

A large library with cutting edge design. It is as large as a ten story building. It has a courtyard in front of it. Many of the walls are made of glass. There are also numerous plants growing up the walls. At the top is a large glass dome.

I felt inspired by the possibilities of sustainable, avante garde design; however, something wasn’t quite right. 

I gave ChatGPT some constructive feedback, and it returned a much improved second version. 

 “Thank you. This looks too large for a city’s library. Remember budgets are a factor. Please try again, and make it smaller and less imposing.”

 

And here’s version 2

A smaller library with a modern design. The library is around 5 stories high. It has solar panels on the roof, many of the walls are made of glass, there is a large balcony on the first floor.

This process of iterative improvement can be continued until you get an image that sufficiently matches your vision. 

Pricing 

ChatGPT Plus currently costs $20 per month for a single account. 

Key considerations 

I have set out what we believe to be some key considerations below: 

Training the model on your content 

You can control whether your content is used to train the model by going to your profile (top right-hand corner) >  settings > Data Controls. And then clicking on ‘improve the model for everyone’ and select your preference (on or off). 

Allowing your data to be used to train an AI tool is a complex matter and you may need to take advice to fully understand the implications of this. 

Guarantees 

OpenAI has been cautious with its legal language. While they provide usage rights for the images generated by DALL-E 2, they do not offer any indemnity against copyright claims, meaning users must exercise caution and due diligence when using generated images, particularly in a commercial context. 

Responsible Sourcing 

OpenAI has not disclosed specific details about the datasets used for training DALL-E 2, which has raised concerns about the potential use of copyrighted images without proper authorisation. They do claim to follow fair use principles. As above, users must exercise caution and due diligence when using generated images, particularly in a commercial context. 


Find out more by visiting our Artificial Intelligence page to view publications and resources, join us for events and discover what AI has to offer through our range of interactive online demos.

For regular updates from the team sign up to our mailing list.

Get in touch with the team directly at AI@jisc.ac.uk

 

By Tom Moule

Senior AI Specialist at The National Centre for AI in Tertiary Education

Leave a Reply

Your email address will not be published. Required fields are marked *