top of page
Banner for the post reading page of the blog, featuring a hamster wearing a mortarboard, sitting behind an open book with an ornate, dark green and intricate patterned background, creating a scholarly and elegant atmosphere.

BLOG

Writer's pictureDavid Fairbairn

Can AI be Used to Illustrate a Book?

Updated: Feb 16, 2023

The Creation of "The Duck Bread Baker"

I recently published my first book, “The Duck Bread Baker”, published by Cricetus Cricetus. It is a short, rhyming children's book with bright illustrations, promoting healthy eating and responsible treatment of both ourselves and wildlife. The target audience is children between the ages of one to six.


As a highly visual person, I have struggled with bringing my ideas to life in my writing for over a decade. My mind often says, "This needs a map!" or "What would this look like next to an illustration?" which leads me to spend countless hours searching for mapping tools and trying (often unsuccessfully) to bring my visions to life. I am friends with talented graphic designers and artists, but they are not always available or willing to work without pay. Fortunately, the advent of AI image generation has made it possible for me to create art without being skilled in the traditional sense.


You may have heard of ChatGPT, an AI language model that has the potential to change your life, I used it to proofread and edit this post (to be clear this is not ChatGPT generated content, this is ChatGPT proofed and edited content, I am dyslexic and for editing, spelling, grammar and structure I find it invaluable). Similarly, AI image generation tools such as Dall-E 2, Wonder, Artbreeder, Midjourney, and many others, can take a text description, such as "An apple eating an apple," and generate a picture based on it using machine learning. Some of these tools are free, while others require payment or have limited free versions.


“An apple eating an apple” as generated by Wonder
“An apple eating an apple” as generated by Wonder

“An apple eating an apple” as generated by Dall-E 2
“An apple eating an apple” as generated by Dall-E 2

So, Can AI Be Used to Illustrate a Book?

The answer is yes, but with a caution. AI-generated images are considered authorless and therefore cannot be copyrighted. This means that once generated, they are open-source and free to use. However, anyone else can use these images without permission and there is no legal ground to object. To protect these images, they must be edited in some way to make them unique and identify a human author. This is exactly what I did for "The Duck Bread Baker" as every image has been edited, combined with other images, or corrected to create a new composition. With the help of programs such as Photoshop or GIMP, and tutorials found on YouTube, anyone can turn AI-generated images into their own creations.

PLEASE NOTE: I am not qualified to give legal advice and before publishing or using AI-generated imagery please take appropriate steps to ensure you are within the laws of your region.


Should AI be Used to Illustrate a Book?

There are moral and philosophical debates surrounding this topic (and whenever you bring up AI someone will bring up Terminator and the inevitable machine uprising), but it ultimately comes down to personal preference and skill. If you have the ability to illustrate the book yourself, it might be more fulfilling to do so. However, if you don't have the skill or time, AI image-generation tools can be a useful alternative. Just remember to edit the images and make them unique to avoid any potential legal issues.


Tips and Tricks for AI Image Generation

As the author of "The Duck Bread Baker", I can attest to the fact that generating images with AI can be a bit of a trial-and-error process. I generated over 5000 images for this project alone, saving about 500 of them, and finally using only about 60 of them for the 43 images that make up the book and the cover. Despite using AI image generation for various projects before, this was by far the largest project I tackled. Through my experiences, I learned several key things that may be useful to others.


Persistence is key

Don't get discouraged if the first attempt doesn't yield the desired results. The "regenerate" button is there for a reason, as AI often produces unexpected results. Keep trying until you get the outcome you want.


Set realistic expectations

AI is good and improving, but it still has limitations. If you have a very specific vision in mind, it may be difficult for AI to fully satisfy it. However, if you are going for a more abstract or general image, success is more likely.


Continuity is a challenge

If you have a series of images in mind with a consistent main character, AI image generation may not be the best solution for you. Currently, the systems I am familiar with do not have the capability to maintain continuity across generated images.


AI has difficulty with numbers

AI image systems struggle with numbers. No matter how you phrase it, the system may not produce the desired number of objects. For example, if you ask for "five apples," you may not get five apples.


Be specific and mindful of language

AI will generate the image based on the exact words you use in your request. Consider the language used, as not all AI systems are developed by English-speaking teams. Try using synonyms if you're not getting the results you want. Also, provide as much detail as possible in your description to increase the chances of getting the desired outcome.


Use multiple AI systems

Each AI system will generate images based on its own data sets and algorithms. As a result, using different systems will yield different results for the same request. Additionally, some concepts may be beyond the capabilities of a particular system. To overcome this, try using multiple AI systems and adjusting the image through a single system with a consistent style filter.


Is Using AI Cheating?

This is a subjective question, but in my opinion, no. Using tools to achieve our goals is a hallmark of human progress, and AI is just another tool in our arsenal. The problem arises when these tools are used for unethical purposes. Whether using a computer to create an image (such as in Photoshop or Illustrator) or using AI to generate an image, the tools were developed by humans to make our lives easier and more efficient to allow us all to achieve more of the things we want to do in our limited time on this earth.


Resources:


169 views0 comments

Comentários


bottom of page