Creating Illustrations

Ideally, for my children’s book, I would like to create my own illustrations. However, right now I do not think I have the time or resources to do that as I prepare for my first teaching practicum! I am proud of the book I have created so far, and would like to use my own art for my final copy. I enjoy engaging in artistic pursuits, and think it would be very rewarding to have my own artwork in the book I created. Learning how to create illustrations is a challenging and time-consuming process, so for now, I enlisted the help of AI.

I started by researching AI tools that are useful for illustrations. The challenge I ran into first was that many of these illustration tools cost money (or give you a short free trial and then cost money), and I did not want to pay for an AI tool that I may not end up wanting to use. I tried out CapCut AI first, but found that the images it produced had errors, and it was challenging to prompt it to generate images in the same style with a consistent representation of the main character. However, CapCut did provide one image when I prompted it, in which I liked the illustration style and portrayal of my main character, Gia. I then took that image and used Gemini AI to develop more scenes from my children’s book.

While Gemini AI produced images in a similar style, I found that it was taking multiple prompts to get the image I wanted. Also, the portrayal of my main character was inconsistent: Gemini was varying the illustration style of my character, and making her skin colour lighter the more I prompted it. As a result, I was having to make multiple corrections for every image. This could also be my own fault; I could be more specific and detailed in my chat inputs. Another problem I ran into with Gemini was that after a certain number of images, it stops you from creating more for 24 hours. So, I only got a few pictures generated before having to stop, and generate more images the next day. In the end, I only got a few images I determined usable, as I was having to correct Gemini many times before getting an image I was happy with, and even then, it was not exactly what I wanted. I think in the future I do need to practice being more specific with my prompts, which would hopefully make the AI output more accurate, and mean that I would not need to correct the output multiple times.

See my Gemini conversation here.

Based on the images Gemini created, I created the beginnings of my children’s book. However, I do not plan to use these images in my final copy due to the inconsistency of the illustration style and portrayal of my main character.

See a rough version of the first few pages of my book here:

I tested CapCut a second time, and using my Gemini images as a reference, I tried to prompt CapCut to create more images for my children’s book in the hopes that it would do a better job of maintaining a consistent illustration style and main character appearance. Once again, I found it challenging to prompt AI to create the exact image I wanted. However, CapCut AI did do a better job of maintaining a consistent illustration style and main character throughout the conversation.

See my CapCut conversation here.

Although I will not be using these images in my final copy, Gemini and CapCut did give me some good ideas for what illustration style I would like to use in my book, and I have a solid idea for what I want my main character to look like. I plan to incorporate similar images in my final copy. Also, I plan to continue to explore different illustration tools, including AI tools, to see what else is out there, and if there are more effective tools I could use.

References

CapCut AI. (2026). CapCut AI Graphic Design Studio [AI-generated image]. https://www.capcut.com/

Google. (2026). Gemini (Gemini 3 Flash version) [Large language model]. https://gemini.google.com/