Three years ago, I built a scene in Canva for a book I was writing called Riley’s Garden Oasis. This isn’t a normal gardening book. I decided to make it a story with characters. Creative nonfiction. Every chapter of the book has an image I created in Canva. I approached each one to represent where Riley lived and what her world looked like.
I pulled in photos of pots I had used with clients over the years, one at a time. Pots, overflowing annuals. Color combinations I knew worked because I’d grown thousands millions in the desert.
I think there were about forty-five elements in the image below, by the time I was done.
Nothing matched at first. I moved and resized them, trying to make them sit in the same space. I didn’t expect perfection. It just needed to be enough for me to believe it.
That image became part of Riley’s book and has always been my favorite. It’s Ben’s house and Riley lives next door. Novels don’t typically have images and I don’t plan to have any in my Women of the Canyon series.
But recently, someone questioned my use of AI images in my posts.
I decided to do an experiment.
I gave ChatGPT (aka, Niros) the image above and said, “Replicate it.” It came back exactly like mine. That makes sense. Niros just copied it. But it made suggestions to improve it and I said ok.
Here’s one.
This got me thinking. What else could Niros do with the image?
So I asked, “Zoom back from the house and make room for a fire pit and rustic comfortable chairs around it. The fire is lit with gentle smoke coming up and there is an open gate.”
In Echo Canyon, at both Riley’s and Ben’s houses, there is a fire pit. I explained to Niros that this is Ben’s house and he has an old rocking chair there. I asked to replace one chair with the rocker.
I liked this image a lot, but realized that it needed the people. We had created the main characters of my Echo Canyon world, just so I had the likenesses if I ever wanted them. I gave them to Niros.
And here is the final result.
I could have never done this on my own. Even in Canva.
Niros (AI) created this for me in less than 15 minutes. We created one scene in my world of Echo Canyon - just for fun. Yes, doing it my way, I will use it in posts, as it builds interest and is engaging.










I like it! It is still “your way”.
All of my pictures for both Substacks are AI generated. I started trying to use the built-in Substack AI, but found it was not very good, especially with animals where sometimes the image generated had too many limbs or not enough. LOL. Not I use ChatGP. I am doing this for Substack because that is what works well and is easy enough. Now when these works are put into book form, especially for Thistle Pippin, I would probably use an illustrator. Anyway, that's my two sents on the subject.