Visualizing Through AI

Descriptions are some of the best parts of a story. They can be the difference between a vague understanding of the setting or character and a vibrant image that pulls the reader/listener into the world of the story. My favorite books are filled with blooming imagery; Erin Craig is a master of descriptive imagery and, in books like hers, it is critical to true understanding of the story. In A House of Roots and Ruin, the story is told from the point of view of a young painter. Everything that she sees is in swathes of color and depth--Craig's descriptions give the reader a view, not only of the setting, but of the mind of her narrator. This is similar to how Sarah J. Maas uses Feyre's artistic view to describe the world around her. 

The narrator or character doesn't have to be a painter or artist for descriptive imagery to be almost a character in and of itself. Robin McKinley, J.R.R. Tolkein, Jules Vernes, Mary Shelley and so many others use description to create a world for their readers. Arguably, every author uses imagery to a degree, but some rely on it more than others. Some rely more heavily on dialogue for character and world development. It is the pictures that I can see when reading that truly transport me, however.

Sometimes, however, I need a little extra help visualizing my characters or worlds, especially in order to remain consistent. I don't know if other story tellers have the same problem, but sometimes I can't remember if my character has blue eyes or hazel, brown hair or blond. I get tired of flipping back and forth on pages or creating a spreadsheet of previously used descriptions so I don't utterly confuse my reader.

Enter my recent discovery of AI: it has been a brain saver! I use NightCafe (I am not paid or affiliated and actually use the free version!) and have so far had an absolute blast! Full disclosure, I honestly have no idea what I am doing when it comes to AI, as is evident in some of my creations. I enter exhibit A:

The description entered was:

a gray cat getting a treat to eat from a maid in a black dress standing in front of a 1920s stove

While the image is technically not wrong, it was certainly not what I was going for. Sometimes you realize how vague your descriptions are when you see it from someone, or something, else's eyes.

Then there is exhibit B:

The description entered was:

an athletic Asian woman with shoulder length hair wearing a long gray tunic split up the side to reveal pants and leading a gold pony up a hill

I am not exactly sure what happened here, but it did not work out super well. I can't tell if the horse is menacing or dreamlike. Either way, he isn't gold and the woman is not wearing pants.

I eventually gave up on both of these topics and moved on to other possible descriptions. AI supposedly learns from what you type and how you describe things. My phone has learned my accent for speech to text and, hopefully, this AI generator will learn from my previous descriptions. Seeing these images, however, I am worried about what I have taught it.

A far better turn out is exhibit C:

The description here was:

a gray cat with blue fairy wings sitting inside of an opening flower

It was short, it was simple, it was to the point and it turned out perfectly! Who can say this isn't the cutest kitten-fairy they have ever seen?

Another success (though not perfect); exhibit D:

The description:

an abstract portrait of a Hellenistic queen with dark hair and light tan skin, wearing a red toga and pearls

I wouldn't consider this abstract, but it is a portrait and she looks relatively Hellenistic. It is certainly a woman with dark hair and wearing pearls and a red, toga-like outfit. Small victories. 

The main idea is to get the point across. Does the face resemble something of what I was trying to convey? Are the dress or hair or wings the right color and style? I can envision certain things, but I cannot always explain them well. Having an image to base a scene or character off of has been more helpful than I can say.

AI is, for me, really the only way to do this. I can see it, I cannot draw it. I have so many friends who can draw anything they want in 2 seconds flat. My artistic talents peaked in 2nd grade. My stick figures are on point, but that is the extent of my abilities. I never thought that I would be able to use technology like AI to get my thoughts and ideas into a real image. It is tangible enough and I can easily refer back to it for consistent descriptions. It is also impossibly fun to play with!

Want to try it yourself? Go to NightCafe!