Today I've been trying to make good AI images for my novel Rainbow Lullaby. This work features a magical woodland creature called a Lagu, so that's what I've been focusing on. In the story, Burblehoot the Lagu is described as similar to a slow loris, but I didn't have any luck at all generating images of slow lorises (or sloths, which look similar). Using prompts like 'a magical creature with big eyes in a forest of ferns' produced slightly better results with a stable diffusion engine.
I added text to the image above for use on Twitter and on my Facebook author page, but this isn't my favorite image from today's session. That would be the image below. It doesn't really work as a book image, but it sure is fun.
I generated several dozen images to produce these five pictures, using mage.space so I couldn't create them in batches. When drawing manually, I'll often make several light penstrokes approximately where a line should be, then select the best of these approximations as the artwork's true line. With AI, pushing the image generation button is like applying a pen to paper. Deciding which images to use is like deciding which of a drawing's lines are true.
Along with producing art, I did all of my usual things today. A little work. Coffee at the coffee shop. Laundry. Lamenting the miles between me and my lady. My biggest challenge of the day has been navigating the thirty foot stretch of glare ice between our back door and the laundry room. The risk of slipping is extreme. It's super dangerous.
Read my novels:
See my NFTs:
