The Next Step in AI Imaging
The Wholly "Nailed-It!" Aspiration
If you follow these posts, you know I’ve been doing a lot of experimenting with AI images. If you have been following for quite some time and have a good memory, you may recall that I wrote a picaresque novel with talking animal characters (and a few humans) entitled Taterskin & The Eco Defenders who travel through time and space to right wrongs, enact justice, and make the world safe for continued human and animal habitation. I sought artists to create images for this book (it’s actually two books in one, or two long novellas combined). But my budget doesn’t allow for hiring an artist, and I wasn’t able to find anyone willing to create the images on a “share the profits of the results” basis.
Anyway, I wrote all that to write this: With the advent of impressive AI image-generating capability, I’m thinking the time might come when I will be able to produce the images myself. There is one catch, though: While AI (specifically Dall-E 3, which I access via Microsoft’s Bing Image Creator) creates some stunningly beautiful images, the characters may appear different from image to image. In other words, if I generate an image of the human male protagonist, I want him to look the same in every picture; the same goes for the animal protagonists, and even vehicles. For that to happen, I would need to feed the machine/robot at least one “yeah, this is the look I want for him/her/it” image so that it could use that as a basis for subsequent images (in my case, the time-and-space travel vehicle is also a “character” that needs to look the same in every image of it). AFAIK, that’s not yet a feature available to me (uploading pictures to the AI image generator to use as a basis for follow-on images).
That capability will one day become available, though, I imagine, so think of the possibilities this will open up for writers: Authors will no longer need to do without images for lack of a willing artist to “take the plunge” with them or due to an ongoing dearth of expendable funds (IOW: they can’t afford it).
I was able to generate a few images that are at least “close enough” to illustrate Taterskin & The Eco Defenders, though. These, shown in no particular order, are just a starting point:
Raja “Stripes” Kipling resting on Tubthumper Modoc
Ooga Ah-Ooga riding E.O. Hippus; Raja “Stripes” Kipling and Marmalade
Raja “Stripes” Kipling and Marmalade
Animals gathering near Mount Roraima in Brazil
Dolly Poise Hayduke
Pard
Chumbawumba Modoc (Tubthumper’s brother)
Jubatus
Terry Wingfinger
Ooga Ah-Oooga
Jowls Behemoth
Yookie
Taterskin, Rovette, and their four Goldador pups (Lab / Golden Retriever)
Scarlet the Scarlet Macaw
The Arodnap, the Time-and-Space traversing VTOL
In the (near? distant?) future, I expect to be able to copiously illustrate my book, as will many other authors. Bad for illustrators? Perhaps, but in most cases, they wouldn’t have gotten the work, anyway, for the reasons noted above. Good for writers? Yes. Good for the reading (and picture-gazing) public? Also a yes.
Would the next step after that be for the AI robots to read a novel and not only create appropriate images for various scenes in it, but also generate an entire movie based on the novel? The possibilites seem to be practically ∞.
