In this blog post, I'm continuing my investigation of computer-generated images and how to use them to help me illustrate my science fiction stories. The image to the right was generated by the WOMBO "Dream" software in response to this text prompt: "Visi-Sonor technology, light and sound, Isaac Asimov, science fiction, futuristic technology, telepathy, Foundation and Empire, mutant humanoid, Michael Whelan style, futuristic hand-held instrument". Figure 1 was actually subjected to some post-AI image manipulation using Photoshop.
|
Figure 2. hand-held...? |
Give Mr. Wombo a hand. In the original AI-generated image there was something strange going on with his hand (see
Figure 2). The very next AI-generated image had a more interesting
Hi Tek™ device, so I copied that hand-held device over into his hand to generate
Figure 1.
I then began using the reference image shown to the right (
Figure 3).
Figure 3 was originally made as a way to depict
Reality Viewing. Reality Viewing was a concept that Isaac Asimov invented for his time travel story,
The End of Eternity.
|
Figure 4. AI-generated Visi-Sonor
|
I was favorably impressed by the results of combining the
Figure 3 reference image with my Visi-Sonor text prompt (above). The resulting image (
Figure 4) is again a slightly altered image. Post-AI-generation, I copied that dude's AI-generated face into the scene to replace another face that freaked me out.
|
Figure 5. "He kills with his eyes."
|
The other face is shown to the right (
Figure 5). There is a line in the novel
Foundation and Empire when
The Mule says, "
He kills with his eyes." That is a lie; disinformation spread by The Mule, but apparently believed by Mr. Wombo.
The Mule is a "mutant" human with telepathic abilities and he can use the Visi-Sonor to amplify his mind-control powers. I'm not sure why Mr. Wombo tends to use male faces when I ask for images that illustrate science fiction scenes. This is particularly annoying when I provide a reference image which contains a woman's face (
Figure 3). I'd rather see young budding scientists such as Tyhry or female mad scientists such as
Dr. Arky. Also, in many cases, Mr. Wombo is amazingly quick to start making images of robots and cyborgs.....
|
Figure 6. mutant humanoid?
|
Mutants. In this case, I asked for a "mutant humanoid", but Mr. Wombo was soon generating images with cyborgs (see
Figure 6).
Figure 6 is actually a composite image in which I made the background more interesting by inserting some of the
Hi Tek™ gizmos that the image-generating software created.
Figure 6 illustrates the sort of weird things that Mr. Wombo does with hands. I guess there could be a science fiction story in which mutant humanoids might all have big claw-like robot hands. I decided to abandon my effort to get a "mutant humanoid" for this scene.
|
Figure 7. a series of AI-generated images for the Hi Tek™ console scene
|
|
reference image strength of influence
|
Tyhry the Working Girl. I changed my text prompt to "
futuristic technology, colorful light show, science fiction, cute woman bent over a console, wearing shorts, bare legs, tight sweater, Michael Whelan style, devices with futuristic fractal complexity" and got the results shown in
Figure 7, above. Mr. Wombo's true colors come out when you select "Weak" for the amount of influence that your reference image has on new image generation. A woman who is working bent over at a
Hi Tek™ console is apparently a recipe for Mr. Wombo to generate NSFW images, so I gave up having Tyhry leaning over the
Phobos Base teleportation console.
|
Figure 8. Now standing up.
|
I switched to this text prompt "
futuristic technology, colorful lights, science fiction, cute women, futuristic control console, wearing shorts, bare legs, tight sweater, devices with futuristic fractal complexity" and the reference image shown in
Figure 8. Mr. Wombo immediately wanted to darken Tyhry's hair so I shifted to "
futuristic technology, colorful lights, science fiction, blond woman, futuristic control console, wearing shorts, bare legs, tight sweater, blond hair, devices with futuristic fractal complexity". There are always strange surprises with Mr. Wombo on the job.
|
Figure 9. Tyhry's hair nanite mannequin
|
In this case, I got an image for which what looked like a cloned head of Tyhry appeared, floating above the console. I could not resist turning this freak occurrence into
Figure 9. This is the kind of lucky accident that can lead to good things.
The development of human telepathy is a major issue explored in my Exode Saga stories. The Exode Saga began with a fan-fiction story in which I tried to imagine what might happen next after the endings of Asimov's two novels Foundation and Empire and Foundation and Earth. It is a big challenge to find good ways to illustrate stories about science fiction topics such as telepathy and invisibly small nanites.
|
Mr. Wombo is creating! Click image to animate.
|
In 2022, I began a new science fiction story, called
The Cythyrya Investigation, that I planned would tell the story of repairing the dysfunctional teleportation equipment at Observer Base (see
this blog post from January). Before I could take
The Cythyrya Investigation in the planned direction, I realized that it made little sense to cram teleportation technology into a story about Alastor Cluster.
Jack Vance never depicted teleportation technology as being used in Alastor Cluster.
My 2023 story Echo of Cynym solved the teleporter repair problem, even before Tyhry and Marda had a chance to visit Alastor Cluster in search of teleportation system control nanites.
|
mannequin head fail |
So, if
The Cythyrya Investigation is no longer going to concern teleportation system control nanites, what should be the science fictional focus of the story? Why would Tyhry be holding a model of her own head? I began working with
Figure 9 as a new reference image and this text prompt: "
futuristic mannequin, colorful lights, science fiction, blond woman holding a mannequin head, tight sweater, blond hair, devices with futuristic fractal complexity". Mr. Wombo seems to have a problem with the concept that someone could hold a mannequin head.
|
Hair Nanite Experiments
|
I tried to be more specific in describing this scene: "
wig designer, a mannequin head being held by a woman, colorful lights, science fiction, blond woman holding a mannequin head, tight sweater, blond hair, devices with futuristic fractal complexity". I was rewarded with images such as the one shown to the left.
|
Why masks?
|
Pandemic Conditions. I also got more images with people wearing masks. Maybe Mr. Wombo knows that if you are working with hair nanites then you should wear a mask.
|
Figure 10. Tyhry turned into a mannequin. |
Mr. Wombo could not resist turning Tyhry into a mannequin, too (
Figure 10), so I changed the text prompt to: "
photorealistic, a woman to the right side of the scene, slim hips, colorful lights, science fiction, with a copy of her own head, tight sweater, blond hair, devices have fractal complexity". For some reason, this text prompt often created images that were called NSFW by Mr. Wombo. Another common problem was that Tyhry was sometimes turned into a waitress holding a tray.
|
Figure 11. Educating Mrs. Wombo.
|
Or holding a computer. I could not resist calling the image to the right "The Education of Mrs. Wombo".
|
Figure 12. Tyhry in the nanite laboratory.
|
The image to the left shows the type of image that I was trying to obtain from the AI image-generating software. Tyhry is in the lab, learning how to program hair nanites. She learned some of these techniques from Farrax while on the planet Sermyth, but her expertise is honed during her visit to Alastor Cluster.
|
variations on the hair nanite theme; 1: just hair, 2: head on a tray, 3: halo nanites, 4: a nice head
|
|
Leaning in. |
|
Forked tongue and extra finger
|
As shown above in
Figure 11, I've started experimenting with AI-generated poses for characters in which they break through the personal spatial boundary that seems to be a default setting for AI-generated images that contain more than one human figure. Text prompt: "
photorealistic, two women kiss, pretty women, narrow waist, flat abdomen, slim hips, small hands, smooth and seamless tight jumpsuit, blond hair, blurred background fractal complexity, five fingers". It seems strange that I have to beg Mr. Wombo to only put five fingers on human hands.
|
take me from behind
|
|
Tyhry flowering.
|
Sometimes the AI-generated images are simply strange. When I asked for an image with a hand on Tyhry's hip, the poor image-generating software got confused. It is amazing how often the WOMBO AI-program reverses the orientation of human body parts. The software seems to have some kind of "chunking" algorithm for body parts, but only a limited overall sense of the physical layout of a human body.
|
kiss me
|
|
a kiss in the stars
|
I've been rather alarmed by what Mr. Wombo does to my backgrounds. Usually the backgrounds fade away to nothing, but once in a while the software comes up with an unexpected background. I was relieved that this text prompt: "
photorealistic, two women kiss, narrow waist, hand on crotch, slim hips, small hands, smooth and seamless tight jumpsuit, blond hair, blurred background plants with fractal complexity, five fingers" did not always trigger the NSFW warning.
|
Yes, I do have a green thumb.
|
|
Stop talking now.
|
One of the goofy results from Mr. Wombo is shown to the left. In this image, one of the women was given three thumbs. Through the wonders of nanites, the botanical lab technician made one thumb green. I made the other two thumbs red and blue. I'm not sure what system Mr. Wombo uses to suddenly remove parts of the clothing (see the image to the right).
|
reference image
|
|
Auto-red eye correction
|
I
previously complained about the inability of Mr. Wombo to make human figures with symmetrical features (such as left and right eyes). I tried to make a robot and specified that the two eyes should be different. Text prompt: "
alien technology, futuristic nanotechnology, bright colors, fractal complexity, 16k photorealistic color photo, very fancy robot, left eye is green, right eye is red, darker blurred devices background". Mr. Wombo prefers to give robots symmetrical eyes, even in defiance of the text prompt.
|
AI-generated alien text
|
Sometimes in a science fiction story I want to include some fragments of an alien language. WOMBO text prompt: "
alien font, futuristic font, fancy script, ornate font, 16k photorealistic image, invent an alien font". The image to the right shows an example of an "alien font" that was generated by Mr. Wombo. Below is my official WOMBO language consultant, Wombotron.
|
Wombotron |
|
alien science fiction
|
I asked Wombotron to design an alien book cover: "Alien book cover, science fiction, text is an alien font, pretty face, ornate font, 16k photorealistic image, curvaceous alien, show a humanoid alien on the cover, complex device in the dim background". Most of the results were "far out". I don't think "humanoid" means the same thing to me and Mr. Wombo.
|
Figure 13. Earth Observation
|
I changed the text prompt to: "
book cover, (16k photorealistic image), text is an alien font, (upper left display screen shows pretty face), curvaceous humanoid on the cover, (upper right display screen shows the Earth and Moon)". Combining that prompt with a new reference image, Mr. Wombo made some interesting alien book covers. I spliced together parts of five different AI-generated imaged to make
Figure 13. In my imagination, this book is about aliens who are secretly observing Earth.
|
WOMBO error
|
For some reason, rendering this kind of image was very challenging for Mr. Wombo and often generated an error message. I've seen other commentators mention that long text prompts can confuse the image-generating software.
|
Figure 14. Mission: Earth
|
I made an alternate version of the alien book cover (
Figure 14). This alternate cover came about when I reduced the influence of my reference image and Mr. Wombo forgot about the whole "alien humanoid" thing. I don't mind making the "aliens" look quite human because in the Exode Saga, the aliens have carefully designed the human species so that we look like the aliens.
|
UIWOI |
A third version is shown to the left for which Mr. Wombo made her head larger. I added another face at the upper right because I imagine humans being caught in a struggle between the bumpha and the pek. I also added a snippet of text that was AI-generated in another rendering of this scene.
|
Deleks watching Earth?
|
Sometimes Mr. Wombo seems to forget what we are doing, as in the image to the right. Mr. Wombo really likes robots.
|
Thinking of Earth.
|
Why not substitute a planet for a head (image to the left)?
|
Translation, please
|
Mr. Wombo sure likes red lips on chicks (image to the right). This image really challenges me to imagine what the alien text says.
|
First Contact.
|
The image to the left looks like the Museum of Alien Contact in the
Ekcolir Reality, when aliens arrived on Earth in 1950. I love how Mr. Wombo totally morphed the entire scene so as to include a museum floor covered with alien text.
|
Dreams of Earth.
|
Dreaming of Earth. She has been trained by aliens for her mission on Earth (image to the right).
I decided to use another text prompt for a movie poster rather than a book cover:
|
Figure 15.
|
"
movie poster, (16k photorealistic image), text is an alien font, smooth seamless dress, slim girl, (upper right display screen shows the Earth and Moon)". Again, this is a movie for people on some distant exoplanet, so the text is in an alien language. For
Figure 15, Mr. Wombo made an interesting choice in putting text right on top of the blue-skinned alien woman. The second humanoid figure on the right was added in using Photoshop.
|
Figure 16.
|
I've long struggled with the challenge of depicting teleportation in a still image.
Figure 16 shows an alien teleporting. This image was also mostly AI-generated, but then I used Photoshop to insert the alien behind the Earth woman.
|
Figure 17. |
Figure 17 shows another example of how Mr. Wombo enjoys placing text and other things onto clothing. I added the pink/white human figure with Photoshop to imply that there are human-alien interactions in this imaginary movie.
|
Figure 18. |
For
Figure 18, Mr. Wombo left a large amount of empty space on the poster which I filled in with the AI-generated spaceship, the blue alien head and the alien text.
|
Figure 18, Tyhry and Marda?
|
I began imagining that the two figures in
Figure 19 might be Tyhry and Marda, all fancied up after the application of hair and clothing nanites. For my story
The Cythyrya Investigation, Tyhry and Marda must travel to Alastor Cluster where they can study how nanites were used for fashion and cosmetic purposes.
I imagine that it would also be easy for Tyhry and Marda to change the color of their skin. Jack Vance depicted humans of the future space age as often using skin coloring techniques, with Palis Atwrode altering her skin coloring after work when going on a date with Kirth Gersen.
Next: AI-images breaking bad
No comments:
Post a Comment