Back in October, Molly Templeton asked, "What Do We Want From Prequels?"
My answer to that question can be illustrated by what Asimov accomplished with his Foundation stories. Ten years ago I blogged about Forward the Foundation, one of a pair of prequel novels that Isaac Asimov wrote for his Foundation Saga. The original Foundation stories did not feature robots as characters, but eventually Asimov found a place for positronic robots in the Foundation Saga.
Asimov had the benefit of four decades to think about the origins of Psychohistory, and given the popularity of his robot stories, why not blame the invention of Psychohistory on robots? Maybe the secret to creating a really amazing prequel is plenty of time to think about how to combine two seemingly distinct ideas.
Age of the Prequel. Apparently, the first use of the term "prequel" came just when I was born, so my life has been lived in the age of the prequel. Somehow, my mind naturally functions like a machine for crafting prequel stories.
Back in 2020, I had a blog post called Writing Backwards. A common experience for me is writing a story and then almost immediately feeling the need to write a prequel. I fell into this pattern during 2022, which caused a big disruption in my telling of a new science fiction story, called The Cythyrya Investigation.
Future. When I started The Cythyrya Investigation, my intention was to resolve the mystery of the dysfunctional teleportation equipment at Observer Base. After the Bimanoid Interface 2.0 was activated, the teleporter stopped functioning.
In
Part 1 of
The Cythyrya Investigation, Rylla and Rey went into the Asimov Reality Simulation in search of advanced nanites that might be able to activate the Observer Base teleporter. I originally conceived their mission in Deep Time as a fun way of elaborating on an old idea: that the Demon Prince Howard Alan Treesong was a manifestation of nanites gone bad.
Present. In Part 2 of The Cythyrya Investigation, the setting is about 20 years before the events depicted in Part 1. Knowing that Rey has been extracted from the RSS, Eddy begins to ask if he can transfer nanites out of the RSS and into the Final Reality.
Past. I suspect that Angela Fersoni could receive information from the future that was (or, will be) shared by Brak by placing it into the Sedron Time Stream. Then that information about the future was placed into the Ivory infites that found their way into my brain.
Eddy has two visions of the future that provide him with two versions of how he is able to attain control over his telepathic mind (see "
Someday Mars" and "
Echo of Cynym"). This allows him to realize that the Sedron Time Stream not only contains information about the future, it has information from multiple Realities. Eddy hypothesizes that the Reality Simulation System makes use of the Sedron Time Stream. He begins testing all the nanites he can get his hands on for their possible ability to unlock the "future end" of the RSS.
When the Huaoshy were preparing to put an end to time travel, they knew that the Sedron Time Stream would remain as a way of accessing information from the future. However, the Sedron Time Stream also contains a wealth of information from all of the past Realities of Deep time.
 |
| icon |
Mystery Icon. Take a look at the image to the right; with two crosses.
✝ Make a guess as to the meaning of this icon. Later, near the end of this blog post (
Figure 8) I'll explain what happens when a user clicks on the button with that icon.
 |
Figure 1. Creating Quyn.
|
New for 2023.
I'm currently writing a new science fiction story called
The Hua Nanites in which I explore a way to link data about the past from the Sedron Time Stream to the Reality Simulation System. Tyhry is able to use the Simulator to obtain advanced nanotechnology from the Hua, one of the ancient biological species that gave rise to the Huaoshy.
For The Hua Nanites, I wanted to create an illustration for Quyn Vye, an older character from the story. I began with an image called "Elena9" that was made by Eclesi4stiK and which was available for use under the CC BY-NC-SA 3.0. I processed Miss Elena through the age progression software that is available at the BoredHumans website. Another similar AI software program estimated that the original Elena is about 19 years old (see Figure 1).
 |
Animate GIF; Elena aging.
|
When I allowed the software to age Elena (see the animated GIF to the right) I was rather horified. The Hua have advanced medical nanites, so I did not want the older version of Quyn to look
too old.
I selected one of the computer-generated images of a moderately older Elena that best matched my mental image of Quyn's appearance and used that to create an image of Quyn for use in The Hua Nanites (see Figure 1). Quyn is a Hua, an alien creature who has a different sort of skin and eye pigmentation than what we humans have. The full-sized image of Quyn can be seen at my Gallery of Movies, Book and Magazine Covers.
 |
Figure 2. Three computer-generated aliens with pink skin and blue eyes. My version: lower right.
|
 |
Figure 3. More AI-generated aliens.
|
Inside the Reality Simulation System, an alien species such as the Hua can be made to have an appearance that is closer to the human norm than might actually exist on some exoplanet such as Sermyth. I asked a text-to-image program to create some humanoid aliens with pink skin and blue eyes (see
Figure 2, above). Maybe one of those aliens is close to the true form of the Hua.
Two Heads are Better Than One? I wonder how the AI program "decided" to create an alien with two heads stacked vertically (see Figure 3). In this case, I asked for long hair, but I did not specify the color and I got a rainbow.
 |
AI in 2023... lipstick on a pig android
|
Chat bot fail. It would be nice if an artificial intelligence program could collaborate to create science fiction stories. Maybe in 2123! Here is some
chat bot output in 2023... 😒 Tyhry: Do you know how to repair the teleportation equipment? Bot: I don't know what teleportation is for. Tyhry: I want to use teleportation to travel from Earth to Eternity. Bot: Interesting. I want to find out more about you. Tyhry: My name is Tyhry. What is your name? Bot: Tyhry, nice name. Call me Bot. Tyhry: Can a robot composed of zeptites use a teleporter? Bot: Some of them can. Tyhry: How do you know that some robots can use a teleporter? Bot: It is a ritual I do the same way every time. Tyhry: What is the transmission range of your teleporter? Bot: I've never been asked that before.
THE END
 |
| Three AI-generated humanoid aliens with long hair, "blue eyes" and pink skin. |
 |
Figure 4. Old blue eyes (see).
|
I tried using the DeviantArt AI image generator DreamUp. I again requested a humanoid alien with long hair, blue eyes and pink skin (see Figures 4 and 5). I suppose the most interesting result was that this AI does not seem to know what "blue eyes" means.
 |
Figure 5. More blue eyes.
|
The Bottom Line: Ca$h. DeviantArt gives me five free text "prompts" and then I'm expected to start paying for the privilege of using an AI that resists my attempt to have it draw blue eyes. What a deal. The AI does seem to have an artificial sense of humor; it included what looks like a penis for the alien in Figure 5.
I've seen complaints about AI-generated art on the DeviantArt website and on Twitter and even some thoughtful discussion. I finally took some time to begin an investigation of the AI-generated art being produced by "Deviants" and I recently created my own favorites folder for AI-generated images on DeviantArt. I've been following a DeviantArt group called "A-I-Generated-Images" since September. That group links to "Dream by Wombo".
 |
humanoid alien? (made by DreamUp, see)
|
Game On. One issue that interests me is this: who holds the copyright on an AI-generated image? So far, I don't see that DeviantArt is placing any kind of copyright statement on DreamUp images such as
this one. When I submitted
this image for inclusion in my DeviantArt gallery, I saw no available options for assigning copyright or for placing the image under a copyleft license.
I like the idea that if we ever meet a space alien it will probably be some kind of artificial lifeform, maybe one that would be composed of nanorobotic components rather than biological cells. Would it look like the creature shown in the AI-generated image above?
 |
| "A humanoid alien sitting in front of their steampunk castle at night with dynamic colors and god rays." |
Based on the results from my second prompt (above) I guess "sitting in front of" means nothing to the DreamUp artificial intelligence system. I modified one of the AI-generated images shown above to make it more suited for my needs in The Hua Nanites (see Figure 6).
 |
Figure 7. Mym waiting.
|
The image to the left shows some detail with Mym, an alien Hua, sitting in front of her zeptite laboratory (
Figure 7).
This image was made using "
Mym's House" by DreamUp and "
AltCun06" by
Eclesi4stiK under the
CC BY-NC-SA 3.0 license.
For someone like me (minimal artistic ability), AI-generated images are an interesting option for helping me create images to illustrate my stories.
 |
Figure 8. Get your programmers to DreamUp a user fuck the user interface.
|
Figure 8 shows the DreamUp user interface.
If you click on the button with two green crosses, then you can select the image aspect ratio.
I hate user interfaces where there are meaningless, unexplained icons on buttons. At the very least, if you hover your cursor over the button it should have a small pop-up that tells you what the button is.
Back
in 2020 I complained about the meaningless icon epidemic, but this is a problem that is not going away. 😔
Speaking of annoyance... I wanted to make an aged image for the character Shya in my story The Hua Nanites. The AI-program added glasses (Figure 9), but Shya lives in a culture with advanced genetic engineering and medical nanites and NO GLASSES. Ug.
 |
Figure 10. Made by WOMBO.
|
Figure 10 shows an example of another "humanoid alien" made using the
WOMBO software. My request for "long hair" and "pink skin" turned into "long pink hair". However, Mr. Wombo seems to understand "
blue eyes". I was trying to create an image that would depict the character
Farryx who has a role in
The Hua Nanites.
Seeing the Future. Figure 10 is actually a composite of two WOMBO-generated images. The original alien face is shown to the right. I dislike the steady stream of latex aliens that have been turned out by Hollywood over the past 80 years. I now fear that for the next 80 years AI software will be annoying me in the exactly the same way 😖.
Above and beyond. I made the mistake of giving my email address to one of the online AI software companies that wanted to trade "free" access to a demo of their software for my email address. "Free". Ha! Now my inbox is apparently going to be forever spammed by their annoying marketing emails which they send from "personal" email accounts so as to defeat filters for commercial emails. I'm told that most people now have a garbage email account that they use as a designated target for such annoying AI-driven email marketing scams.
 |
Figure 12. AI-generated using
"photorealism style" in the prompt
|
Three Legs ARE BETTER Than Two. In an attempt to use an AI image generator to obtain an less freaky depiction of Farryx (better than
Figure 10), I gave up asking for a "humanoid alien". I eventually was able to produce
Figure 11. Farryx is an hermaphroditic alien Hua, so I wanted "her" to look alien, but not too alien.
In my experience, AI-generated human figures usually look alien enough without me having to ask for an alien (example: Figure 12).
 |
Figure 13. No Sale.
|
I'm Not An Early Adopter. I know nothing about sales and marketing, but in seems strange when companies selling access to their AI systems don't provide instructions for how people can use their software, provide demo access to their software which crashes (Figure 13) and returns unexplained errors while they continually try to get you to $pend money on their crappy product. I suppose there are millions of people who have only ever used crappy software and they don't know any better. In the case of Stable Diffusion, the software can be run on your own PC (see).
The availability of Stable Diffusion for home PC use now provides my first serious motivation to upgrade from my mid-2011 iMac to an
M1 device. For now, I'm waiting for updated instructions on how to properly construct text prompts.
 |
Figure A. trouble with anatomy
|
There is an annoying
PDF guide for text prompts that is still dated October 2022 as of mid-February 2023.
The first example in that PDF is shown to the right. However, as a science fiction story writer, I have little interest in cute cows.
Prompt for the image to the left: "cute girl, reading a book, soft smooth lighting, evening colors, 3d render, polycount, photo-realism, physically based rendering".
 |
Figure B. not using "polycount"
|
The "
prompt book" provides 16 examples that make use of the term "
polycount", but there is no description of what that term accomplishes.
My guess is that use of "polycount" in a Stable Diffusion text prompt is a way to direct the software to give preference to a subset of training images derived from the polycount website, but who knows?
Images A and B show two images that share the same text prompt except that image A includes "polycount" while image B does not.
The Stable Diffusion training set of images not only was weak on human anatomy, but also provides an "artificial intelligence" that seems to not "understand" the structure of a book. Figure B does have reasonable hands.
 |
Figure 14. AI-generated. (see)
|
Physics and Biology. It is quite clear that someone needs to teach these AI image generators some fundamental physics and biology. The results can be quite comical when the software does not have any actual intelligence or understanding of what it is doing.
Figure 14 shows an image that was generated using this text prompt: "
slim woman riding a pony with four legs, near a mountain, in moonlight, bareback, show an owl, show the moon, flying owl, no saddle, owl in sky, photorealism style". I had a devil of a time getting the software to include an owl in the image. The first time I got an owl, it was the form of this fantasy owl woman. I never imagined that it would be so hard to get a machine to draw a horse with four legs. It seems to prefer three or five legs. Also, the AI seems to enjoy putting two moons in the sky at the same time; one in full phase and one as a crescent. I had to specify "slim" because the AI seemed to enjoy putting comically large women on little ponies.
 |
| AI-generated |
Bottom line. All of my usual software complaints and whining to the side, using these AI image generators can be fun, even addicting. Sometimes I need a
fractal or a
rendered human figure, so I make use of software that can produce those kinds of images. I'll certainly continue to experiment with using the available AI image generating programs to help me create images that illustrate my science fiction stories and I look forward to their improvement in the future.
 |
AI-generated (see)
|
I went through about thirty iterations with an AI image generator and significant post-processing with Photoshop to produce the image shown to the left. Insistence on blue eyes eventually yielded some freaky blue eyes. Sometimes it feels like these AI programs must be caught in a chaotic state space where multiple users are pushing and pulling the learning matrix in different directions. More WOMBO images.
 |
Figure 15. Dryad hair-nanites.
|
Plant People.
Figure 15 shows a three-legged hair-nanite salesdryad. This image required some modification with Photoshop. The two white-coated technicians did not originally have green dryad heads. Also, I added in the flying nanite objects near her hand.
 |
Figure 16. Dryad distilling.
|
Druids and Dryads. Figure 16 shows a druid (in the background) getting a lesson in the dryad art of distilling.
 |
Figure 17. Druid and Dryad gardening.
|
Figure 17 shows another Druids and Dryads scene. All of these forest scenes are down under a deep canopy where Hi Tek™ artificial light sources are used by the technologically advanced dryads. All the fancy Dryad technology usually baffles the druids, but once in a long while a young druid begins to learn about nanites and can get into mischief...
 |
full sized image here
|
Just One More. I'll end this blog post with one more forest scene; using advanced Dryad technology, a druid has begun a series of cloning experiments (image to the right). As usual, this involves the druid making use of Dryad nanotechnology. What could go wrong? The cloned Dryad is having her under-developed legs placed into a nutrient bath. The clone's left arm is growing too rapidly. That's a green nanite containment vial resting in her hand. The longer the druid works with Dryad nanites, the more she takes on the appearance of a Dryad!
Balance. These "AI" image-generating programs have no intelligence at all, so working with them can be frustrating. However, they also are a source of surprises which can be a good stimulus for my creative processes.
Related Reading: Google image-processing AI
Next: Part 3 of The Hua Nanites.
 |
I'd settle for some human intelligence: make a useful image search function.
|
No comments:
Post a Comment