Book-Scrolls and Plant-Machines

Figure 1. Alien time travel device by Gemini.

 Last month I tried to make some AI-generated images depicting time travel (see the blog post: The Time Travel Game). I ended up with some images that I could use to illustrate The Historical Archive, a time travel story that I wrote back in 2022. While trying to illustrate science fiction stories, I constantly find myself fighting against the natural tendency of AI-powered image generators to simply vomit forth variations on images that are in the software's training set. The question becomes: how can I trick the software into visually depicting some future technology that exists only in my imagination? 

Figure 2. By Wombo Dream.

I asked Gemini to generate an image depicting: "futuristic advanced technology discovered on a distant exoplanet, intricate details in the fractal complex alien design of a time travel device". One of the four generated images is shown in Figure 1. I liked the image in Figure 1 because it seems to depict relatively complex alien environment. However, when an AI is fed the word "exoplanet" you can bet on being provided with in image that depicts a sky full of some absurdly unlikely combination of moons and giant planets.

The image in Figure 2 was generated by Wombo Dream and is centered on an object that is quite similar to one of the alien mega-structures that is included in Figure 1. I asked Gemini, "Why do AI-image generators tend to depict scenes on imaginary exoplanets as being in dry, desert-like settings with very little plant life?"

Figure 3. Alien plants by Gemini.

Gemini: "Deserts are often seen as being visually striking, with their vast expanses of sand and rock. They can also be seen as being mysterious and alien, which may appeal to people's sense of wonder." 

Waste Not. While responding to my question, Gemini also generated several images with plants, including the one shown in Figure 3. A week or so ago, I had to twist Gemini's arm to get it to generate images. Today, Gemini insisted on continuing to generate images even when I did not ask for them. However, I did find a way to make use of Figure 3 as a background for Figure 13, below.

alien fossil? Click image to enlarge

I asked Gemini: "In the case of a human artist trying to depict imaginary alien technology on an exoplanet, might the artist intentionally avoid including visually-interesting plant life so as not to distract viewers of the image from the alien technology?" Gemini instantly generated several images that showed human hands drawing supposedly "alien technology"... or something that Gemini considered to be alien. One of these images is shown to the left. 

← In my imagination, the strange stone that Gemini had generated looked like what might possibly be an alien fossil that has been discovered on a distant exoplanet.

fossil by Mr. Wombo

I was puzzled by Gemini's image of an "alien fossil"; there was no real indication of what the human hand was doing there. Maybe the human hand in the lower right corner of the image is the hand of an exopaleontologist taking notes on the fossil's discovery... maybe making a quick sketch of the find. 

I had Mr. Wombo make an alternative version of this alien fossil (see the image to the right). Text prompt: "a fossil discovered on a distant exoplanet, the fossil has some fractal complex colorful crystal agate mineralization, plants are seen in the background". In Mr. Wombo's version of the scene, the hand has been converted into a rock. Mr. Wombo was continually teetering between making images that looked like rocks and images that looked like living creatures. It was hard for Mr. Wombo to generate images that actually looked like fossils.

fossil #1

While using no reference image, I asked Mr. Wombo to generate images for this simple text prompt: "a fossil organism in limestone". There are "styles" available with Wombo Dream that are less likely to add color to images, but I actually like the idea of a colorful "fossil".

fossil #2

Mr. Wombo did not hesitate to make the colorful fossil that is shown to the left. 

When Mr. Wombo generates images such as the one that is shown to the right, I start to wonder if the software remembers my past history and used my past topics to guide new image creation. I'm totally into the idea of finding some sort of strange alien fossil. I suppose Mr. Wombo finds it hard to make a skull without eyes in the eye sockets.

fossil #3

One of the new "styles" for WOMBO Dream is Diorama v3, which I used to make the image that is shown to the left. Tails high, brothers!

fossil #4

Here is one more result with the Diorama "style" (see the image to the right). 

For me, anything that is put under glass always seems to look more important. I only wish that Mr. Wombo had embedded that cute display case in a complete museum scene. 

Including a human hand that is depicted as holding a colorful and detailed fossil is a nice touch. As usual, Mr. Wombo struggles to make an anatomically-correct thumb.

Exobiologist. Click on the
image to enlarge.

I'm always trying to enhance images by including human or alien figures in a scene. The image shown to the left includes a modified version of the "alien fossil". There are some teeth-like structures close to her right hand. This could be the exoplanet Elemacha-z where the high temperatures make it unlikely that working scientists will wear much clothing.

Exogeology

For the image to the right, I was using the "weak" setting and Mr. Wombo generated what I'll call a human Exogeologist with an interesting mineral sample being collected on an Exoplanet.

Figure 4. Alien device and plants.

 Botanical Adventures. I asked Mr. Wombo ("Mr. Wombo" is my pet name for Wombo Dream) to include some plants in a scene along with an alien time travel device. One of these images that was generated by Mr. Wombo is shown in Figure 4. In this image, Gemini seemed to physically merge the alien time travel devices with some of the alien plants. 

 Plant-Machines. The ability of the WOMBO AI-image generator to intermingle two categories of objects; 1) machines and 2) plants (types of objects that the human mind usually keeps distinct) reminds me of Gemini's recent invention of the book-scroll. The image in Figure 4 might be a depiction of the "plant-machine", the plant kingdom's equivalent of the cyborg. 

Figure 5. Alien humanoid.
Full sized image here.

However, this is just the sort of strange alien life that I am trying to imagine for the exoplanet Elemacha-z, so I am in the mood to mix together plants and machines.

For illustrations of science fiction stories, I usually want either 1) a human being or 2) some alien creature in the image (if not both). For the image to the right in Figure 5, I used this text prompt: "futuristic advanced technology discovered on a distant exoplanet inhabited by pretty humanoids, intricate details in the fractal complex design of a time travel device made by humanoid aliens, there is a rich diversity of alien plant life surrounding the time travel device, the background contains a landscape full of strange plants".

Figure 6. An alien.

I devised a composite image that could then serve as the reference image when Mr. Womb generated Figure 5. The image in Figure 6 shows the typical output from Mr. Wombo any time when the word "alien" is included in a text prompt. This kind of alien is exactly what I meant above when I mentioned AI-image generating software that vomits forth the contents of its training set.

Figure 7. Humanoid alien.

However, by substituting "humanoid" for several copies of the word "alien" in my text prompt, I could get some images such as the one shown in Figure 7. This image was contaminated with features that I don't like (particularly, look at her neck) so I found a way to make Figure 7 into a suitable reference image.

Figure 8. A human.

I pulled out one of my pre-existing images (see Figure 8) from when I was having Mr. Wombo make scenes set in ancient Rome (another example).

Figure 9. Composite reference.

By cutting the human figure from the image shown to the left and pasting her into the scene of Figure 7, I got the new reference image that is shown to the right. The resulting composite image, Figure 9, was then used as the reference image when I had Mr. Wombo generate Figure 5, above. 

Figure 5 was generated by having Mr. Wombo use the "weak" setting for how closely to follow the reference image. By using the "weak" setting, details like the "crystal balls" were lost, but Mr. Wombo was able to transform the human figure into the kind of alien humanoid that I was hoping for.

Figure 10. "regular" setting

 Figure 10 shows the kind of image that was generated if the "regular" setting was used, allowing Mr. Wombo to more closely follow the reference image (Figure 9). In this image Mr. Wombo really went with the "crystal ball" theme, but I did not feel that the lady in Figure 10 was "alien enough".

Often I have Mr. Wombo generate humanoid aliens with green or blue skin, but the orange skin in Figure 5 was also interesting and different enough from normal human skin pigmentation for me to believe that the orange lady is an alien humanoid from some exoplanet, not a human from Earth.

Figure 11. Human scientist.

 More Orange. The image to the right shows a human figure with red hair that Mr. Wombo generated for this scene on a distant exoplanet. It is not know why Mr. Wombo will take a reference image with a blond (Figure 9) and transform it into an alternative image with a redhead (Figure 11). Maybe Mr. Wombo just "likes" redheads. Maybe Mr. Wombo "thinks" that red hair goes good with the green plants in this scene.

Figure 12. A small pumpkin.

 Hallucinate. On the subject of unexpected orange appearing in AI-generated images, the image shown in Figure 12 was one of several in which Mr. Wombo spontaneously inserted a small pumpkin or gourd into a scene that was supposed to be set in ancient Rome. In this case, there is what looks like a bird feather stuck in the pumpkin. (see also: more pumpkins)

I don't know if this kind of visual oddity can be classified as the visual equivalent of an AI-hallucination. Another one of these Roman tiny pumpkin images is shown in the composite image at the bottom of this blog post. Are AI-hallucinations simply errors, or can they be beneficial sources of creativity when an artificial intelligence is helping a human create and illustrate a science fiction story?

intentional hallucination by Gemini

I asked the Gemini large language model to try to hallucinate. "I'm intrigued by the possibility that a LLM such as Gemini could 'simulate' the occurrence of an AI-hallucination upon request of a human user. For example, if I asked you (the Gemini LLM) to imagine a scene in a library with an open book on a table, you could generate an image depicting such a scene. However, what if I asked the Gemini LLM to simulate an AI-hallucination and insert something unexpected into the scene with the open book. Could Gemini simulate a visual AI-hallucination?" One of the "intentional hallucination" images that was generated by Gemini is shown to the right.

Figure 13. Alien plant person.

 Exoplanet Elemacha-z. Since I'm currently writing a science fiction story called "The Fesarians" in which there is a type of intelligent plant-like creature with telepathic powers, I'm in the market for images of "plant people".  The image in Figure 13 depicts first contact between a human and a bipedal alien plant creature. Figure 13 was generated by Mr. Wombo by using the "jungle" image from Figure 3 (above) in the background along with this text prompt: "actress Kate Jackson as a tall slim woman in a jungle, Kate Jackson is seen from behind as she looks at a humanoid plant person, Kate Jackson had long black hair, Kate Jackson is twenty years old, the plant person is a plant-like humanoid walking in the jungle, Kate Jackson's face is seen in profile view".

Meet the plant people.

Another "plant person" image that was generated by Mr. Wombo is shown to the right. In "The Fesarians", Xylo has some clone sisters. This could almost be a scene in which two of the clones meet a native plant person of the planet Elemacha-z, but I don't really imagine the natives of Elemacha-z having this sort of appearance.

The alien hybridization experiment.

In the image to the left, I like to imagine some kind of genetic experimentation that leads to hybrid human-plant people. However, once again this fanciful scene does not really fit with the plot of "The Fesarians". 

In my imagination, the natives of Elemacha-z are part of a distinctly different evolutionary line than are we humans, so a hybrid would be essentially impossible. Luckily, Mr. Wombo is oblivious to such facts and happily makes alien hybrid human-alien images upon request. Sadly, Mr. Wombo often wants to turn such hybrids into some kind of cyborg. 😝

Talia makes a splash.

The surface temperature of Elemacha-z is higher than what we are used to here on Earth. However, Port Cooway, the human colony on Elemacha-z, was carefully positions so that is located adjacent to a cold water current that brings relatively temperate water directly from the north polar ocean. The image to the right shows Talia wading in the relatively "cool" ocean of Elemacha-z.

This was one of the rare situation where Mr. Wombo was able to generate some nipples on a female figure. I manually darkened the nipples, making them easier to see than in the original AI-Generated image.

Talia on the beach.

The image to the left is a depiction of Talia sitting on the beach during the day. The only way to survive on the surface of Elemacha-z without an environmental suit during the day is to be in the cool water. Maybe ultraviolet sunscreen has been perfected at this point in the future, but she must be wearing protective contact lenses to prevent damage to her eyes.

Talia at night.

The image to the right is a night time visit to the beach. For "The Fesarians", I imagine that Talia eventually discovers that the native Merovae spend most of their time in the ocean, but they do occasionally come onto dry land. It is on the beach, where the Merovae can quickly slip into the water, that they allow themselves to be seen by Talia and her clone sisters.

Talia on the beach II.

To the left is another depiction of Talia enjoying the ocean. During the day. Mr. Wombo prefers to depict cute women as just standing around looking cute, so this was an easy image for Mr. Wombo to generate.

 Power System Engineer. Talia is in charge of supplying electrical power to the human colonists on Elemacha-z. I was tempted to try to modify the background of some of these images depicting Talia on Elemacha-z so as to show some of the solar/wind power collectors that Talia has installed on the planet.

Figure 14. Generated by Gemini.

Another of the "time travel devices" generated by Gemini along with Figure 1 (above) is shown to the right. In my imagination, this could be an alien going through a time portal.

I used the image in Figure 14 as a reference image for Mr. Wombo along with this text prompt: "a very young Kate Jackson has long black hair, Kate Jackson a pretty twenty year old girl, Kate Jackson dressed Victoria's Secret style in a low-cut silk shirt, Kate Jackson inside a technologically-advanced time travel portal, the time portal has fractal complex control panels built into its arched structure".

Figure 15. By Mr. Wombo.

The type of time portal that Mr. Wombo generated is shown in Figure 15. Basic elements from Figure 14 are retained in the images that were generated by Mr. Wombo such as the glowing orange entry arch of the time portal.

Figure 15. Book cover version.

Starting with a similar image, I added text for a book title and author name to and created an imaginary book cover (see Figure 15) for  story called Time Portal. 

More Orange. I don't know if Mt. Wombo gave her an orange shirt to match the orange entry arch of the time portal. I was relieved that Mr. Wombo did not insist on having her wear high heals.

Figure 16. White boots version.

 White Boots. Mr. Wombo did sometimes switch off to white boots as shown in Figure 16. 

Figure 17. Rest time.

 On the Time Thrown. Eventually, Mr. Wombo decided that this intrepid time traveler must be tired from running through all of these time portals so he generated an image in which she was allowed to sit down and rest for a while (see Figure 17 to the right). For this image, Mr. Wombo decided to dress her entirely in white.

I asked Gemini, "To what extent can we view the human brain as having evolved so as to be able to function as a "novelty detecting device?" Upon my mention of the brain, Gemini generated several images of brains. I incorporated one of Gemini's brain images into a scene that is set on a distant exoplanet...

Figure 18. Gamester brains.

 Say 'Ah'. I've been re-watching the original Star Trek and the image shown in Figure 18 reminds me of the episode called "The Gamesters of Triskelion". There are unusual alien life forms on planet Elemacha-z, but sadly no giant brain creatures. I was amused by the way that Mr. Wombo has one of the humans put her hand into the alien's mouth.

Figure 19. Haunted forest.

In a sense, Elemacha-z is haunted by an ancient telepathic life-form. One of the forest images that was generated by Mr. Wombo included a white ghostly figure, so I emphasized that feature in the the image that is shown in Figure 19. Mr Wombo also liked to convert parts of plants into convoluted brain-like structures.

Figure 20. Mr. Wombo draws a brain.

 Image Composite. As discussed above in this blog post, Gemini generated some images that were intended to be depictions of how an artist might draw an alien device. I modified one of those images by inserting 1) a human brain that Gemini had also generated and 2) three human figures. One of the resulting images is shown to the left in Figure 20.

For Figure 20, the three small brain bits near the bottom of the image and the three green brain-like fruits along the upper left side were all manually pasted into the scene. I also manually added in the eraser on the end of the pencil. Originally there were sharpened pointy tips on both ends of the pencil. Mr. Wombo seemed to prefer having these big brains floating on the surface of water.

Brain tissue by Mr. Wombo.

 R U Hungry? Some of the images that Mr. Wombo generated for this scene were rather disgusting, such as the one shown to the right. My text prompt included: "a large human hand is holding a pen and drawing a human brain", but in this image the hand got seriously morphed into the contorted pink stuff above their heads.

Mr. Wombo draws a brain-hat.

In the image shown to the left, Mr. Wombo merged the woman and the big brain, giving her what I'll call a brain-hat. Maybe after eating some brain food, you need to take a nap and dream about how big your brain could grow. This is a great illustration of the stable diffusion process that can make it almost impossible for AI-image generating software like WOMBO Dream to keep two nearby objects in a scene separate.

Repair needed.

In the image shown to the right , I imagine that the woman with her brain falling out is trying to encourage Mr. Wombo to repair her head.

Giant with orange hair.

 More Orange.
By having asked Mr. Wombo to draw a giant hand in these jungle scenes, I received many AI-generated images with green giants lurking in the forest. For the image that is shown to the left, Mr. Wombo gave this giantess eye-catching orange hair. This is apparently the valley of the green giants and the large brains. 

Shocks Brain.

There was an episode of Star Trek called "Spock's Brain". In that episode, after his brain was stolen, Spock could walk around by "remote control" like a drone being guided by signals from a joystick. 

For lack of a better title, I'll call the scene that is depicted in the image to the right "Shocks Brain". Maybe the woman on the far right is delivering electric shocks to the pink and green brain. Bonus points for Mr. Wombo: a rare bare breast, sadly only seen in profile view.

I Want Your Brain.

I turned one of Mr. Wombo's "big brain" creations into an imaginary book cover (see the image to the left). I resisted the temptation to make the title be "I Want Your Brain!", but I did add the commentary ( in white text ) about artificial intelligence and included an exclamation mark. The full sized image can be viewed either here or here.

The image that is shown to the left is actually a composite image; the pink "brain-hat" in the lower left was from one AI-generated image while the green "brain fruit" on the right was from another image, the same one in which the brunette had her hand on the blond's head.

Contemplating the power of the human brain.

My text prompt called for a human hand to be in these scenes, holding an implement and drawing a brain. In the image that is shown to the right, Mr. Wombo made the hand hold a small brain and bring it to the attention of the women while the larger brain was still there in the background.

Next: Science Fiction Baseball.

Search for the elusive book-scroll. Visit the Gallery of Movies, Book and Magazine Covers