On page 1 of Jack Vance's novel
The Star King, there is part of an interview of
Smade, the proprietor of Smade's Tavern, the only building on a distant exoplanet known as Smade's planet. Smade is asked, "
Whatever impelled you to settle here? A rather dismal world, on the whole, isn't it?"
Smade replies: "Beauty is in the eye of the beholder. I don't care to run a vacation resort." Twenty pages later, Vance's character Hildemar "Beauty" Dasce arrives at the tavern. Back in 2023 when I began experimenting with text-to-image software, I tried to get Mr. Wombo to make a depiction of Dasce. The image shown in Figure 1 is my most recent attempt to get WOMBO Dream to do justice to the depiction of Dasce that was made by Ed Emshwiller in 1963. The free version of WOMBO Dream has a strict length limit for text prompts, so I used: "a science fiction character with red skin, spiky red hair, and blue eyes without eyelids. Around his eyes there is a mechanical apparatus: retractable metal shields and tiny nozzles designed to spray moisture onto his lidless eyes. His face has chalk-blue cheeks and a cleft nose".
Bloody AI. I've yet to reach a prompt length limit with
ImageFx, so to generate the image of Dasce that is shown in
Figure 2, I used a longer text prompt with ImageFX: "
Retro pulp science fiction illustration in dramatic Ed Emshwiller style. Close-up of the grotesque character Beauty Dasce with bright crimson skin, spiky red mohawk hair, and blue eyes without eyelids. Around his eyes, add mechanical apparatus consisting of retractable metal shields and tiny nozzles designed to spray moisture onto his lidless eyes.
His face should have chalk-blue cheeks, cleft nose, and a tall round head on a red tubular neck. Include detailed texturing to show the interface between flesh and the eye-moistening equipment. In the background, a tall glowing cyan/blue cylindrical tower or beacon rises against a dark starry night sky. A figure in a mustard yellow jumpsuit stands near the tower's base structure with antennae, while another figure in similar attire lies dead on the rocky alien terrain with a small pool of blood. Use high contrast, textured brushwork, and vintage sci-fi horror aesthetic reminiscent of 1960s pulp magazine covers", which had been suggested by Claude. The type of background that was generated by ImageFX is shown to the right. I did not like all the blood that was made by ImageFX, so I had Mr. Wombo make an alternative background for Figure 2.
 |
SSS Dasce
|
I provided
Whisk with the inputs that are shown to the left. The "subject" image was
Figure 2 and the "scene" image was generated by ImageFX using the text prompt: "
A stunningly detailed, photorealistic rendering of a science fiction exoplanet surface where a tall slim woman in a mustard yellow jumpsuit stands near a futuristic tower's base, while another figure in similar attire lies dead on the rocky alien terrain within a small pool of blood. The futuristic tower has glowing windows and fractal complex features including antennae. Use high contrast, textured brushwork, and vintage sci-fi horror aesthetic reminiscent of 1960s pulp magazine covers". The initial image generated by Whisk is shown below in
Figure 3.
 |
Figure 3. Dasce's planet by Whisk. Version #1.
|
Modified Description. I changed the original Whisk-generated description for this "storyboard" to:
"
A stunningly detailed, photorealistic rendering of a science fiction exoplanet surface featuring warm earthy browns, tans, and oranges in the landscape, contrasted by cooler blues and greens in the night sky. A tall, beige multi-level futuristic building with antennas and glowing lights stands on a dark, rocky landscape under evening stars. In front of the building stands a tall slim woman in her twenties with beautiful shoulder-length blond hair and light skin tone, wearing a dark green jumpsuit, looks down at another woman lying on the ground near a pool of blood.
The second woman, also in her twenties with reddish-brown hair and light skin tone, wears a similar dark green jumpsuit. Close-up view of the face of a surgically disfigured man with skin that is mostly dyed red, a spiky red tuft of hair on the top of his head and blue eyes without eyelids. Around his eyes is a mechanical apparatus consisting of retractable metal eye shields and some spray nozzles that are designed to spray moisture onto his lidless eyes. His face has chalk-blue cheeks and an ugly cleft nose. His face is positioned to the side of the scene, rendered with hyperrealistic detail. The background scene behind the man’s face is detailed with a wide angle showcasing the vast landscape. The lighting suggests a late evening sun, casting long shadows". With this new description, now Whisk generated images such as the one shown in
Figure 4.
 |
| Figure 4. Dasce's planet by Whisk. Version #2. |
When I changed the Whisk "subject" image to Figure 1, Whisk generated the new "storyboard" image that is shown in Figure 5. Here is the Whisk description of Figure 1: "A painting depicting a scene in what appears to be a futuristic laboratory or space station. In the foreground, a humanoid alien with reddish-orange skin and a spiky, reddish-orange mohawk is prominently featured. The alien has large, light blue eyes, and pale, almost white, circular patches on its cheeks. It wears what looks like a pair of goggles or a head-mounted device with circular lenses. The alien's attire consists of a light blue, metallic-looking suit or uniform with visible mechanical components. To the left and slightly behind the alien, a human male, appearing to be middle-aged, with light skin and short, dark brown hair, is visible.
He is wearing a yellow spacesuit with dark accents. The background features a complex array of metallic machinery and tubes, with glowing blue lights suggesting technological activity. The overall color palette is dark and metallic, with contrasting bright blue and yellow accents. The ground appears to be a reddish-brown, possibly rocky terrain. A large, cylindrical, transparent container filled with a yellowish-green liquid is visible behind the human male. The container is part of the larger machinery in the background".
I had Mr. Wombo make some alternate versions of this "humanoid alien". The "alien" was clearly given some female features by Whisk (without being asked to do so), but I specified a "woman with spiky red hair" in the text prompt for Mr. Wombo.
 |
| Figure 5. Dasce's planet by Whisk. Version #3. |
 |
| Figure 6. Dasce's planet by Whisk. Version #4. |
Full Disclosure. The "cleft nose" in
Figure 2 was manually inserted by me. I had Mr. wombo generate an image in which the man had a cleft nose and then I pasted that nose into the ImageFX-generated image. As seen in
Fugures 3-
7, Whisk refused to make any cleft noses.
 |
SS cleft nose
|
For the Whisk-generated "storyboard" shown below in
Figure 7, I provided Whisk with the "subject" and "scene" images shown to the left. I had manually pasted in those blue eye covers (cut from another Whisk-generated image) that are shown positioned above Dasce's eyes (image to the left, top panel). In
Figure 7, I was impressed by all of the drops of water that Whisk placed running down his face. Whisk also seemed incapable of giving Dasce blue cheeks, even when I put "
His nose is cut open and he has blue cheeks" into the Whisk text field for "Add additional details".
Making use of WOMBO Dream, I modified the woman in Figure 7, giving her the appearance of Tyhry. I've previously imagined Wendy using the Reality Simulation System to visit the Asimov Reality and meet Dasce, so why not also Tyhry?
 |
| Figure 7. Dasce's planet by Whisk. Version #5. |
 |
SS Pallis
|
The complete inability of both ImageFX and Whisk to generate the features of Dasce's face (features as envisioned by Jack Vance) was disheartening. However, I moved on and had ImageFX generate an image depicting Vance's character
Pallis Atwrode. Text prompt: "
A stunningly detailed, full color, photorealistic rendering of a pretty young woman with arch piquant facial features. Her facial skin is artificially toned a shinny metallic silver. Her hair is a beautiful dark cloud of ten thousand very tiny varnished spikes. She has a wide mouth with a hint of an upward twitch at the corners. She is dressed in a brightly colored sun dress." The ImageFX-generated image of Pallis is shown as the Whisk "subject" to the left along with a composite "scene" what was also generated by ImageFX in three parts: 1) Pallis, 2) the galaxy and 3) the reception desk for the Department of Galactic Morphology at Sea Province University on Alphanor. With these inputs, Whisk-generated images such as the one in
Figure 8.
Figure 8 is actually a composite image, the silver face being copied and pasted from a second Whisk-generated image.
 |
Figure 8. Whisk-generated image of Pallis Atwrode, receptionist, the Department of Galactic Morphology.
|
 |
Figure 9. Whisk-generated image of Pallis Atwrode meeting Gersen.
|
 |
SS Gersen
|
I then used the image shown in
Figure 8 as a new "scene" image with Whisk along with a "subject" image of a man (
Gersen). I modified the description of the Whisk "storyboard" to include: "
The receptionist is looking at a man who stands in front of the reception desk. The man has short, dark brown hair, light skin, and a serious expression, wearing a black t-shirt and black leather jacket, stands in front of the reception desk with his back to the camera". With these new inputs, Whisk generated images such as the one shown in
Figure 9.
Figure 9 is actually a composite image with the head of Pallis copied and pasted from another Whisk-generated image. No matter how hard I tried to tell Whisk to give Pallis normal skin for her body and sliver skin for her face, Whisk could not do it.
Previously, I imagined Honorary Provost Gale Warpweave as Gersen's adversary in the Asimov Reality. I provided Figure 9 to Whisk as a new "scene" image along with an image of the fashion model that I previously used to represent Gale. I added to the "storyboard" description: "In the background near the wall there is another light-skinned woman with shoulder-length brown hair who is wearing sunglasses and gazing towards the man".
With these new inputs, Whisk generated images such as the one shown in Figure 10. Once again, Whisk refused to generate Pallis with a silver face and a body with normal skin tone, so I manually pasted her face into Figure 10 from another Whisk-generated image.
 |
Figure 10. Whisk-generated image of Pallis Atwrode meeting Gersen. Gale Warpweave in the background.
|
 |
| SS Nautilus |
After meeting Pallis at the Department of Galactic Morphology, they go out for dinner at an under-sea restaurant called the
Nautilus. Vance states in
The Star King that after work, Pallis re-tints her skin green. I provided Whisk with the "subject" image that is shown to the left and asked ImageFX to generate an image depicting: "
A stunningly detailed, full color, photorealistic rendering of a futuristic restaurant that is 100 feet below the surface of the ocean. An escalator carries customers to undersea dinning rooms with glass walls and sea life beyond those transparent glass walls". The resulting ImageFX-generated image is shown in the lower panel of the image to the left.
With these new inputs, Whisk generated images such as the one shown below in Figure 11. As shown in Figure 11, sometimes Whisk got confused about how to render an escalator. I pasted into a similar Whisk-generated image a depiction of Tyhry (see Figure 12).
 |
Figure 11. Broken escalator at the Nautilus.
|
 |
Figure 12. Composite made by pasting in Tyhry (wearing the blue gown).
|
 |
SS Tyhry
|
Using that composite image from
Figure 12, I allowed Whisk to generate images that included Tyhry... the first such was the one shown in
Figure 13.
I then modified the "storyboard" description to include, "A tall blond woman dressed in a beautiful blue gown stands in front of the escalator gazing in the direction of the woman who has green skin". Whisk then generated Figure 14.
 |
| Figure 13. |
Using the Reality Simulation System, Tyhry observes Gersen and Pallis at the Nautilus.
 |
Figure 14. Two escalators and two Tyhrys at the Nautilus.
|
 |
Tyhry visits the Nautilus.
|
For some reason, Whisk enjoyed making two copies of Tyhry to include in the scene inside the Nautilus restaurant. I used WOMBO Dream to make an image of Tyhry facing the camera and manually pasted that into the right hand side of the Whisk-generated image that is shown in Figure 15, covering up the "second Tyhry" that had been put in that position by Whisk, and, like the other copy, was facing away from Pallis.
The image that is shown to the right was made by WOMBO Dream. That could be a depiction of Tyhry while she is on Alphanor and following Pallis and Gersen to the Nautilus restaurant.
 |
Figure 15. Tyhry watching Pallis and Gersen at the Nautilus.
|
I was fairly satisfied with this image (
Figure 15) but I was left unable to understand why it is that on some days Whisk seems to take the task more seriously when asked to render images in a photorealistic style. I suppose if there are no photos of green women in the training set then it becomes quite difficult to generate a photo-realistic green woman. In my next blog post, I see if
Leonardo can generate a more photo-realistic depiction of Pallis in the
Nautilus.
Next: Leonardo goes green.
%20&%20WOMBO%20background.png)
.png)
No comments:
Post a Comment