Archive for February, 2022


Not a review of “Superintelligence”

I’m reading this book “Superintelligence” by Nick Bostrom (I apparently have the 2014, not the 2016, edition). This isn’t a review, because I haven’t nearly finished it yet, but I have some Thoughts.

First of all, the book is taking far too long to get to the “What we should do about the prospect of things that are sufficiently smarter than us coming to exist nearby?” part. I’ve been plodding and plodding through many pages intended to convince me that this is a significant enough probability that I should bother thinking about it at all. I already believed that it was, and if anything the many many pages attempting to convince me of it have made me think it’s less worth worrying about, by presenting mediocre arguments.

(Like, they point out the obvious possibility that if a computer can make itself smarter, and the smarter it is the faster it can make itself smarter, then it can get smarter exponentially, and that can be really fast. But they also note that this depends on each unit of smarterness being not significantly harder than the last to achieve, and they give unconvincing external arguments for this being likely, whereas the typical inherent behavior of a hard problem is that it gets harder and harder as you go along (80/20 rules and all), and that would tend to prevent an exponential increase, and they haven’t even mentioned that. Maybe they will, or maybe they did and I didn’t notice. They also say amusing things about how significant AI research can be done on a single “PC” and maybe in 2014 that was a reasonable thing to say I dunno.)


But anyway, this isn’t a review of the book (maybe I’ll do one when I’m finished, if I ever finish). This is a little scenario-building based on me thinking about what an early Superintelligent system might actually look like, in what way(s) it might be dangerous, and so on. I’m thinking about this as I go, so no telling where we might end up!

The book tends to write down scenarios where, when it becomes Superintelligent, a system also becomes (or already is) relatively autonomous, able to just Do Things with its effectors, based on what comes in through its sensors, according to its goals. And I think that’s unlikely in the extreme, at least at first. (It may be that the next chapter or two in the book will consider this, and that’s fine, I’m thinking about it now anyway.

Consider a current AI system of whatever kind; say GPT-3 or NightCafe (VQGAN+CLIP). It’s a computer program, and it sits there doing nothing. Someone types some stuff into it, and it produces some stuff. Some interesting text, say, or a pretty image. Arguably it (or a later version of it) knows a whole lot about words and shapes and society and robots and things. But it has no idea of itself, no motives except in the most trivial sense, and no autonomy; it never just decides to do something.

So next consider a much smarter system, say a “PLNR-7” which is an AI in the same general style, which is very good at planning to achieve goals. You put in a description of a situation, some constraints and a goal, and it burns lots of CPU and GPU time and gets very hot, and outputs a plan for how to achieve that goal in that situation, satisfying those constraints. Let’s say it is Superintelligent, and can do this significantly better than any human.

Do we need to worry about it taking over the world? Pretty obviously not, in this scenario. If someone were to give it a description of its own situation, a relatively empty set of constraints, and the goal of taking over the world, perhaps it could put out an amazing plan for how it could do that. But it isn’t going to carry out the plan, because it isn’t a carrier-out of plans; all it does is create them and output them.

The plan that PLNR-7 outputs might be extremely clever, involving hiding subliminal messages and subtle suggestions in the outputs that it delivers in response to inputs, that would (due to its knowledge of human psychology) cause humans to want to give PNLR-7 more and more authority, to hook it up to external effectors, add autonomy modules to allow it to take actions on its own rather than just outputting plans, and so on.

But would it carry out that plan? No. Asking “would it have any reason to carry out that plan?” is already asking too much; it doesn’t have reasons in the interesting sense; the only “motivation” that it has is to output plans when a situation / constraints / goal triplet is input. And it’s not actually motivated to do that, that is simply what it does. It has no desires, preferences, or goals itself, even though it is a superhuman expert on the overall subject of desires, preferences, goals, and so on.

Is the difference here, the difference between being able to make plans, and being able to carry them out? I don’t think it’s even that simple. Imagine that we augment PLNR-7 so that it has a second input port, and we can bundle up the situation / constraints / goal inputs with the plan output of the first part, feed than into that second slot, and PLNR-7 will now compare the real world with the situation described, and the plan with whatever effectors we’ve given it, and if it matches closely enough it will carry out the plan (within the constraints) using its effectors.

Say we give it as its only effectors the ability to send email, and make use of funds from a bank account containing 100,000 US dollars. We give it a description of the current world as its input, a constraint that corresponding to being able to send email and spend a starting pool of US$100,000, and a goal of reducing heart disease in developing countries in one year. It thinks for awhile and prints out a detailed plan involving organizing a charitable drive, hiring a certain set of scientists, and giving them the task of developing a drug with certain properties that PLNR-7 has good reason to think will be feasible.

We like that plan, so we put it all into the second input box, and a year later heart disease in developing countries is down by 47%. Excellent! PLNR-7, having finished that planning and executing, is now just sitting there, because that’s what it does. It does not worry us, and does not pose a threat.

Is that because we let humans examine the plan between the first-stage output, and the second-stage effecting? I don’t think that’s entirely it. Let’s say we are really stupid, and we attach the output of the first stage directly to the input of the second stage. Now we can give it constraints to not to cause any pain or injury, and a goal of making the company that built it one billion dollars in a year, and just press GO.

A year later, it’s made some really excellent investments, and the company is one billion dollars richer, and once again it’s just sitting there.

Now, that was dangerous, admittedly. We could have overlooked something in the constraints, and PLNR-7 might have chosen a plan that, while not causing any pain or injury, would have put the entire human population of North America into an endless coma, tended to by machines of loving grace for the rest of their natural lives. But it didn’t, so all good.

The point, at this point, is that while PLNR-7 is extremely dangerous, it isn’t extremely dangerous on its own behalf. That is, it still isn’t going to take any actions autonomously. It is aware of itself only as one element of the current situation, and it doesn’t think of itself as special. It is extremely dangerous only because it has no common sense, and we might give it a goal which would be catastrophic for us.

And in fact, circling back around, the book sort of points that out. It tends to assume that AIs will be given goals and effectors, and notes that this doesn’t automatically give them any kind of instinct for self-preservation or anything, but that if the goal is open-ended enough, they will probably realize in many circumstances that the goal will be best achieved if the AI continues to exist to safeguard the goal-achievement, and if the AI has lots of resources to use to accomplish the goal. So you end up with an AI that both defends itself and wants to control as much as possible, not for itself but for the sake of the goal that we foolishly gave it, and that’s bad.

The key step here seems to be closing the loop between planning and effectuating. In the general case in the current world, we don’t do that; we either just give the AI input and have it produce symbolic output, or we give it effectors (and goals) that are purely virtual: get the red block onto the top of a stable tower of blocks on the virtual work surface, or get Company X to dominate the market in the marketplace simulation.

On the other hand, we do close the loop back to the real world in various places, some having to do with not-necessarily-harmless situations like controlling fighter jets. So that’s worth thinking about.

Okay, so that’s an interesting area identified! :) I will watch for, as I continue to read the book, places where they talk about how an AI might get directly attached to effectors that touch the real world, and might be enabled to use them to carry out possibly Universal Paperclips style real-world goals. And whether not doing that (i.e. restricting your AIs to just outputting verbal descriptions of means toward closed-ended goals) might be a Good Thing To Do. (Although how to you prevent that one jerk from taking the Fatal Step in order to speed up his world domination? Indeed.)


Pencil on White

Modern pencil sketch on lined paper, graceful power
Modern pencil sketch on lined paper, dancing wind
Modern pencil sketch on white paper, dancing zephyr
Modern pencil sketch on white paper, swirling phantom
Modern pencil sketch on white paper, joyful motion
Modern pencil sketch on white paper, elegant choreography
Modern pencil sketch on white paper, dancing zephyr
Modern pencil sketch on lined paper, ballet choreography
Modern pencil sketch on lined paper, battle choreography
Modern pencil sketch on lined paper, ballet choreography
Modern pencil sketch on lined paper, pair dance choreography
Modern pencil sketch on lined paper, royal dance choreography
Modern pencil sketch on lined paper, dancing air notes
Modern pencil sketch on lined paper, loving grace in movement
Modern pencil sketch on lined paper, strength
Modern pencil sketch on lined paper, circle of dance
Modern pencil sketch on white paper, wolves of air

The Colors of Chaos

I chose the title “The Colors of Chaos” randomly for an experiment with Gustave Doré, and that went so well (and so amusingly) that I tried a bunch of other artists as well. My impression is that the AI knows all of these artists, and can at least gesture in the direction of their styles. This is more difficult, and the results perhaps less interesting, for artists who would not generally deal with the colors of chaos! But overall I’m pleased with this tiny collection.

In the style of Gustave Doré
In the style of M. C. Escher
In the style of R. Crumb
In the style of Thomas Cole
In the style of Pablo Picasso
In the style of Hieronymus Bosch
In the style of Salvador Dali
In the style of Pieter Brueghel the Elder
In the style of Lisa Frank
In the style of Grandma Moses

Two Very Meta Books

Pulling my head out of the Infinite Art Museum of NightCafe, I will talk (perhaps briefly) about two books that I read recently, both of which were extremely meta.

This is probably because I tend to buy books that sound meta from the back-cover text and so on, likely on some theory that they are more likely to be clever or surprising or delightful than something more normal and/or less meta.

This theory may or may not be true.

I bought both of these, along with some others, at Split Rock Books the other day, when we actually went there physically, M and I and the little boy, in our first venturing out together for fun in maybe literally years I’m not sure. One of the main things we did was buy books at Split Rock Books, because apparently the huge quantities (at least in my case) of bought-but-not-yet-read books we already have isn’t yet huge enough.

Spoilers for both of these books, I suppose, in some sense. Or perhaps not. You will learn, at most, that not a lot that I found noteworthy will happen in the first one after the initial framing, and that a particular puzzling thing will happen in the translator’s notes to the second one.

The first book is called “[title]” and the author is given as “[name of author]”, which makes it rather difficult to find on the web. The one clue is a note that “cool-sounding indie label” (or equivalent) is owned by Sublunary Editions (perhaps it turns out you can’t publish a book without giving at least that one piece of true data), which led to a page about it, which aside from proving that it actually exists doesn’t provide any new information.

The page does give a feeling for the overall experience of reading the book, in that it never really abandons the meta perspective, and only indirectly tells us a few things about the Protagonist, the Author, the Narrator, and one or two (possibly imaginary) other characters in the book.

Nothing really happens in the book; the Protagonist appears to become very upset at some points, but as one is so removed from the direct action by the meta gimmick, I didn’t find it especially evocative. The gimmick is sort of cute, and in a way it was a fun read, but it never really went outside the box, or otherwise advanced beyond a few obvious (if interesting) challenges to traditional assumptions about the reader, the writer, the characters, narrative, books, and so on.

The second book, which isn’t nearly as obviously meta, is Mrs. Murakami’s Garden, by Mario Bellatin. Or probably by Mario Bellatin. Probably written in Spanish and translated to English by Heather Cleary. Assuming Mario Bellatin and Heather Cleary both exist.

The text is a little odd, a description of various things happening in a quasi-Japanese place that is not Japan. It has various small earnest footnotes defining some of the Japanese and pretend-Japanese words in the text, but also the word “Formica”. Some of the footnotes refer the reader to footnotes later in the text. The story is basically the life of a woman who is constrained and not constrained in many ways, familiar and exotic, and how she allows and doesn’t allow various constraints to constrain her. It was pleasant to read.

Then there’s a translator’s note at the end, that implies subtly that Bellatin himself translated the text to Spanish from some third language, and also undercuts itself by suggesting that the original from which he translated it was illegible.

Did the translator add that on their own hook? Was it some collaboration with Bellatin? Does the translator exist, or is the translator’s note written by Bellatin also? Was the entire work in fact written in English? Or does Bellatin not exist, and is Cleary the true author of the entire thing? Or was it all written by some third person, with Bellatin and Cleary as characters of an implicit frame story?

I investigated this on the web, violating various theories of textual criticism and obeying others, and I’ve decided that most likely Bellatin really did write the book in Spanish, and Cleary really did translate it to English and add the odd stuff in the translator’s note just for fun. Whether Cleary had Bellatin’s blessing, I dunno.

But I admit it was fun to be confused by! And it’s fun to know I could easily be wrong.


Ink on Paper

I am having Various Thoughts about what these mean for the nature of creativity, the relationship between the artist and the viewer, and all like that. Whoosh!

(The US copyright office just ruled that an AI can’t hold copyright in stuff; but we humans probably still can, as long as we don’t make too big a deal about how the AI really did all the work, heh heh.)

An Old Riverboat
Lights Along The River
Thoughts of the Land
Thoughts of the Country
Thoughts of the Past
Reaching High
Deep Earth
The Marketplace (with shopping carts!)
Quiet Valley; Birds Resting

A few bonus smaller ones:

The Cathedral
Stone Pillars
Signs and Omens

And finally a travel poster (not that we want too many tourists!).

Quiet Valley; Travel Poster

There are more, of course. A few more that I’ve already made (found? discovered? elicited?), many more that I will soon, and an unthinkably huge number that exist potentially, just keystrokes and a few minutes away, the vast majority of which no one will ever see. I don’t have a way to relate to this, mentally! How do I imagine, conceptualize, this huge collection, and this tiny subset of it that I have seen, that anyone will ever see?


Imaginary Isles

Sometimes I try a random prompt that springs into my head, and fun unexpected things happen. Why this particular style of illustration results from things like “Detailed colored ink on paper, illustration of … island”, I dunno! Maybe there is a subgenre, or a few prolific flickr posters?

(Harder to search the web for similar images than it is to search for texts similar to GPT-3 outputs, but nothing obvious came out when I tried.)

But anyway, I kinda love these isles, from the dreams of a neural network… :)

Sunrise. Suns scattered around like big egg yolks.
The Towns
Sunset; good night!

And a few bonus images, with slightly smaller / sketchier settings. I wanna go!


And all like that there

Back in the day (and especially back in the day), I used to just, y’know, chat a lot more than I do now. Rather than posting a particular thing about a particular thing, or a bunch of AI-generated images or whatever.

Speaking of which:

Ink on Rice Paper: Cozy

Why that is associated with “cozy”, I don’t know. Which is part of the fascination of this; exploring the odd mind of the AI. (See also of course GPT-3 and additionally Semantle for that matter, in the textual sort of area.)

I don’t think I’ve linked to semantle before, so there you go! I’ve become very much in the Wordle habit (even in the New York Times period here) (haven’t lost one yet!), and Semantle is also fun. It’s much harder, but on the other hand you get an infinite number of guesses. I realize in writing this that I got distracted and didn’t get yesterday’s, although I was within like two or three words of it.

(Long pause here while I do today’s Semantle in a mere 442 guesses, hahahaha. Toward the end there I was just typing random words that sprang to mind. In retrospect, though, it makes sense and I should have gotten it quicker.)

I also I used to use other fonts and stuff more often, because that is fun, and I used to write in raw html rather than in this WordPress environment here, which gets easily confused if you try to do anything too fancy, and often just deletes random markup that one might add in raw mode, oh well. Also it often looks radically different on my phone than on the computer here, in ways that I don’t have the patience or energy to look into to understand.

(Like, will this be in a different color everywhere? I dunno!)

I am reminiscing about Back In The Day, because for some reason that I can’t recall at the moment I was digging around in the ancient weblog on the personal site (whose front page still says “COMING SOON” and I really ought to fix that) looking for a particular funny story, and I read all various old entries while looking for it.

And also, I found it! Here is a copy of it, as well as that link there. The context (also interesting in itself) is a journal that I needed to get to someone in Indiana.

Then after lunch I took it down to the “Post Office”, where a lady agreed to take it to Indiana for me. I also had to buy her a little paper package to carry it there in. (That seemed sort of odd; why didn’t she just factor the package into the price of the service, the way she presumably does with gas fare and stuff?)

I’m not sure she’s really got a viable business model going here; the price was so cheap! I mean, she agreed to take the journal to Indiana (a particular place in Indiana, even) for like six bucks (including the price of the little paper package). I gather that there’s an economy-of-scale thing here, that she waits until she’s got a bunch of things all going to Indiana, and takes them all at once, to save on travel costs. But she also promised to get it there in just two days, so if no one else comes in and wants her to take something to Indiana by Wednesday, she’ll have to go there on just my six bucks, and there’s no way that that’ll even cover her costs.

(She’s probably got to pay someone to keep the place open while she’s going to Indiana and back, also; and then there’s rent, and utilities, and all sorts of other stuff.)

Probably she’s selling at a loss right now to get people used to being able to send things cheap, and she’ll raise prices later, once she’s got mind share. Sorta like did so well!   *8)

Still, it was nice that she was still in business today, so I didn’t have to, like, drive to Indiana myself…

Apparently this lady or her successors are still in business, which is nice!

That journal was part of a project which is probably (although not certainly) the same as this 1000 Journals Project, about which there is apparently a book and a documentary and stuff, which is pretty cool. I had (briefly) number 278, about which the old weblog talks a bit. Do go take a look! I wonder if it still survives, somewhere out there. Or even in the book or documentary!

There is some extra space for eyes

So that was all fun to discover and reminisce about. (Ha, I’d forgotten that I scanned the whole thing and burned it onto CD (remember “CDs”?) and included a copy of it in a little paper slipcase inside the journal itself. How clever of me!!

Hi-Tech Comix!

One thing that’s rather different from Back In The Day (in addition to having less control over the HTML, and not being in the habit of doing one catch-all entry per day, with the date as the title), is that I no longer feel guilty about putting more-or-less-huge images in the weblog here. On the assumption, I guess, that hardly anyone is reading this on like a 4096 bps modem or something. (But if anyone is, or is otherwise bothered by all of the huge images, do let me know! You can even leave a comment right here in the weblog and I might notice it.)

I am still very enamored of NightCafe and the inexpressibly vast universe of images that it’s willing to create. It feels (still) like I’m a tiny kid wandering through a vast library of lavishly illustrated books, dashing joyfully from shelf to shelf, delighted and overwhelmed.

I felt roughly that way about GPT-3 for quite a while, too, only with words instead of pictures; and, as the structure of these words hints, I don’t really feel that way anymore, at least not at the moment. I’m not certain how or why, but one theory that I have is that I gradually came to realize that there is no “there” there; that is, as far as I can tell, when interacting with a GPT-3 or NovelAI model, it’s easy to feel like the funny and crazy and suggestive things make sense in light of some thoughts or model of the world or way of thinking, and that by continuing to interact with it, one will come to know more about that way of thinking, and that’s cool. But after interacting with it for awhile, that seems less likely, or at least it’s more like “this writes stuff superficially like random fanfiction but without even that much sense to it” than it is like “wow, this crazy alien being I’m talking to sure has some interesting ideas!”.

I can easily imagine the same kind of thing happening with NightCafe; looking at the panel from “Hi-Tech Comix!” up there, my first fascinated reaction is that those enigmatic machines and blurred speech bubbles must be About Something, must be Saying Something in some enigmatic language and universe into which the image offers us a preliminary glimpse. But perhaps one ultimately realizes that it isn’t; it’s just a mindless imitation of stuff on Flickr or whatever.

If indeed that’s all it is. :)

What else what else? The family still exists and prospers. The little daughter is working remotely from Queens for an IT company (“IT company”) and going to tango events again now that there are vaccines and things are opening up somewhat. M and the little boy and I still live here in the ‘burbs, but I went in to work in Chelsea a couple of days last week (yay!), and I intend to do the same this coming week.

It’s wild how, after living up here and barely visiting the City at all for decades, once I started working down there I fell in love with the place, and have been going somewhat crazy not being able to get in for months and months.

Ah, Chelsea, how I missed ya!

I skipped my usual bowl of cold cereal at home, and got a Bacon Egg an’ Cheese on a Roll, Salt, Pepper, Ketchup from the vendor on the way to work the other day. It was glorious.

So that is a nice completely random weblog entry, as in Back in the Day. Now I will probably generate many more images (Twitter link assuming it works, and NightCafe link similarly), and perhaps play some Computer Games (although I’m kind of plateaued on everything I play, including WoW and Satisfactory and No Man’s Sky, but that doesn’t always stop me), and perhaps read some books (Oh, I was thinking I should do a weblog entry on a couple of rather meta books that I finished lately, maybe I will some time), fight some entropy, and/or go for a walk in the sunny chilly day.

Blessed be!


The Adventures of Hugo and the Lamb

A comic strip from NightCafe. In a variety of styles.

Episode 1: We meet Hugo, and the Lamb
Episode 2: At the Bridge
Episode 3: A twist!
Episode 4: The Sign
Episode 5: Return to the Source
Episode 6: The Lamb Considers
Episode 7: The Puzzle
Episode 8: A Dark Turn
Episode 9: The Nightmare
Episode 10: The Baker You Meet In Heaven
Episode 11: Resolution
Episode 12: Returning Home

Special bonus content: Coming This Summer, Hugo and the Lamb, the Motion Picture!

Publicity Still

Manuscript with ominous drawings

… all that was found at the scene was …


The Stormracer VII

I thought it would be fun to see what the Nightcafe “Artistic” engine would do with a slightly ambiguous product name (and basically nothing else). So we present: the development of the Stormracer VII!

(The caption on each image is the entire prompt passed to the AI engine.)

The big wheel in the concept art was apparently dropped during development.

One thing (of many) that strikes me here, is that the logos on the three documents that have logos, seems to almost spell out the product name. That seems to say something about the level of understanding that the AI has; other aspects say similar things, more subtly.


AI-generated AI prompts

So, obviously, I gave AI Dungeon a prompt with a few prompts that I’ve given to Nightcafe Studio:

Here are some good prompts for an AI program that generates images from text:
- vray rendering of my feelings for you
- vray rendering of diamond tentacles in space
- At the Cabaret; oil on canvas
- Illustrated plans for an interstellar cyberpunk yacht
- Victorian colored sketch, lush comfortable; Sunlight through the window
- dark neon alien city; The Cost of Memory
- Modern realism retrofuture gernsback vivid vintage street scene
- Detailed steampunk cyberpunk noir library interior

And it continued:

- Detailed art deco fantasy architecture
- Detailed sketch of magic artifact
- Fantasy gargoyle, black and white
- vray rendering of SciFi city
- Medieval cathedral interior
- Sketchy cyberpunk bar interior; The Bartender
- Digital concept painting, renaissance era
- Colorful fantasy city scene
- Sketchy cyberpunk bar interior; The Bar
- Fantasy medieval castle exterior
- High fantasy city at night, watercolor painting
- Detailed concept art of alien city

And it would have continued except that I seem to have a thing about dozens lately.

So here we go! Images generated by an AI, from prompts generated by an AI. :)

Detailed art deco fantasy architecture
Detailed Art Deco Fantasy Architecture
Detailed sketch of magic artifact
Detailed sketch of magic artifact
Fantasy gargoyle, black and white
Fantasy gargoyle, black and white
vray rendering of SciFi city
vray rendering of SciFi city
Medieval cathedral interior
Medieval cathedral interior
Sketchy cyberpunk bar interior; The Bartender
Sketchy cyberpunk bar interior; The Bartender
Digital concept painting, renaissance era
Digital concept painting, renaissance era
Colorful fantasy city scene
Colorful fantasy city scene
Sketchy cyberpunk bar interior; The Bar
Sketchy cyberpunk bar interior; The Bar
Fantasy medieval castle exterior
Fantasy medieval castle exterior
High fantasy city at night, watercolor painting
High fantasy city at night, watercolor painting
Detailed concept art of alien city
Detailed concept art of alien city

Not extremely notable: the images are more or less what we’d expect from the prompts, which are more or less what we might expect from the meta-prompts.

But still! Fun!


Victorian Sketches

Victorian colored sketch, lush comfortable; Sunlight through the window
Victorian colored sketch, lush comfortable; Strolling in the sunny street
Victorian colored sketch, lush comfortable; Cakes in the Bakery Window
Victorian colored sketch, lush comfortable; Women's Hats
Victorian colored sketch, lush comfortable; Sunny Garden Gate
Victorian colored sketch, lush comfortable; Flowers by the Roadside
Victorian colored sketch, lush comfortable; Library Interior
Victorian colored sketch, lush comfortable; Leather Lounge
Victorian colored sketch, lush comfortable; City Street Scene
Victorian colored sketch, lush comfortable; The view from the window
Victorian colored sketch, lush comfortable; The Street Leads Home again
Victorian colored sketch, lush comfortable; Sunset over the Palace

Dark Neon Alien City

Its name is inaccessible to us.

dark neon alien city
Welcome to █████
dark neon alien city; Public Square
Public Square
dark neon alien city; The Squalid Bazaar
The Squalid Bazaar
dark neon alien city; The Squalid Warriors
Warriors of the ████x██
dark neon alien city; The Cost of Memory
The Cost of Memory
The Lake of Industry
The Lake of Industry
dark neon alien city; The Palace of the Three
The Palace of the Three
dark neon alien city; Library Interior
Library Interior
dark neon alien city; The House of Fires
The House of Fires
dark neon alien city; The King's Procession
The King’s Pro████
dark neon alien city; Surreal Hotel
██████ Hotel
dark neon alien city; Rainbow Sunset

Travel Posters from Nowhere

One from each of the realms / prompt-prefixes that we’ve been playing with.

The NightCafe “Creative” engine (VQGAN+CLIP) is pretty darn good at posters, as long as you don’t mind the text being illegible.

Yeni Cavan Travel Poster #yenicavan
Yeni Cavan
Cogspin Travel Poster #cogspin
Neocine City Travel Poster  #neocine
Neocine City
Miller's Bend Travel Poster #millersbend
Miller’s Bend
Kranija Travel Poster #kranija
Technopolitan Travel Poster  #technopolitan
Travel Poster  #nightlife
Night Life
Travel Poster #moebius
In the Style of Moebius
Le Monde Moderne: Travel Poster #mondemoderne
Le Monde Moderne
dark neon alien city; Travel Poster
Dark Alien City

That last one is a new prefix that I was just playing with a little. But it made a good travel poster for it!

I will probably (or perhaps) write more words about all of this stuff that the AI is making; but at the moment I’m just wallowing in all of the imaginary pictures…


12 Jazz Posters from the Imaginary World


That Mœbius Style

We’ve been playing around with styles in NightCafe Studio (obviously), and recently with the styles of some specific artists.

(This is still all in “brand-new image” mode; somehow I’m less interested in the “style transfer” mode, where you ask the AI to apply some style to an existing image. That might be cool, also, but my primary fascination at the moment is more “let’s look inside the mind of this crazy emerging AI” and less “let’s use this AI as a cooler Photoshop”.)

One artist that the NightCafe AI (the “VQGAN+CLIP” one especially) does a notable job with is the French cartoonist (“cartoonist”) Jean Giraud, aka Mœbius. I discovered his work in Heavy Metal magazine back in the day; while I was often not crazy about the stories, the art was novel and exotic and immersive.

So anyway! I pointed NightCafe Studio at a bunch of prompts starting with “Detailed realistic image in the style of Moebius”, and this is the kind of thing that came out; the dataset that the AI was trained on must have had a lot of stuff by or referencing him. Enjoy!

Detailed realistic image in the style of Moebius; riverside
Moebius bedroom
In the style of Moebius: Starport
Starport buildings
In the style of Moebius: Starships I
Starships I
In the style of Moebius: Starships II
Starships II
In the style of Moebius: The Growing City
The Growing City
In the style of Moebius: How People Dress Now
How People Dress Now
In the style of Moebius: Meeting with the weapons dealer
Meeting with the Weapons Dealer
In the style of Moebius: Lift off
In the style of Moebius: By the Lake
By the Lake
In the style of Moebius: Meteorite
Weapons of War
Weapons of War

Night Life

Frightened business men with lamps
concept art film noir, library with lamps
concept art film noir, wonderful mechanisms
concept art film noir, street scene
concept art film noir, dancers
concept art film noir, cafe
concept art film noir, We are so happy to see you
The Latest Thing
concept art film noir, The nuns return to the air
concept art film noir, coffee
concept art film noir, Night Life
concept art film noir, hearts of sweet gold
concept art film noir, singers of fire and ice
concept art film noir, the hum of the crowds
concept art film noir, when the parade goes by
concept art film noir, finally to sleep


I could just produce one of these posts a day, for months. Or one an hour, for that matter. Will I get tired of it? Probably. Eventually. But not today! As always, courtesy of NightCafe Studio.

Retro-futurism, Hugo Gernsback style.

Technopolitan #technopolitan
Technopolitan street #technopolitan
The Bustle of Modern Life
Technopolitan downtown #technopolitan
Heading downtown
Technopolitan library #technopolitan
Knowledge Is Progress!
Technopolitan dancers #technopolitan
Sophisticated Entertainment
Technopolitan Lovers #technopolitan
Modern Love
Technopolitan technology #technopolitan
Technology Makes The World Better!
Technopolitan city hotel #technopolitan
City Living
Modern realism retrofuture gernsback vivid vintage nightclub
Technopolitan Nightlife

(Interestingly, I failed to get a sunset image out of it that I felt was sufficiently in-theme. Maybe I should have kept trying, but I think it’s telling that while “nightclub” and “technology” and so on hit immediately, “sunset” is more of a stuggle. Not much room for nature in this aesthetic?)


Libraries and Sunsets

Continuing our explorations of NightCafe Studio (still like 30 credits left!), we now have five (what?) microstyles based on prefixes of the prompts sent in to the engine, and imagined as five fictional places:

  • Yeni Cavan, a sort of noir steampunk / cyberpunk sprawl,
  • Cogspin, a lighter and happier steampunk region,
  • Neocine City, a bright nanotech-based cybermetropolis,
  • Miller’s Bend, a generic Old West town on a river,
  • Kranija, a dark but vivid dreamworld, perhaps somewhere in the Carpathians.

There are a few more in the offing (a fantasy video game, and a set of pencil sketches), but we’ll take these five for now.

For fun and to organize my thinking, let’s look at a library, and a sunset, from each of the areas.

Libraries first. For whatever reason “library interior” seems to work better than “library”, so I’ll mostly use that.

Yeni Cavan library #YeniCavan
Yeni Cavan Library Interior
Cogspin Library #cogspin
Cogspin library
Neocine library interior #neocine
Neocine City library
Miller's Bend Library (incl. horses) #millersbend
Miller’s Bend library (incl horses)
Kranija town library #Kranija
Kranija town library

It’s noteworthy, I think, that the Miller’s Bend library is probably quite a bit too large (and has horses wandering around inside it (and a giant hole in the roof)), as is perhaps Kranija’s, and that Neocine City’s library would likely not contain recognizable books at all; maybe something more like crystals with writhing patterns within. But the AI doesn’t really understand that stuff, it just makes what it thinks of as a normal library, in whatever artistic style the prefix suggests.

To be fair, leaving out the word “interior” in some cases produces… different things:

Bright detailed nanotech cyberpunk library
Neocine City Library

This detail of the Neocine library shows not just paper-like things, but also odd little organo-techie things, which seems appropriate.

old west detailed oil painting library
Library dream, Miller’s Bend

And this might be the actual Miller’s Bend library, a few books stacked untidily on a desk, with dreams of the open plains and also a bigger library with neat shelves, floating above. (I like this one!)

On to the five sunsets, an appropriate way to close out this post.

Yeni Cavan Sunset #yenicavan
Yeni Cavan Sunset
Cogspin Sunset #cogspin
Cogspin Sunset
Bright detailed nanotech cyberpunk sunset
Neocine City Sunset
Sunset, Miller's Bend #millersbend
Sunset, Miller’s Bend
Kranija sunset #Kranija
Kranija Sunset

I rather adore all of these…


Styles in NightCafe Studio

Although I’m toying with the idea that all of these text-to-image GAN images have a similar high-level style (intriguing, everything a bit off, not quite making sense), I’ve also been playing with more specific styles in NightCafe Studio in particular (their landing page really ought to notice that one is already logged in).

Each of the images of Yeni Cavan comes from a prompt that is “Detailed steampunk cyberpunk noir” followed by one or more additional words to more or less get some specific thing, like “hotel” or “mechanism” or “street scene”. And the result is quite consistent in overall style, colorway, and so on, I’d say:

General dark-brass overall, lighting mostly magenta with some glary greens and occasional blues, lots of smoke / fog, and of course lots and lots of leather and mysterious quasi-organic machinery.

It turns out that other descriptive phrases do the same sort of thing, but with different subtleties, even if the basic genre is similar.

For instance, the prefix “Bright detailed nanotech cyberpunk” produces this sort of thing:

Bright detailed nanotech cyberpunk street scene
Bright detailed nanotech cyberpunk dancers
Bright detailed nanotech cyberpunk dance party
Bright detailed nanotech cyberpunk restaurant and bar
Bright detailed nanotech cyberpunk lovers
Bright detailed nanotech cyberpunk mechanisms

A significantly different palette and overall feel, even in roughly the same SFish genre.

Closer to Yeni Cavan, but still significantly different, the prefix “Photorealistic happy steampunk” gives for instance:

Photorealistic happy steampunk skyline
Photorealistic happy steampunk vehicles
Photorealistic happy steampunk street scene
Photorealistic happy steampunk spaceship
Photorealistic happy steampunk lovers

That last one (which is, by the way, “lovers”: note the cute heart!) could almost be from Yeni Cavan, but seen with the other ones from the same prefix, it’s clearly closer in spirit to those; still brass and leather, but brighter, more (what?) Victorian, and less threatening.

So, yeah, I’ve been burning through a whole lot of the credits that I’ve earned and bought on NightCafe. :) And I’m wishing for better ways to organize my creations within their site (although to be fair I’ve been ignoring the various suggestions to download things or install apps, and just used the web pages, so maybe I’m missing out on features).

Shall we try a few images from another randomly-selected prefix, while we’re here? Maybe something significantly different, like “old west detailed oil painting” (I don’t know why I like “detailed” and “photorealistic” things; I guess I want to avoid sort of abstract blobs). Let’s try a few!

old west detailed oil painting street scene
Street Scene
old west detailed oil painting dancers
old west detailed oil painting lovers
Lovers (wut?)
old west detailed oil painting saloon interior
Saloon Interior

Pretty plausible, I’d say. There appear to be horses in the saloon, and I don’t know what is going on with Lovers, but it feels like there’s a coherent theme going on. It would be interesting to see what a subtle change of prefix, to like “1789 frontier realistic oil painting” say, would do. Maybe later. :)