The Power of Hate

I am still working on the same scene from Sunday. Damn work-life balance. Time to think about retiring.

Okay; I did go back and revise the Atlas scene to try and focus the emotional arc and make more of a coherent mini-crisis out of it. And expanded and revised the Apocalypse Log.

And hit a plot question in the scene I'm on; I couldn't figure out why this Apache writer of westerns she meets with in Cloudcroft wouldn't mention the guy who is gonna turn out to be the murderer. That's a bigger clue than I wanted, but it ain't like the whole conversation can't just be about his hat.

I'm not good about writing red herrings.

Really, I never set out to write mysteries. Turns out it is baked in to the basic premise. I wanted to do adventure (archaeological adventure but anyhow) but with real history and real archaeology. And mostly an urban setting, meaning this is a lot less human-against-nature and more human-against-human.

Being realistic about history and science bled into being realistic about how the world works, and how the world works is; if you have bad guys running around with gun in hand a hell of a lot of cops show up. Have a murder or some other big crime, and some very careful police work and forensics gets into the picture. And even if they don't have time to solve it properly (or at all), that sort of attention does a really good job of shutting out the amateur.

So the best crimes are those where not only does nobody know who dunnit, they don't even know what was done, or even if anything was done. And that means neither does my protagonist. Instead of being a "The Zebra Killer has struck for the thirteenth time -- we need to find him and stop him!" it is more the Isaac Asimov quote about the usual sound of scientific discovery; not "Aha!" but "That's odd..."

Which plays hob with having a driving plot, ticking time bomb, or for that matter having villains nice and visible and easy to hate.

I'd be failing there anyhow. Not only have I consistently failed in creating villains to properly dislike, I also have been failing in having people dislike my protagonist. I mean, that is her method. She talks people around, gets them to like her, finds ways that cooperation is more effective. She's a regular Power of Love shoujo heroine.

Making me want to work with the Blue universe. Because there I could have a protagonist who isn't being supported, who is actively being blocked, even, both by actual bad guys and by people who are technically on their side of the great game.

Didn't finish the book

Just finished Part III and there's only 15K or so left in the novel. But also, there's 15K of word-smithing and the final act of the novel and there's only two days left of the holiday vacation.

Some vacation, when it was cold and raining and everything was closed. Why can't we have that time off in spring instead?

I was starting to really crank. Got up to around a thousand words a day over the last few days. Part of that, though, is that a lot of things changed from outline to the actual scenes, and the choices that made sense for the "Duel" scene, Atlas, and the War Zone were all choices that simplified the scenes, took out the need for more research, and made the things shorter as well.

I'd been intending an epic climb back out of the silo. But as I wrote the scene, the emotional crux and the important story beat had already happened and I didn't need this for the story. And it didn't feel important enough to spend another day or two writing it, so I skipped over it. 

(I have enough trouble finding enough to say about driving down a road through flat empty desert. Spending enough pages on climbing up rusty steel in the dark to make the scene feel properly epic? I don't feel up to it. How much can you write "right foot up, then left hand up?" I don't even enjoy playing it that much, in the increasingly obligatory "Uncharted" sequences.)

That may have been a mistake. In the War Zone, dealing with the place in any detail went too far aside from the main thrust of the plot and brought up too many directions I couldn't afford to go, so I ended up just doing a very surface-level depiction. No long conversations about the economic woes and the drug epidemic. Nothing at all about the sunny side of the neighborhood, the people living their lives there (increasingly, behind security gates); no attempt at a balanced or nuanced depiction.

Which also got rid of any need for a scene with "Michael Rennie" as he warns her not to go there, or cops with a similar warning, or a local who argues for a more nuanced view. Instead I could have a fairly simple asshole who gets right to the point, show it all from a single point of view, hand Penny the next clue and letting her move on to the next stage of the plot.

I think that was the right choice. I am much less sure about how the beats worked out in the Atlas scene, and that's why I think that instead of pushing through to get as much as I can of the Bar at the End of the Universe sequence written before vacation ends, I should back off and do some higher-level planning again.

When I wrote the first book of the series, I wrote a mid-act turn-around as an essential part of the structure. This is not uncommon. If you chose to sort the story beats into three boxes and call them "acts," there are two big changes of direction in the narrative occurring at the act breaks.

Act I introduces the elements of the story (setting, story, conflict). Act II can be said to begin when the protagonist first acts. This is when the problem is now clearly stated, and the cast begins to attack it. In the more Campbellian schemes, this is when the hero leaves their village and sets out into the world (or "enters the underworld" in Campbell's terms).

Act II is the middle of a paragraph or the middle of a symphony. It is where the themes are explored. Lots of "stuff" happens but the conflict does not get resolved. Saggy middle syndrome is baked in to the very structure of the story.

The II-III act change is when things change. What the protagonists were trying to do isn't working and they have to do something different. In more of a martial story, the bad guys have been attack, attack, with the heroes on the back foot, and II-III is when they come up with a plan to take the fight to them.

But then there's the strange one. There's a nadir, when the protagonist falls and everything looks hopeless. Then something happens (often a resolution of some other plot arc) and from here to the end of the story is the rising action and climax.

Which is why it usually (or something like this) happens in the middle of the act. Because you can't have the heroes only winning for all of Act III; that would be boring.

The place I went with many of the Athena Fox stories is that this crux, coming out of this nadir, is when Penny becomes most Penny. Sometimes pure Athena Fox, sometimes her own peculiar spin, it is the moment where she stops trying to win the stupid way, and commits to winning her way. It is the Darkwing Duck moment ("let's get dangerous") -- and that show followed that model quite closely.

I already knew going into Part III that I'd messed it up. Way back in the early outline, she would have gone a little crazy when Lon was found dead, have the "Footloose" clash at Blake's Lotaburger, and over the long long drive to WIPP do a bunch of soul-searching and be back in a place of sanity when she arrived. But I couldn't make the sequence of clues or even the geography work with her also visiting one of the Atlas-F sites near Roswell, much less my late-game decision to throw in Albuquerque's "War Zone."

(I was doing some clean-up at my favorite brunch place yesterday. Five minutes after a re-wrote the Jennifer Beale namedrop, "What a Feeling" came on their radio. Laughed out loud.)

So I've sort of scattered all the emotional beats and any act-like turn-around Penny could have across the entire act. And I feel like I could tighten it and make something where she really is bringing together the pieces and coming out of it stronger and more ready to finish the adventure.

At least there's no more long drives to write.

(An actual game on Steam. I'd get it -- four bucks -- but it is a dead game now, abandoned by the independent developer).

War Story

 

Sorry, that's Battlezone, not War Zone.

Trying to see if I can push through and finish the New Mexico book over this holiday's break. That break is short, though. A lot shorter than it was when I first joined the company. 

Anyhow, I got through the Duel sequence and now the outline calls to go back to Albuquerque.

But I got hung up on the best way to present the idea of what locals call the War Zone. Some locals disagree with that name, although they aren't willing to go as far as the city, which rebranded it as the International District. Point is, crime got so bad there by the '90s the city put up barricades. As of the period of the story, it is still one of the most violent places in America.

Of course I want to paint a more nuanced picture than some depraved hive of scum and villainy. 

Up until this morning, I figured "Michael Rennie" would warn her about it before she went...and she wouldn't run into any trouble. (Penny runs into a lot of people she never gets a name for. This guy, she imagined a resemblance to the actor and lead of The Day the Earth Stood Still.) 

Before that, I was tempted for her to actually run into trouble.

Now I'm tempted to have the cops pull her over for suspiciously driving slowly around the kind of place that, in their words, a nice girl like her had no business being. I like this because she's in her mind on the run (she did flee the scene of an accident) and having to interact with the police makes a good dramatic moment.

But other than that, I don't like it. So I'm thrown back into watching videos shot by various visitors and locals (like Peter Santenello, a stalwart for this sort of thing), gathering up any kind of interesting whatever that will make having this as a location have a point. Since her next stop is near Roswell, not here.

Kay Vess is bad at her job

And so is Ubisoft, and so am I -- which is why I'm playing the game anyhow.

There are games that deliver a riveting story. Games that deliver memorable atmosphere. Games that deliver an emotional ride. And then there are comfort games. Usually open-world games with fun environments and lots of little side quests and other things to do. Games you don't try to "win" so much as go and visit. Skyrim. Horizon Forbidden West. Subnautica.

And Star Wars Outlaws. When it works, is when you are just wandering around the marketplace, hanging out in the cantina looking for interesting gossip to overhear, or sometimes out exploring.

It made me miss travel a little less. These are good environments, good enough that Ubi was actually smart enough to put in a couple of rest spots that are just there to sit and take in the scenery.

And while there is a main campaign, it isn't really an arc, there's nothing big at stake (Kay is offered membership several times in the Rebellion and turns them down), there's no character development. Kay is early Han Solo, just living the life -- the life being, steal everything that isn't nailed down, and quite a few things that had been, complete with coded locks and armed guards. Hang out with her low-life buddies, buying blaster parts from back-alley traders between betting on the fathiers and games of Sabac, and eating at street stalls.

And I'll get back to this little scene in a bit.

Kay is probably the right character for this, although she can be annoying. She's a neophyte, a jumped-up street rat trying to earn that street cred. Which is great for the player because you get to build skills, contacts, familiarity with the streets and of course that important reputation.

Which means the most important skill is putting up a good front. Acting tough so people don't mess with you, pretending experience so they will hire you, bluffing or intimidating as the need may be and, if all else fails, fast-talking your way out of trouble.

And she is just so bad at it!

Even Nix, when sent to distract enemies, is so adorably bad at it. And no use pretending that Kay's lame "I'm really good at this, trust me" isn't reading across languages. You can tell when a jawa is being sarcastic, in this game. There's even a fast-talk skill that takes multiple missions and challenges to fully unlock, which resolves as Kay going "Uh, hey guys, I was just..." for maybe two seconds before they draw their blasters anyhow.

Maybe she gets better at it in the course of the game but it takes a bit of the fun out of it when she, the player, and everyone else is completely aware she's not the hardened outlaw she pretends to be.

Only maybe she is. Or maybe it is just the universe is lame. The big patch Ubi put in was to remove mandatory stealth. Well, I started the game on High difficulty because I already knew the combat was too easy. It turns out just difficult enough that stealth is a good idea. Oh, and while Kay may be able to pick locks and crouch-walk, her take-downs are the least stealthy thing you can imagine. It really does become ludicrous, when you send your pet to attack a guy's face, kick his legs out from under him, then slam him into a table, four feet away from an oblivious Imperial Stormtrooper.

Who you can then knock out by hitting his helmet with your bare fist. What planet did Kay come from, anyhow?!

In open combat, at least the Stormtroopers have good aim. Uncanny good aim. If you try to fight everyone in an Imperial base it is going to be one tough fight. (If you can find a hiding spot, though...it may take a while, but eventually you get that old "It must have been the wind.")

So it isn't completely immersion-breaking. There's enough difficulty and tension to make you feel like you accomplished something. Something random, that is. These are for all intents and purposes Radiant quests, even if only the Faction quests really do seem to be radiant. (The faction quests are also notable in that most of them, and all the high-paying ones, are "betray this other faction." Which is why your faction approvals go up and down like the VU on an '80s stereo.)

There's even a mechanism by which right as you complete a mission, you can choose instead to betray the faction that sent you on it!

So it all sort of works, and aside from just wandering the markets, there are some fun missions; infiltrating Jabba's Palace was about as good as the game got for me, although the fight with the rancor at the end was not fun.

Because with all this betrayal going on, it totally makes sense the game would betray you. Remember that it used to have mandatory stealth missions? Well, many of the missions it doesn't matter if you have perfect stealth, when you grab the McGuffin it will go to cut scene and at the end of it the entire place is alerted and shooting at you. Plus your reputation with that faction is shot (I don't think I'm ever getting back in good graces with the Pike cartel).

And this is Ubisoft. King of ridiculous control schemes. If you've stolen a bracelet and want to pawn it, hold down the space bar. It slowly scrolls until the sale is completed. Have two? You tap the spacebar to go to a screen where you select how many. Then tap the icon of the space bar on that screen? No; that icon is broken. Select the icon with the mouse and click on it to complete the sale.

(Or maybe not. The controls in this game are peculiarly opinionated for how long you need to hold them down. Take fast travel. Tap too quick, or too short, and it won't happen. This ain't a fast travel mechanism, it's a rhythm mini-game!)

Want to go lower on the grappling line you are hanging from? S. Unless this is a place where you can swing, so you have to use E instead. To get off, press C. Or maybe space. Or maybe space makes you jump to your doom. Sometimes it is space, sometimes enter, sometimes control is back, sometimes it isn't. Your gun auto-switches ammo type so the civvie you tried to stun just got a blaster bolt to the face.

Bringing us back to the rancor fight. Ubi really loves doing these action sequences where they introduce a brand-new mechanic they've never used before, with a brand-new control for which the icon might or might not flash briefly. For the grappling hook, it is a paragraph. A list-out of six different commands. In small type. For about a quarter of a second. At the moment you start your swing across a chasm.

And that was the food stall experience. A delightful little sequence with nice animations and a cute cutscene and delicious looking street food and Nix of course being adorable...and you spend the whole thing squinting at your screen as ten different buttons are called for in different orders and fast succession.

Turns out you can at least turn that shit down. This is a modern enough game to have a tweak screen for "difficulty," even if the Ubilords phrased it all in terms of "If you are too much of an idiot to play our perfectly-designed game the way we designed it, and have to ruin everything, and if you are really sure about this, then you can turn this one little bit down just a hair." Just for the eating mini-game. Because even Ubi realized how shit that was.

In so many small ways they are trying to make this the same rote looter-shooter they turned Assassin's Creed into (in fact, so much of this is just retread). Even Tatooine is cluttered with icons for meaningless bits of loot and XP.

They had six hundred people working on this game. It makes sense that they could have some wonderful costume set and even hair design that really captures the old-school Star Wars vibe (if the entire thing took place in a back alley behind the cantina -- seriously, where is the good part of town on any of these planets?) Creature animation, environments, a few basic but amusing puzzles. And then lame and annoying story, gameplay, and control scheme decisions that cramp your experience.

AAA is balanced right now at the point where they have to put enough random crap and collectibles in, and enough mandatory sequences that slow the game down, so people feel like they got their money's worth. Even if the return for a longer game is you get a shittier game. The raw economics, or the way the companies interpret the economics, are spiraling into a Corey Doctorow of games that are increasingly expensive yet increasingly less fun to play.

Fortunately, Steam has frequent sales. I would never have paid the full dollar for this one. At slightly less than the cost of breakfast out...I am not unhappy.

The Fifth System of War

I am leaning heavily on Brett Deveraux here, who extracts from John Keegan a way of referring to styles of warfare as systems. 

In my own simplifications, First System is raid-based, where the aim is exploitation/denial of resources (and the land that has them) through raids and attrition. This thinking dominated the Plains Indians even while the United States and Mexico had moved into the Civil War -- Second System warfare -- and arguably still holds in some forms of guerilla warfare in areas (like the mountains of Afghanistan) that are difficult for other than small lightly-armed groups to move in.

Second system arises with agriculture (another thing we can blame on agriculture!) and assumes fixed population centers where the exploitable resource is not just the fertile land but population that farms it, along with their infrastructure.

This is the Sid Meir's Civilization model, where control of cities is the aim. Armies become large if often seasonal because siege happens. And with armies and the static nature of siege, armies themselves clash -- again I refer back to Afghanistan, say the Retreat from Kabul, where an army designed to fight armies meets a force designed for raid-and-retreat.

What Brett variously refers to as Third System or Modern System is when technological advancement, particularly artillery and air power but well-heralded by the machine gun, makes forming up in the field a bad approach. This is almost a situational definition; the First Gulf War saw a fully equipped Third System army fall before a Third System army with a few more decades of technological advancement behind them, as if they were an artillery battery at Woking against a Martian Tripod.

He suggested, very tentatively, (and well prior to the invasion of Ukraine), a Fourth System, where drones and cyber-warfare come to the front. "If you can see it..." is replaced with "If you can hack it..."

Horizon Zero Dawn has a heartbreaking meeting of systems in this regard. Semi-autonomous robots and cyber-warfare were a mature technology in 2064. Self-replication, and hacking of opponents to suborn them, were big selling points of Ted Faro's "Chariot" line. In re the latter, Sobeck herself referred to the Hartz-Timor swarm as being an apex predator.

Which meant all of humanity ended up on the back foot of trying to fight conventionally, without any of the digital assistance they'd become used to, in the costliest delaying action of all history. Actually and implicitly handing out rifles to anyone who could lift one and throwing them into the meat grinder just to buy the Zero Dawn project the time it needed.

Which, as a fictional situation that bends a few things to make itself happen, leads to my facetious designation of a Fifth System. This is the implicit (and almost never explicit) premise behind, say, Halo, Star Wars, or, well, any number of games and movies.

You could call this the Sergeant Zim system of war; "if you can get a man with a knife into the same room with it..."

And I've made that implicit and recognized in-world for two of the novel concepts I've been playing with. In the vampire-werewolf-love-triangle (in space!) I was tinkering with, it is recognized as being the only thing that super-soldiers actually might be useful for. Even if the people who came up with the cyber-boosted genetically manipulated wolf-humans had no idea (they bought into the stupid idea that a physically strong but very expensive soldier would be a really good thing to have).

In the Blue universe, this is implicit in the currently low-intensity, rapidly involving situation, as well as the kind of technologies in play and especially the lack of mature military doctrines in the use of any of this.

But basically, this isn't a realistic strategy. Sure, you can recognize that just as with any hacking, the human layer is most vulnerable, followed by the physical layer. And you can suggest that a small infiltration force, which is highly trained and equipped in expensive ways, could get in to subvert or destroy the systems. This is in fact done all the time.

Enough so that any military force that isn't completely new to the game is going to have countermining against those sappers, close-in guns for those "small, agile snub fighters" and so on. Dropping in some SEALs is just smart strategy. Pinning your entire war plan on them personally bringing down the enemy's stronghold is not.

In a fictional setting, you could both handwave and lampoon by expressly considering this as a clash of systems. In far too much fiction, insurgency tactics (because the heroes are usually the underdogs) are remarkably successful. It is rare when it is recognized that the weight is sometimes more moral than strategically effective.

That, and in fiction this is almost never the plan anyone wanted. At best, it is a Hail Mary. More usually, the heroes just happen to be in the right place at the right time, and while the world is dependent on them pulling off this impossible stunt, it was never in anyone's war plans.

In any case, many of the depictions are a technological imbalance between more-or-less similar systems. James Bond simply has better gadgets than the guys trying to stop him...and it is a close enough battle that their Bonds often succeed in doing a lot of damage to "our" side.

Expanding from that idea of changes of system with changes of society and technology, this Fifth System arises when you are in a kind of cyber-trench warfare. And in fact, perhaps the biggest explicit depiction of Fifth System warfare in fiction was cold war spy exploits; No Man's Land here being the MAD doctrine and the fact that open warfare itself was impossible.

Any halfway-realistic setting that has casual interstellar travel is up in power levels that make actual warfare incredibly and mutually destructive. If there are counter-measures or defenses, they are themselves also technological, and thus both front lines (as it were) are complex enough that it becomes a theoretical impossibility to plug all the holes.

So warfare, whatever the actual form of it is, involves exploiting the inevitable vulnerabilities that show up as both sides continue to evolve and react to continue their technological detente. To lose the arms race is to be destroyed, but to continue the arms race opens up those vulnerabilities to special forces. And this atmosphere of constant technological change means the opportunities are constantly changing.

The first side to invent superheroes wins. For a few months -- then everyone has them and status quo returns.

Research not done

Not going to watch all five seasons of Breaking Bad. Not going to finish The Tewa World View, or the book on nuclear secrecy, not even going to buy Yellow Dirt. Not making another research trip to actually go into Holloman AFB (much less the Atlas-F site outside of Roswell). There's a dozen books I only started (some of them I only read the sample chapters). Not going to finish them.

The world is big and complicated. I'm trying to kick out a novel once a year, so even though it may look like I'm doing a lot of research, I'm really only doing a fraction of what I probably should.

For the New Mexico book, most of the book is behind me now. I still have some stuff to look at for WIPP, the Atlas-F site, and even Egtved Girl. But everything else, the scenes are written and there's no point continuing (or even starting, for some things.)

It is still doing better than the Paris novel. But then, the Paris novel was all about Paris seen through the eyes of others; through film and literature, by visitors, etc. There were several set-piece scenes that explored and unpacked some work connected to Paris, within a Parisian setting. Plus had a little action in it.

The climax takes place on the roof of Notre-Dame des Paris, and references the Hugo novel; how it interpreted this building with its own real history, and how Disney and others in turn interpreted Hugo's novel. Through the lens through the lens.

I didn't finish reading the book and I've never seen the Disney animation.

The sequence at Opera Garnier got closer. Like Notre-Dame, I've been inside myself. And I at least know the Andrew Lloyd Weber score via cast recording. I've never watched Phantom all the way through and I didn't finish the book, either. And as with the Hugo novel, the scene takes note of the many interpretations and expansions throughout the years, from the novel Erik to Phantom of the Paradise.

There's a sneak scene during a Van Gogh show. At least I've seen one of those shows, and I know a little of his art. But, no, I didn't read a biography or really study his art for the scene.

In fact, the last time I can think of that I actually finished ONE book (out of several resources that were available) was the Japan book. I read a book on the Takarazuka Dance Troupe. But I've never seen even a clip of one of their shows, much less all of "Beru-bara" (The Rose of Versailles).

Here's how bad it is, in a nutshell. I bought a coffee table book at the visitor center of the White Sands National Monument. I have barely opened it. It isn't even on my desk as I finish this novel.

The ghost in the machine

There's something weirdly detached about AI art. This is beyond the various "tells" that range from obvious to edge cases. It is this sense of there being no there there. A soulless-ness that is of a different quality than corporate art or mass-produced art or been-through-too-many-committee-meetings art. That's the backdrop that makes those glimpses of a real soul in there so striking.

The first level is seeing those human signs that are divorced from their original context. In Saberhagen's Berzerker series, the titular machines do not have vocoders or some other synthesized speech. Instead they talk using scraps recorded from the various humans who have been its prisoners. From word to word the age, accent, gender, and emotional content of the voice changes. And you get glimpses of those original personalities, the fearful, the resigned, the aloof intellectual, the child in terror.

It freaks me out as someone who has tried to learn visual arts, and who has studied the style and careers of various artists. Much of what the AI was trained on is effectively anonymous, as this was the output of commercial artists on contract or the various cogs in a studio system, or just artists you haven't yourself learned to recognize. 

But you still see those things, those brush strokes, those choices in line, the way a mouth is drawn. Things that in context are part of a complete style and approach. That come out of a philosophy (quite possibly one that is ever-shifting, as an artist approaches different projects with different intents, and their career changes over time as well).

But you don't see the same thing across the image. One line is drawn one way. Another line, right by it, is drawn in some different way. Not completely different (the AI is more selective than that) and not usually pure (it always blends a little). So you probably won't see an Arthur Adams mouth or a Rob Liefeld foot, not perfectly preserved in amber. But you will have a blurry glimpse, a funhouse mirror version, enough to know without the slightest doubt that this particular artist was in the training material.

And this lack of a clear overall purpose is also visible in something else. Something that is most marked when some semblance of it is present.

And that is in composition. On the largest scale, on the scale of the total image, the AI will probably have a harmonious composition. Most of the material it was trained on had that, and enough of it was using similar rules (or, more precisely, a small set of different compositional approaches that themselves have a small set of fairly well-defined rules). 

And some of this is because the current most popular approach of generative AI for visual art is progressive approximation, working from the broadest outlines in. So you could say the AI starts with blocking, with a massing study, and that is probably decent. Where it falls down is because there isn't a guiding purpose that makes sure that everything else unfolds according to it.

That's why AI is currently so shit at lighting, by the way. And yes, the AI-boosters will point out that it is "getting better" and will continue to get better. But it is doing so by brute force and band-aids. When the Hildebrandt's painted, they had a clear vision of sunlight (usually) coming from one direction, the reflection of what was often green earth coming from another. This was in their heads with every shadow, every shading of every limb or rock or tree.

And it was in the planning. When an artist like Jack Kirby penciled, he knew where there would be blocks of solid black, and he planned and placed them as part of the composition. They didn't just fall where you'd expect a shadow to fall.

The AI can capture this statistically. It places the shadows where they usually fall. It shades because the training examples shade. It only gets it right most of the time because most artists are doing the thing in similar ways and with enough samples it will probably not be led into producing an outlier.

But that's all aside of where I was going with this. There are AI images which go beyond simplistic harmony in their composition and actually move on to telling a story within it. Where there is a focus and eye-leading, where the pose isn't just a collection of average limb positions but actually communicates intent. When a character is engaged in a purposeful action that can be read.

And the reason is because there was an original image. I recently watched a video about a young artist (who has learned better since) who did quite nice drawings then sent them into AI to make them look more polished. His style was perfect for this, by the way. AI gets confused by linework and isn't going to save it anyhow, but blocks of color tells it the ideas you want it to flesh out.

There are other routes to this soul transplant. One is by over-prompting the AI to copy a single original artwork you've been inspired by. You might ask it for an astronaut on the moon and with any luck at all you will get one of the three top images from Apollo 11 that appear over and over across media. With a few screwy details, because it is still mashing together multiple sources and not all of them came from the photograph in question or even from the real space program. Or you might outright name the artwork and/or artist, and surprise surprise, you can absolutely get the Mona Lisa back. 

But there is also image to image. And that is, in the essentials, using AI as a Photoshop filter on an existing artwork. Or photograph, or whatever.

In fact, there are entire models that are absolutely and unapologetically based on this.

In ComfyUI, one workflow -- right there on the front panel of the application -- is to take an image of a person (which could be the actual original photograph of a real person) and a screen-grab from a movie or a TikTok, and put the choreography of the one on the other. This isn't an edge case or an abuse of the model, this is how it is intended to be used. ComfyUI supports plenty of models which are text-based and even in the above the final results can be shaped by a more typically generative process, but this is absolutely making a paste-up of two real (stolen) things. It is necessarily so; the toolchain doesn't work if you don't provide these two pre-existing things for it to mash together.

What is sad here, though, is the glimpses of a potential partner. There are parts of making art where the amount of expression you put in is too low a ratio for the effort put in. Inking is wonderful stuff. Erasing the pencils, not so much. And few people want to grind their own pigments.

Somewhere in there is the ability to partner, to have a dialog, to use technology to draw the inbetweens or fill in the blocks. Some of it, we have. Some of that has been with us for a while. There's a reason why Photoshop survives the shitty business practices of Adobe. 

Just, this thing we currently have that we call AI is being applied to the wrong parts of the process, and too often for the wrong reasons. We could have a tool that would allow us to work as artists, and automate the parts of the process that are least human. Instead we have a tool we are using to bypass the parts of the process that are the most expressive of artistic intent; to do those things that should be human.

(And, in the end, fuck up the small details we might have turned to it to do.)

Also... have you looked at the price of RAM lately?