Friday is the traditional Grab Bag day around here but I'm not convinced
I'll have time for one tomorrow and I do have a few random thoughts I could do
with clearing out of the bottom drawer of my mind, so Thursday it is. There's no
theme and I don't have any plan so we'll see where we end up.
Of course, if, as so often happens, some kind of structure emerges as I go along, I'll come back and edit this so it looks like it's what I meant to do all along. And no-one will ever know. [Evil laughter...]
Where shall we begin? Better not start with anything to do with AI or there'll be people who who'll never read any further. Probably shouldn't even have mentioned it.
How about this?
Ananta
I'd never heard of Ananta until Nimgimli posted about it a couple of days ago. (Fake News!! I posted about it last December but I only realized it when I went to add a tag and found there was one there already.) MassivelyOP followed up with a news item about it the next day, by which time I'd already been to the official website to register my interest. Or "Pre-Register" as they put it.
I mean, who wouldn't, after watching this?
That's seven minutes of action I find hard to believe is taken directly from the game. But it is, apparently. Obviously it's been edited but as far as I can tell everything you see is gameplay footage.
And so is this, which just landed on YouTube.
That's ten minutes of "Official Extended Gameplay" that absolutely looks like a movie. Specifically, the center-piece car chase from a pretty good movie. A movie I would certainly watch.
My question isn't "Is this game going to be any good?" or "Will it be Pay-to-Win?" or any of those obvious things gamers always ask. It's not even "Will it run on my aging PC?", although that's a valid concern.
No, my question is "How the hell is that even possible?" Apart from the oddly stilted walking animations, the video looks like a bunch of clips from a movie. MOP described it as a "sizzle reel" and that's really what it is.
After watching it, I found myself thinking back to something that happened more than thirty years ago, which I've never forgotten. I happened to go into large store in a nearby city and I was walking through the electronics department when I saw, on a large screen (For those days.), a video game playing. It was a castle in a fantasy forest and it looked like just a movie.
I never found out what the game was or what it was running on and I can't even remember now how I knew it was a game not a film. I'm pretty sure if I could see it now it would look like something you'd see in a PS1game. It probably was a PS1 game. It would have been around that time. Whatever it was, it looked like the future. Well, it looked like the past, what with it being a fantasy castle and all, but the past seen through the lens of the future.
Technologically, we are all living in the future now and we take it for granted. Things we pay no attention to or even get anoyed by today would have seemed like literal, actual magic just thirty years ago. If I find it hard to believe something like Ananta can even be real today, imagine how I'd have felt if I could have watched that trailer in 1995!
Something I hear often these days is that improved graphics can only take the games industry so far and that the limit may already have been reached. And that's probably true, as far as it goes. But what's hard to believe about this trailer isn't so much the visual quality, impressive though it is; it's the world-building.
That's what draws me in to all of these new games, all these new worlds, from Wuthering Waves to Neverness To Everness to Ananta: their cinematic, novelistic storytelling. I see clips and I want to know who the characters are, what their lives are like, what they're doing. It's not new but it seems like it's new. It's the new future of our long, storytelling past.
Also, someone's bound to be wearing bunny ears in one of these things and that's always a bonus. I'd say you don't see that in real life but I saw two people wearing bunny ears in the street only this week...
On MMORPGs And Why They're Kinda Done
As big, mass-market affairs, that is. Not as small to medium sized niche services servicing small to medium size niches. They're still good there.
It was something Redbeard said in response to Wilhelm's post about Ship of Heroes yesterday that started me thinking. For some reason, the conversation took a detour into Guild Wars 2 and how a big part of the revenue system there relies on persuading players to part with cash to remove the inconveniences the devs have added in, seemingly for that exact purpose.
It's a well-known mechanic in many MMOs. The very unpopular version is to put up obvious pay-walls for in-game services and systems most players believe should be part of the core gameplay. Star Wars: the Old Republic used to be the leader in that awkward field, charging money for everything up to and including enough hot bars to fit your skills.
They toned that down over time and recently they've cut it right back to something closer to what players expected in the first place. It still leaves a whole lot of less controversial money-making opportunities in place, though, one of which is always storage.
GW2 has always had a serious storage problem. The game isn't overly generous with inventory space to begin with but it does allow you to expand it in various ways, all of which are technically free. As Redbeard points out, though, its much, much easier to get out your credit card and pay.
I never did and that's because I played GW2 like an old-school MMORPG. I played for a decade, virtually every day, often for several hours at a time. Just playing normally and not spending any real money, I ended up with a huge amount of inventory space across multiple accounts and characters. On my most-played characters I had all the bag and bank slots you were allowed. I hit the point where even real money wouldn't have gotten me any more and I hadn't paid a dime.
It still wasn't enough. I made bigger bags and that wasn't enough. Nothing was ever enough. GW2 doesn't just have a storage problem, it has a loot problem. There's far, far, far too much of it.
Which is also the key to not spending any money on storage. You have the choice, as Wilhelm suggests, of just throwing everything away. That'll fix it for free and right away. Or you can sell all of the good stuff you find and use the gold to buy Gems, the cash shop currency, then use those Gems to buy more storage space, which is what I did.
People complain all the time about not being able to make enough gold in GW2. I made thousands, easily. I still have most of it because I made far more than I ever needed. I didn't make any special effort, either. I had just two tricks for making gold:
- Never, ever make any Legendaries. Sell all the mats for them you find to to those who do
- Play every hour god sends
The first isn't a problem. It's just sensible. No-one needs Legendary items in GW2. They do nothing for gameplay. They're convenience items and cosmetics. Let others waste their time and pay you for the privelige.
The second is the reason MMORPGs are and always will be a niche interest. They absolutely do not respect your time.
Redbeard has inventory problems he can only, rationally, solve by paying real money because he doesn't play GW2 several hours a day, every day, year in, year out. If he did, the problem would solve itself, at least until there were no more slots left to buy with all that gold.
All MMORPGs are designed to be played for several hours a day, every day, forever. That's the business model. The reason the genre is being lapped by open-world RPGs and survival games is that those offer the same level of intensity and immersion that made MMORPGs successful in the first place but they don't go on forever and ever and ever like an eternal treadmill. They don't necessarily come with a Game Over title card but they have clear end-points that feel like a satisfying place to stop. And they have clear on-ramps every time they add DLC or major updates, meaning you can dip in and out without worrying about falling far behind.
Or something. It's a theory. I haven't really thought it through. Feel free to elaborate and iterate on it.
AI Ruins Everything
That's made a lot of people happy, I bet. I don't mean that it does. I mean that I said it. Do I mean it, though? Depends what you mean by "AI", doesn't it?
Greg Morris, whose blog I picked up in Blaugust, has a good take on it. He, like me, isn't existentially opposed to AI. He seems some merits and uses, as do I. He's just not very impressed with what people are doing with it just now.
"Widespread usage of generative products, particularly ChatGPT, by many people I come in contact with is making my life more difficult" he says.
Yes, mine too. All of ours, I bet.
Like Greg, I am not as sanguine as I could be about my doctor's surgery moving to an AI triage system. Neither am I convinced that having AI answer the phone when I call to try and buy something is the best way of handling a commercial transaction. Unlike Greg, I'm happy to say I don't encounter AI in my work although no doubt it's only a matter of time, but I have very low confidence in any of the current LLMs being able to send emails that wouldn't have gotten me into trouble if I'd written them.
What I do think, perhaps over-optimistically, is that the AIs will either have to get a lot better or most of the companies and services that have adopted them so quickly will have to drop them again pretty quickly, probably as soon as they notice how much money they're losing and how many lawsuits they're picking up. Those No Win No Fee lawyers must just love AI.
This is all LLMs, though. They're the bad AIs. Generative AI as a whole may be morally and ethically bad but only some of it is practically bad, too. The real problem with the rest is that it's getting to be too good at what it does.
I've been a fan of graphic art since I was a child. I've spent well over half a century developing a fairly well-trained eye and a reasonably informed, historical understanding of line art and illustration, particularly comics. I have a reliably-formatted emotional and aesthetic reaction to every such image I see that allows me to parse it and contextualize it by way of decades of exposure.
And I like some of the AI images I'm seeing these days as well as or better than many, many similar pictures I've seen that were drawn by humans. Humans draw a lot of really bad line art and a lot of them don't know how bad it is.
It doesn't matter if I know an image is by AI, either. It doesn't put me off. The image is the image (cf. Intentional Fallacy, as always.) It's not like I'm being tricked into thinking a human did it. AI images of a certain kind stimulate the same level of endorphin production or synaptic snap or whatever it is that makes me look at a line drawing and get a warm, tingly feeling.
Doubly so if it's one I've caused to be made myself.
If I was making my living by drawing pictures, that would piss me off. Or maybe it wouldn't. Maybe I'd just see AI as a new kind of pencil and start using it. At the moment, there's resistance to commissioned AI art but will that attitude last? In the end, don't people mostly want the right image for the purpose they've commissioned it? Is the provenance of the picture that important? And if your primary (Or ancillary.) motivation is to give work to the artist, does that have anything at all to do with aesthetics?
To some people, yes, obviously. But to the mass market? I doubt it. And I was never going to commission anyone to draw anything anyway so it's a moot point as far as my money is concerned.
The same applies to generative AI and music, another field where I can claim a longstanding interest. Or at least it applies to music I've made with AI, which I currently prefer to listen to over almost anything else.
I have to say that I have yet to hear anything made by anyone else using AI that has the same impact. Most of it is horrible. But I suspect that says more about the musical tastes of the people who are making it than the capacity and potential of the AIs. When people who like the kind of music I like start making music with AI, I'm pretty sure I'll find plenty to enjoy.
Always assuming I even know, that is. If people don't say they used AI, who's going to know?
There are some free apps around that purport to be able to tell if a song has been made by AI or not. I tried a couple. I'd link them but I forgot to bookmark them.
One is extremely accurate but that one works by analyzing the frequencies of the sample you submit rather than trying to make any aesthetic judgments. It's undeniable that AI has sound you can hear once your ear attunes to it and there's a technical accuracy to the timing that human musicians don't always achieve. I imagine that could be coded out if the people making AI wanted to hide it.
That app, unsurprisingly, pegged everything I sent it as having been made by AI. Which it was. I should have dummied it with some human music but I was just dicking around, not running some kind of double-blind experiment.
The other app, which attempts something more along the lines of a critical analysis, looking at things like the structure and the lyrics as well as the music, was much less certain whether an AI was involved or not. It tended to think most of the songs I was sending it were mostly done by a human.
And so they were. I wrote all the words. I came up with all the melodies. I gave the AI a guide vocal, which I made sure the AI followed very closely. In my opinion, those songs were indeed made by a human. A human using AI as an instrument.
I suspect most of the AI music being made doesn't have that degree of human input behind it, though. An astonishing amount of AI music is being uploaded to streaming sites every day. Millions of songs. Almost all of them made purely by AI from a short prompt.
You can see why the platforms might want to filter it out. Except, if their customers are happily listening to it, why would they bother?
Bad pushes out good is the old saw. But what if the bad gets good and the good isn't as good as all that?
If We're All Sitting Comfortably...
Gemini has a new trick. It'll make a picture-book for you. I was curious so I fed it the basic plot, such as it is, of a little series of vignettes I wrote in the 'nineties. It's the last of the fiction I'm turning into songs so it was in my thoughts at the time.
Gemini did an okay job. I've seen considerably worse published picture books in my time as a bookseller. I can definitely imagine using it to make picture books for small children, who'd absolutely love to see their pets or friends appearing in an illustrated story-book.
I think if I was going to produce an illustrated book of something I'd written, though, I'd keep my own prose and have one of the really good image generators produce the pictures. This is a bit gimmicky in my opinion.
If you want to take a look at the whole thing, here's the link. Don't expect too much.
And here's a song I made from using the same source material. It's rather better, I hope you'll agree.
Notes on AI Used in this post
Some of this is explained in the text but to be clear, the final image is from Gemini's newish storybook feature. The full prompt for the whole thing was an ad hoc precis of a whole series of fictional pieces I published in an APA back in the 90s. At some point I will be posting the originals, complete with the infinitely superior full text and collaged illustrations, so if you have any interest, you might want to wait for that:
"Please make me a picture book for adults about a girl called Phoebe, aged about 12 or 13, who lives in a rambling old house with a walled garden. She shares the house and garden with a girl in her late teens called Cat, a young woman of indefinable age called Cathy and a woman in her early thirties called Rachel. Phoebe has never left the house and garden. She has no memories of ever being anywhere else. She thinks she had a brother but she doesn't really remember him. She reads, plays chess with Cat, and just generally gets through the day. She believes Cat and Cathy are ghosts. She believes she's a half-ghost. She believes Rachel is fully real. Cat is friendly and kind to her. Cathy is drifty and vague but unafraid of anything and protective of Phoebe. Rachel is often angry because she wants to leave to get back to her friends, about whom she often talks, but she doesn't know how to leave. Rachel, when she's in a good mood, creates firework displays and makes snow fall so she and Phoebe can build snow statues of Rachel's absent friends. One day, after this has been going on for a long time, two of Rachel's friends, Sally and Cado Babe, appear in the garden and soon after they and Rachel are gone. Phoebe is left on her own with Cat and Cathy. She wonders now if she, too, will ever leave. "
The header image and the second one were both done with Google's Imagen 4.0 at NightCafe. For some reason, instead of taking a couple of screenshots from the Ananta videos, I thought it would be amusing to have an AI make them instead. The prompt for the top image was "An anime girl dressed in blue with bunny ears riding in an electric blue sports car along a freeway between the towering skyscrapers of a near-future city. A dark-haired young man weraing sunglasses is beside her in the passenger seat. Bright sunshine, blue sky. Anime line art, bright flat colors." For the second, I used the same prompt but added "A dark-haired young man wearing sunglasses is beside her in the passenger seat." All settings were left on whatever the defaults are.
The third image is one of the daily pictures I'm still producing to keep my NightCafe streak going. I'm still using the same prompt I posted about, "Walking through corn fields Covered in dust Lost in this dustbowl young female figure, old, worn clothing, line art, color, retro-futurism". This image was made with Qwen Image SD.
Posts
No comments:
Post a Comment