Messages

Yes, happy birthday, Tabata!

They do, and they're not good. There was a thread on here hyping them up, and I showed how terrible my results were.
It turns out that they're good if you want useless abstract art. Which are pretty to look at, but useless for game making. But if you want something like what I just said "a photograph and then have it be redrawn in the style of The Curse of Monkey Island" I don't think that'll happen anytime soon. (Try it yourself if you don't believe me, you'll just get an ugly collage.)

I don't think that's a very accurate summary of what happened. (Edit: Actually, it was this thread.) You had a very specific thing you wanted it to do, and it turns out it couldn't do that particular thing.

It works much better with traditional realistic/impressionistic painting (as well as abstract stylization), probably because that is what it has been trained to do:





All these deep neural network tools rely heavily on their training data. So if you trained it on a lot of cartoony art, it could probably do a half-decent job at turning photos in COMI-style paintings, too.

But I think that's really the crux: It's fairly easy to get it to do something that looks pretty good, if you're just messing around and aren't aiming for anything in particular. But in my experience, having played with a number of these tools (Neural Network filters are actually implemented in Photoshop now, including artist style transfer), if you do have something specific in mind, it's very hard to get it to do what you want, and it almost never looks right.*

So I mostly agree that it's not currently a serious threat to most worthwhile artists/illustrators. That could change, though.

* With one exception: ArtBreeder. That offers enough knobs and dials to effectively let you tweak the results towards what you're looking for. Within the scope of what it does, e.g. character portraits, I think it's already a viable alternative to hiring an artist (or at least to do the bulk of the work before a final polish by a human).

Hope you had a good one, Kastchey!

And happy birthday AGA!

I mean, already a couple of years ago we were playing AI Dungeon, which is basically a collaborative AI story-writing tool. You can read some examples of stories written by OpenAI's more advanced GPT-3 system here.

You can see that the current generation of AIs can produce text that reads just fine on a sentence-by-sentence, paragraph-by-paragraph level, and usually remains meaningful and relevant to the given topic. However, it definitely still struggles to maintain coherence and focus over longer stretches. Its stories tend to be rambling and pointless. But it IMO the performance of text-producing AIs is comparable to that of image-generating AIs.

There's also this. (The text in green was written by GPT-3, while the text in white was supplied by the person interacting with it, @funnycats22 on Twitter):

Thanks. I'm hiking in the Norwegian mountains. I had marginal/no connectivity and charging opportunities for the first days, but now I'm back on my phone.

Babar: no idea, but I'll say Superbad. Another hint?

I just read The Good Soldier by Ford Madox Ford (it's great, and the title is very misleading), and found myself humming this song; obviously because the narrator keeps calling it "the saddest story" he ever heard (which was also Ford's preferred title):

https://youtu.be/Z7wSBqOIJBo

[To Catch A Thief]

Fairly sure that's To Catch A Thief.

That's correct.

OK, here's another one. I'm afraid you'll be on your own in a few hours, so if anyone is confident they've got it, please just move ahead:

I'll be going away on holiday tomorrow, so I'll make it easy (starting off with two caps!) and hope someone gets it before then:

I think both statements are true. They wildly overpaid at the auction, with the book selling for almost 100 times its estimated value and what other copies had previously sold forâ€"though as I said, the book is rare, so it's not like you could just at any time scrape together $40k and just go out and buy one. The opportunity only comes around rarely. (At the time, the DAO hadn't raised enough to pay for it, so Soban Saqib, a crypto-millionnaire, provided the funds. Technically speaking he was the one who actually bought the bookâ€"as I understand it he later sold it or otherwise transferred ownership to the DAO.)

Then over the following days the DAO raised more money off the publicity around the sale. Saqib claimed they raised $12M. Earlier this year they apparently held about $3M in Etherium, before the currency crashed and the value sank to about a third: that sum may be the money left over after the purchase and their various other investments.

Has anyone, other than the auction house and the extremely lucky seller, made bank from this affair? I don't know, but with so much money floating around and the vague and unrealistic mandates for how to spend it, as well as the general crypto-currency mindset, it certainly wouldn't surprise me.

I'm no fan of crypto-bros, and there is plenty to criticize about "The Spice DAO" and their purchase of a copy of the pitch book for Jodorowsky's Dune movie, but there is also a ton of misinformation going around about the scheme. (The best reporting on the story has been from The Verge, with this as the latest update.)

I guess the problem with being a decentralized collective is that it makes it harder to establish who speaks for the project. Consequently, a lot of what's been reported about the Spice DAO comes from posts made by random people who might only own a 5$ stake in the project. For example, the story that they planned to burn the copy of the book, make a video of it and sell it as an NFT was never any plan or official statement from the DAO. It was something one guy suggested on their forums, and which immediately got shot down by everybody else. (Imagine if everything anybody says on the AGS Forums was taken to be an official statement by AGS!) In fact, the DAO has made arrangements for proper archival preservation of the book.

Many news outlets have also reported that the book they bought is already available online. That is not true. Scans of a few pages were available, plus a number of lower-quality screencaps from videos of someone flicking though the book (taken from the movie Jodorowsky's Dune), and mixed up with pages from other sources, reconstructions and outright fakes. At an estimate, about three quarters of the 200-page book had never been publicly seen when they bought the copy. That it remains largely unseen is a big part of the book's mystique. The Spice DAO recently made a video showing someone flip through the entire book, but you have to own a stake in the project to see it; they are reportedly still working on scanning it in high quality and somehow making it viewable by the members of the DAO. (This would almost certainly be a copyright violation, but since they in some sense co-own the copy they might just possibly be able to get away with it.) I certainly hope they do, and that the scans subsequently leak so that the book will finally be available.

Also, while no doubt some of the people who chipped in money to the Spice DAO scheme were ignorant of or confused about copyright (after all, this is the NFT crowd), the major investors, Soban Saqib and a few others, were perfectly aware at the time of purchase that buying the copy did not grant them copyright. Though what they thought they were actually doing and would be able to do once they owned the book, and why they would overpay so much for the copy, has not really been adequately explained. It might just be that they really wanted a copy of the book, and given that only a handful of copies exist and there was no telling when another would go on sale, they were willing to pay whatever price. Or it may have been a scam; while they had to pay the bulk of the $3M auction price themselves, they reportedly raised as much as $12M from the community after the fact. (I tend to think that in crypto-world, the line between "scam" and "over-exuberant fantasy" is a blurry one.)

In any case, the plans to try to leverage the purchase into negotiating with the rights holders were always highly unrealistic. The rights situation for the materials for Jodorowsky's Dune is extremely tangled up. In particular, the widow of Jean Giraud (the main artist of the book), Isabelle Champeval, is hostile to Jodorowsky and to anything to do with the aborted movie (apparently Jodorowsky and Giraud had a falling-out in later years), and has been blocking any wider publication of the pitch book for years. There was no reason to believe she would relent.

Since it therefore looks like there's little hope for an official reprint of the book, it would be great if high-quality scans became available, copyright be damned.

That really looks like Dustin Hoffman in bed with the sheep, but It's gotta be Everything You Always Wanted to Know About Sex But Were Afraid to Ask, right? (And apparently Kyriakos is correct that Gene Wilder is in that movie, so that fits.)

Attack on Titan? That's the only anime I know of that features a big wall.

[Why Didn't They Ask Evans]

I'm not disputing that Agatha Christie is a highly skilled constructor of elegant and satisfying puzzles, and I think the misdirect functions well as a twist in these stories (and other times she employed the same trick, particularly in The A.B.C. Murders; The Body in the Library also has a good variation on the gimmick). But are they "believable" in the sense of "these are plans a murderer would in reality make and carry out in order to not get caught" and "these plans would in fact work" (that is, reduce their risk of getting caught)? Hardly.

And again, realistically speaking, criminals do make mistakes, and difficult cases are often solved by fortuitous coincidences. Insisting that the successful solution of the case should not rely on mistakes by the criminal or on coincidence is not to insist on realism or objective believability, but on what is elegant and narratively satisfyingâ€"i.e. artificial.

To get back on topic, I haven't seen many detective mysteries I liked on TV or film recently that aren't on arj0n's list. (Hugh Laurie's adaptation of Why Didn't They Ask Evans was a disappointment after a promising start, with the third episode so rushed as to be almost incoherent.) But let's see…

Series:
Sharp Objects
Big Little Lies
Trial & Error (comedy)
American Vandal (true-crime spoof)

Films:
The Dry (haven't actually seen it, but it got decent reviews and I liked the book a lot)
A Simple Favor
Knives Out (the solution is far-fetched, but still a good film)
A Hero (more of a drama, but has some mystery elements)
Burning

I'll also throw in a book series: The A Good Girl's Guide to Murder trilogy (plus wholly unnecessary prequel novella) by Holly Jackson. Like Veronica Mars, it's a very satisfying mix of YA tropes and proper hardboiled detective mystery (the books ultimately get very dark, as events take their toll on the main character and the people around her), and like Only Murders in the Building it uses the idea of a true-crime podcast host as amateur detective. The books are also really well produced, full of reproductions of the various evidence (photos, interview transcripts, web records, Google Maps printouts, etc.), very much like an updated version of the old-school whodunnits that featured a map of the scene of the crime, and the audio books similarly take advantage of the podcast conceit, with the interview transcripts performed by a full cast. I had some quibbles with each book, both in plotting and prose, but overall they are very compelling. Only, if you do check it out, make sure which version you get: the original version is set in the UK, but the American version of the books relocates it to New England, along with some other changes.

I added in a more explicit discussion of why I don't find one of the plots you mentioned very believable:

Spoiler

In After the Funeral, the family hasn't seen their aunt Cora lo these many years, and don't realize that she's actually an impostor. Fair enoughâ€"perhaps. But then they all meet the impostor again in her real identity within a few days, and none of them recognize her. That I find very hard to believe.

[close]

I think the difficulty many Christie adaptations have had in making believable disguises when the same character appears under multiple identities demonstrates the flaw in those plots: they wouldn't actually work in reality. That objection may be harder to demonstrate when reading a book and you have no portraits to refer to, but it's still a problem with the realism of the story.

As for the other one you mention:

Spoiler

I had huge issues with the logic of the murder in 4.50 from Paddington, but I forget exactly what it was. Perhaps that the killer would obviously fall under suspicion for the mass poisoning towards the end of the book? I do recall being frustrated because Miss Marple doesn't actually solve the case in a meaningful sense: she never demonstrates a chain of deductions or inferences from the evidence at hand that points to the murderer. Instead she goes the cheap "TV detective" way of just intuiting who the guilty person is and then setting them up to incriminate themselves.

[close]

I don't agree that (at least the ones I mentioned) Christie's plots are as artificial as one tends to see in tv shows. At least Christie doesn't have the "killer made a stupid mistake/miscalculation" all that often (it does happen, but isn't central/things spiral down by that point). Compare to something as ridiculous as the 7-season The Mentalist, where essentially half the episodes are solved by "instant hypnotism" 

I'm a huge Christie fan, but I think by any standard her plots are highly artificial. They are very ingenious puzzles, but they are not realistic crimes, or realistic investigations of crimes. (And that's not what she was going for.) And of course, in real crimes killers make stupid mistakes all the time.

To take a favorite, Five Little Pigs:

Spoiler

the crime itself is fairly realistic for Christie, in that it's a pretty straightforward murder that gets obfuscated by a cover-up based on a misunderstanding. And the investigation (with the twist that Poirot has to investigate retrospectively, fifteen years after it happened) is well-plotted and engagingâ€"but it is not realistic. The testimonies are far too accurate, far too revealing (the murderer provides a written testimony that includes key facts that allow the murder to be brought home to them, while other witnesses give away secrets they've desperately tried to conceal), and Poirot's deductions hinge on recollections of details that simply could not be reliably recovered after all that time. (Like, what flower smell was in a room on a certain occasion when you entered it fifteen years ago, even though you didn't note it at the time, or precisely what words were used in an argument you overheardâ€"and misunderstoodâ€"as you were passing by a room.)

[close]

It's artificial as hell, but very satisfying.

(Vague spoilers for various Agatha Christie mysteries.)

I sometimes wonder if Agatha Christie suffered from face blindness or something, because a disturbing number of her plots depend on people failing to recognize people they meet. (Most infamously in one novel where a woman has married the same man twice without recognizing him.) This is also a problem in adaptations, where you can always see where things are going as soon as the outrageous theater disguises show up. Though as Ali has pointed out on Twitter, in the 1957 Witness for the Prosecution they actually do a great job with Marlene Dietrich's makeup; it's the cockney accent that's the problem.

Which is to say that I don't quite agree that the plots of (at least one of) the mysteries you mention are entirely solid and believable, Kyriakos.

Spoiler

In After the Funeral, the family hasn't seen their aunt Cora lo these many years, and don't realize that she's actually an impostor. Fair enoughâ€"perhaps. But then they all meet the impostor again in her real identity within a few days, and none of them recognize her. That I find very hard to believe.

[close]

(In fact, Christie's obsession with this theme goes back to a Capgras-like recurring nightmare she used to have as a child, of a monster she called "the Gunman," who could transform and impersonate anyone she lovedâ€"like her mother or sisterâ€"and would only be recognizable by his pale blue eyes. As she writes in Crooked House: "Because this is just what a nightmare is. Walking about among people you know, looking in their facesâ€"and suddenly the faces changeâ€"and it's not someone you know any longerâ€"it's a strangerâ€"a cruel stranger.")

So...you changed your mind mid-post? 

No, I just think etymology is harder to get away from than naming practices. (Of course this gets into questions like "What is the native language of Asgard?" and "Where did that language come from?" that simply don't have any answers that make sense. We're not talking Tolkienesque worldbuilding here.)

This doesn't really bother me that much, since the MCU Asgard is clearly its own thing with its own, non-Scandinavian culture. Out-of-universe it is of course inspired (loosely!) by Norse mythology, but in-universe (to the extent that it can be rationalized at all) it seems more that ancient Scandinavians somehow learned a few things about this alien realm, and that influenced Norse culture. It doesn't mean that anything that holds true of Norse/Scandinavian culture or language applies to the people of Asgard.

More than the gender, I would question whether "Astrid" (meaning "beloved by one of the Aesir") makes sense as a name for one of the Asgardians/Aesir.

OTOH, I was annoyed in Thor Ragnarok when Hemsworth mangled "Nidavellir" (I think he said "Nivadellir") and they left it in.

I wonder if this is a case where people just aren't aware of the TV show.
If you were told that it was a still from a movie, would that help you recognize the getup?

When people repeat jokes that have been run into the ground. In particular:

"I could tell you, but then I'd have to kill you."
"Actually, it's only X if it's from the X region of France. Otherwise it's just sparkling [description of X]."

Isn't that sometimes the point of such jokes (I suppose not in the examples you gave, though)? To overdo something to the point of hilarity?

Yeah, I don't know if I can defend my annoyance in any principled way, because I don't hate all overdone jokes. "That's what she said" can be funny depending on the context, and I'm always happy when the Loremen do "In Dorset? -'Course I do, I bloody love the place."