Hacker Newsnew | past | comments | ask | show | jobs | submit | coldnose's commentslogin

E621 isn’t even mentioned in the article, but of course it comes before pornhub in the title on HN X3


As someone who considers himself a relatively informed netizen, I hadn't even heard of e621.net until now, and I don't know whether I should be proud or ashamed of that.


E621 is probably the best porn site on the internet tbh. I’ve never seen a site so well organised and searchable.


It's an incredible source of tagged training data - meticulously detailed and hand maintained


There's in fact a Stable Diffusion model trained on most of e621. Uh not sure what HN's linking rules are, but look up Furtastic.

Though I understand and respect furrydom, I'm not a furry... But in its niche of generating anthropomorphic people, I still find the model to be incredible. Its got to be one of the most extensively trained, tagged and coherent SD 1.5 finetunes out there, which doesn't even surprise me TBH.


I'm not usually impressed by generated images. But the furry models are astounding. Right now my best signal for something being AI is that it's incredibly detailed by doesn't have a signature.


Yeah.

As a seperate note, I saw another HNer blog post specifically complain about AI generated furry art as a destroyer of talented furry artists... I get the point, but still have mixed feelings.

If some AI is so incredible at an artistic niche (not just furry), isn't that amazing? If artists are not economically free to pursue new styles and representations the AI doesn't "know," I feel thats a larger problem with society and art platforms, not AI specifically.


I don't think it's as simple as just exploring new styles. AI has gone from generating unrecognizable crap to stunning digital art within a year. And it's not clear how much further it will go. If you are a digital artist who makes their entire income on what was once a very secure job of drawing furry art. You'd be terrified right now. The artists are still quite a way ahead of the AI stuff, but for how long?

I think the most likely outcome is that people stop drawing manually and either switch to making physical products like clothing and fursuits. Or they make use of the generative tools to create much more elaborate art. Perhaps one person pumping out a full comic in the time that would have taken 10 people.

It's still a very scary and uncertain time when peoples incomes are on the line and decades of hard earned skill is at risk of becoming worthless.


Yeah. And forget images or comics, what happens when AI can make a furry TV show, even if it has to be steered by an individual? Its not that far away, all the pieces are there. How do oldschool artists possibly compete with that, even if the AI quality is questionable?

I'm just frustrated society isn't ready for this. Someone who made useful stuff for others, for years, shouldn't have to worry about having a roof over their head when automation is so rampant.


I think you meant "anthropomorphic animals".


True.

Maybe the adjective is mismatched, but "people" is not necessarily a synonym for "human," right?


There's nothing quite like a well tagged booru. There are more out there.


I mean only if you care about furries, for anime Danbooru is probably an equivalent, funny enough I have never found a website like E621 and Danbooru with real photos, the tagging is always terrible.


And here I was hoping some poor intern at Fox had visited such a site and had to explain what it was to their superiors


It's because I'm a furry and it's the only 'porn site' (not all the art is even explicit) I care about. And Inkbunny too I guess.


I'm perplexed that someone would name a pornographic site with the same name as a very popular food additive.


Great, now I am too. Thanks. Forgive me ad-sense recommendations, but I'm going in.

Looks like it's unconfirmed but likely it was specifically named for monosodium glutamate. "This is where we keep the tasty content."

https://en.wikifur.com/wiki/E621

https://www.dictionary.com/e/slang/e621/

https://knowyourmeme.com/memes/sites/e621

(Edit: Apparently there's a sister site called E926 to share SFW content. E926 is yet another food additive (Chlorine Dioxide) which is used to bleach food for a cleaner appearance.)


E621 - Monosodium glutamate (MSG), Flavour enhancer.


Hmm, single life isn’t so bad


I’m wary of these kinds of stories, where someone PR-savvy with a grudge lobs nebulous complaints against a company. Can anyone at BO without a grudge substantiate this “toxicness”?


Check out the Ars Technica story. They confirmed with sources. https://arstechnica.com/science/2021/09/former-blue-origin-e...



> After publication of the essay, Ars spoke with several current and former employees who have provided reliable information in the past about the company. Although it is clear the essay was a product of disgruntled workers, these sources agreed that there were elements of truth in the essay. For these sources, the withering criticism of Blue Origin founder Jeff Bezos, and his hand-picked chief executive, Bob Smith, rang especially true.


This seems like a trivial case of Betteridge’s law of headlines.


I really wish there was a safe way to initialize structs with arbitrary binary data in rust. Perhaps I'm missing something, but it seems intuitively safe, especially if the struct only contains integer primitives.


How would it be possible to make it safe? The compiler can't verify that N random bytes input from an arbitrary source are going to be valid for the types in the struct - you have to tell tell the compiler "trust this even if you can't verify it from the source code" - hence the unsafe keyword.



Oh cool! Thanks for the pointer.


For integer primitives, all possible values are valid. Just allowing binary initialization for structs that consist of integer types would allow compatibility with C structure. As it is, I have to either resort to unsafe code or manually deserialize every struct


Oh I see, that makes sense. Also there's a sibling reply to yours for a cool looking proposal to do just that if you haven't seen it yet.


Hear hear


I'm no fan of RMS, but it does seem cruel to oust him for just one of his (innumerable) faux pas.


How?

Any one of us would be tossed out of our job for far, far less. People here act as if Richard Stallman has been tortured, humiliated, tarred and feathered and sent to the stocks, but the truth is he's been gently mollycoddled and protected from the consequences of his behavior for decades and never even suffered more than minor inconvenience.

I had a post show up on my Twitter feed where someone claimed Stallman licked a colleague of theirs on the arm. Even passing out the "pleasure cards" people here think are so quirky would get anyone here fired. Stallman apparently gets away with touching and licking women at will.


I'd feel better if the mob cited his lifetime of shittiness, rather than focusing on exactly one incident. My favorite was when he gloated over Steve Jobs' death the day after he died: http://stallman.org/archives/2011-jul-oct.html#06_October_20...


N=50


You mean n. N is population size.


What an awful headline


I feel like changing it to "Tim Cook gave a speech" but that would be passive-aggressive.


I think they're just trying to sound "cool" by using millennial colloquialisms. However, besides being less informative than a more nuanced alternative, it ends up sounding cringeworthy to me in a business/finance publication.


No, this is about Apple iPhone and iPads disallowing the Facebook app from collecting a large portion of the data that Facebook believes that it needs to collect in order to do business.

This is not an article about a speech that itself may end Facebook. This is an article about a speech given to formally announce a decision Apple has made to protect its users from Facebook's insane data collection.


Like that GAN that simulates Pac-Man, this is an exciting new level of inefficiency I couldn't even conceive of a couple decades ago.


We need to go deeper! Soon we'll have a machine learning algorithm conjure Mac OS by calculating the next frame to display, taking into account user input, based on watching thousands of hours of normal Mac OS usage


That's brilliant. Gave me the funny thought that GPT-3 could solve one problem we used to argue about back in the day. What's the right order for loading extensions? I don't remember the details, but I do remember my old office would discuss it endlessly. The black magic and superstitions to keep the system going as long as possible before the inevitable crash. Maybe, 25+ years later, there could be a definitive answer.


A neural network is not so much a "definitive answer" as black magic and superstition - but digitized.


Surely there’s a statistical approach to answering that question, isn’t there?


I seem to remember a Cassidy & Greene program that would do all sorts of statistics to your extensions. I mainly used it to do a binary search to figure out which extension is causing the shareware app of the day to crash.


Conflict Catcher! By Jeff Robbin, who later wrote SoundJam MP, which was then bought by Apple and turned into iTunes, and he stayed at Apple for many years (may still be there even(


They also had one of the best classic Mac arcade games of all time.

Crystal Quest and it's sequel, Crystal Crazy

https://www.macintoshrepository.org/5955-crystal-crazy


This was one of the only games I actually played in the Macs they had at elementary school (we spent most of our recess time on Napster...playing 12 different songs simultaneously in the same room, and nobody cared). I spent so long looking for this game, but I couldn't even figure it what to search for, because I didn't remember the title.


Prediction: Spinning beach ball (92% confidence)


On 8.1? More likely “System Error Type 11”.


Type 11 was really just an issue on PowerMacs, and this emulates a 68K Mac. (Type 11 was basically a catchall error for any native PPC exception)

</pointless historical nitpick>


Also I believe 7.6 updated error reporting to native PPC codes so that a lot of the time an application could just crash on its own without crashing the whole system. Not always, but yeah even without protected memory.


There was one release of System 7 where they advertised "Type 11 Errors are virtually eliminated!"

Imagine that being a selling point in an Apple product today...


Huh, TIL.

(I only ever used one real 68k Mac, and that was something donated to my school’s art department).


Guru meditation then? Oh, right CPU but wrong computer...


Weirdly I don't think there's a machine learning algorithm that can actually do this. pix2pix loses cohesiveness quickly, and GANs require mostly aligned input to make any sense of reality. BigGAN is the closest, yet even that doesn't seem to work very well for continuity.


yeah, you’d have to jump into the IL/RL/IRL/infogan tarpit :)


You must like running Doom on Windows 95 inside Minecraft: https://news.ycombinator.com/item?id=23960028


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: