|
dwarf74 posted:The tweet linked above seems to have been deleted, regardless, unless I have forgotten how to click. This much is true.
|
# ? Dec 19, 2023 14:46 |
|
|
# ? Jun 9, 2024 16:52 |
|
https://twitter.com/tegiminis/status/1737045838214648161
|
# ? Dec 19, 2023 15:14 |
|
https://twitter.com/dril/status/134787490526658561
|
# ? Dec 19, 2023 15:22 |
|
https://magic.wizards.com/en/news/announcements/generative-artificial-intelligence-tools-and-magic They released a statement, but yeah, that listing was literally just talking about the stupid Parallaxing poo poo they do
|
# ? Dec 19, 2023 23:02 |
|
GonSmithe posted:https://magic.wizards.com/en/news/announcements/generative-artificial-intelligence-tools-and-magic I still think the statement "we require artists, writers, and creatives contributing to the Magic TCG to refrain from using AI generative tools to create final Magic products" is compatible with hiring people to tweak a bunch of baseline generated content into a final product
|
# ? Dec 19, 2023 23:52 |
|
why would that not be using ai generated tools
|
# ? Dec 19, 2023 23:57 |
|
flatluigi posted:why would that not be using ai generated tools Because if you use the AI tool to create a bunch of drafts then modify them then you've only created drafts with AI, not the final products. They have put that word in there for a reason and it's not a good one!
|
# ? Dec 20, 2023 00:00 |
|
you use drafts to make final products
|
# ? Dec 20, 2023 00:04 |
|
I'd rather go back to old timey non offensive racial slur chat
|
# ? Dec 20, 2023 00:05 |
|
flatluigi posted:you use drafts to make final products I use drafts to increase my collection for constructed. hyukhyukhyuk
|
# ? Dec 20, 2023 00:34 |
|
flatluigi posted:you use drafts to make final products ah well thy made a two sentence post on their website with slightly ambiguous language guess they cleared everything up unrelated i'm curious what happened with layoffs and hasbro within the past two weeks
|
# ? Dec 20, 2023 01:13 |
|
hmm i wonder what happens if i google wizards of the coast and ai art i assume there are zero results
|
# ? Dec 20, 2023 01:18 |
|
curious if mark rosewater ever said anything remotely like “I believe there will come a day where AI is a tool we use for design, but there are a whole host of issues to work through before we get there.”
|
# ? Dec 20, 2023 01:19 |
|
uggy posted:ah well thy made a two sentence post on their website with slightly ambiguous language guess they cleared everything up welcome back to the internet uggy! the tldr is hasbro fired a bunch of people from every single department in the company despite wotc being like their only successful division and they chose to do it right before the holidays because they're garbage. love it when corporations are constantly seeking eternally escalating profit to fill the pockets of their execs especially when it makes no logical sense, but we're fully into the 'get as much as you can before we finish setting the planet on fire' stage of existence it'd be cool if people kept talking about that instead of tilting at windmills because ai art is apparently more important to people than dumping people with decades at wotc on the street, but people gotta keep chasing the new controversy i guess
|
# ? Dec 20, 2023 01:24 |
|
flatluigi posted:welcome back to the internet uggy! the tldr is hasbro fired a bunch of people from every single department in the company despite wotc being like their only successful division and they chose to do it right before the holidays because they're garbage. love it when corporations are constantly seeking eternally escalating profit to fill the pockets of their execs especially when it makes no logical sense, but we're fully into the 'get as much as you can before we finish setting the planet on fire' stage of existence AI also impacts labor??? It's possible to be upset about multiple things at once???? Conversations move through multiple upsetting topics sequentially??
|
# ? Dec 20, 2023 01:33 |
|
rickiep00h posted:AI also impacts labor??? yeah these are all true wow, a lot of people just coming back to the internet today
|
# ? Dec 20, 2023 01:39 |
|
straight up coming across as taking their statement at face value and telling folks they should focus on this other thing instead of ai art when that's the conversation at hand and my point was that hasbro, parent company of wotc, is not really a company that one should trust or assume acts in good faith
|
# ? Dec 20, 2023 01:43 |
|
shockingly i also hate people using ai generated poo poo and i agree with uggy that wotc and hasbro have absolutely considered it in the past. the thing that's stopping corporations from leaning into it entirely is that (until lobbyists decide to throw weight around) the legality is getting established as not in their favor and "getting sued" is one of the few things that'll make an exec pause for a moment the difference here is that i think people going off on the job listing is making poo poo up out of whole cloth and actually teed the company up entirely to make a 'we're not going to use AI' statement with zero actual effort or changed position on their part. grats to whatever's left of their pr department
|
# ? Dec 20, 2023 01:49 |
|
That’s fair I think them issuing a statement shows them feeling like they have to say something cause their actions don’t obviously reflect their now stated position
|
# ? Dec 20, 2023 01:56 |
|
The 'final' jumped out to me too, but really if they were going to lie about it they would just lie about it. Its not like posting a 'please put your pitchforks down' statement is legally binding.
|
# ? Dec 20, 2023 01:58 |
|
A lot of professional artists are using AI to rough draft stuff like background details and I guess Hasbro doesn't want to immediately throw them under the bus?
|
# ? Dec 20, 2023 02:00 |
|
Yeah, I don't just randomly accept everything they say at face value, but as soon as I saw that post about that job listing I said "that sounds like it's just about the parallaxing poo poo they do," and if it WAS about AI they would have had no reason to put the statement out. They could have just ignored it and done it anyway. I loving hate AI-generated images (not calling it art), it's theft, it would be awful if that is what WOTC is doing.
|
# ? Dec 20, 2023 02:01 |
|
TBH it sounds like WOTC actually got off way lighter than the rest of the company. The total layoffs represent 20% and it absolutely does not sound like 200 people got laid off from WOTC specifically. I mean, yes I personally would not have hit the one actually profitable branch of the company at all, but they do seem to have at least taken "these guys are our one moneymaker" into SOME account.
|
# ? Dec 20, 2023 02:04 |
|
Was there a group a thread bans that expired today or something?
|
# ? Dec 20, 2023 02:18 |
|
Nah I posted yesterday
|
# ? Dec 20, 2023 02:19 |
|
uggy posted:Nah I posted yesterday So did toshi
|
# ? Dec 20, 2023 03:00 |
|
There is a theory that current AI art learning models will eventually start converging on their own results as source material for generating new art, resulting in a feedback loop of sorts that enhances stereotypes and deformities from prior results which when you think about it would be funny within the sphere of Magic the Gathering. I for one am looking forward to watching the slow transformation into the inbred art 17th iteration of Jace.
|
# ? Dec 20, 2023 06:30 |
|
it would be cool if there was an AI art tool that allowed the user to define the data set the model was trained on instead of every single one just scraping the entire internet. I get that you need a lot of data to make these things viable but it would be more art-positive in the public consciousness if it was just framed and presented entirely differently. if you're a workaday artist taking commissions and contract jobs and cranking out pieces every day, it would be nice to have a tool you could feed with your art and only your own art for a more effective model from which to create spec drafts. if you have a backlog of stuff that's all the same style, it would probably be a lot easier to touch up whatever the model trained on your stuff spat out. I get ai tools are being put out now but I'm not sure if there's anything quite like what I'm suggesting, where the model itself (the backend from which the AI generated images are created) is customized to your art specifically.
|
# ? Dec 20, 2023 07:54 |
|
So, historic Artisan event on arena made me think about power levels. Imagine a tournament that allows decks from either Pauper, Historic/Pioneer Artisan or Standard. Which would you choose. e: Also I am actually surprised that there are no youtube videos with that gimmick. VictualSquid fucked around with this message at 13:30 on Dec 20, 2023 |
# ? Dec 20, 2023 13:28 |
|
kalel posted:it would be cool if there was an AI art tool that allowed the user to define the data set the model was trained on instead of every single one just scraping the entire internet. I get that you need a lot of data to make these things viable but it would be more art-positive in the public consciousness if it was just framed and presented entirely differently. if you're a workaday artist taking commissions and contract jobs and cranking out pieces every day, it would be nice to have a tool you could feed with your art and only your own art for a more effective model from which to create spec drafts. if you have a backlog of stuff that's all the same style, it would probably be a lot easier to touch up whatever the model trained on your stuff spat out. I get ai tools are being put out now but I'm not sure if there's anything quite like what I'm suggesting, where the model itself (the backend from which the AI generated images are created) is customized to your art specifically.
|
# ? Dec 20, 2023 14:04 |
|
HootTheOwl posted:You can, it does, and no one has enough art to fill it Yeah, there's been quite a few of the engines that say they are "trained on their own pool" that have had to admit they added in other training model's datasets. There's a court case going right now (or to be filed soon? I read the filing) for some DeviantArt artists against them, they figured out a way to get specific images to be generated showing it stole their original art in the dataset even though they were saying they didn't do that.
|
# ? Dec 20, 2023 15:06 |
|
VictualSquid posted:So, historic Artisan event on arena made me think about power levels. Probably Pauper mono blue terror, kuldotha, or UW all that glitters... What answers does historic or standard have to a 5/5 ward 2 on turn 3 with counterspell and spell pierce backup?
|
# ? Dec 20, 2023 15:42 |
|
kalel posted:it would be cool if there was an AI art tool that allowed the user to define the data set the model was trained on instead of every single one just scraping the entire internet. I get that you need a lot of data to make these things viable but it would be more art-positive in the public consciousness if it was just framed and presented entirely differently. if you're a workaday artist taking commissions and contract jobs and cranking out pieces every day, it would be nice to have a tool you could feed with your art and only your own art for a more effective model from which to create spec drafts. if you have a backlog of stuff that's all the same style, it would probably be a lot easier to touch up whatever the model trained on your stuff spat out. I get ai tools are being put out now but I'm not sure if there's anything quite like what I'm suggesting, where the model itself (the backend from which the AI generated images are created) is customized to your art specifically. So, there's two things here. You can train a network on a large dataset, to give it a foundation of what a picture is, and then give it more training which is heavily biased towards a particular dataset. It's how they do deepfakes. They take a model that has been trained to draw human faces on a large dataset, and then train it some more to bias it towards one particular persons face, of which there's not enough samples to train from scratch. So, you can make a model customized to a particular person's art, but it isn't fully trained on that art. Also, you can absolutely download any of the code for generative AI models and train them yourself on your own dataset. It was a common thing people did back when the state of the art was possible on a high-end gpu (now you need to be google or nvidia to have enough compute power to compete). You will probably just run into issues of overfitting. If you were a big company who owned a lot of art (wotc might have enough) then maybe you could do it from scratch on your own dataset.
|
# ? Dec 20, 2023 16:01 |
|
Thom Yorke raps posted:Probably Pauper mono blue terror, kuldotha, or UW all that glitters... What answers does historic or standard have to a 5/5 ward 2 on turn 3 with counterspell and spell pierce backup? I do play mono blue terror in standard, though there having the terror on turn 3 is a god draw. And the most annoying match up is mono-w soldiers featuring Thalia and esper legends featuring Thalia. Or mono-r attacking with a 6/5 swiftspear with monstrous rage on turn 2.
|
# ? Dec 20, 2023 16:14 |
|
Prof is giving/auctioning away stuff on Whatnot right now for at least 15 more minutes. Right now he's signing extra Academics he got from his distributor as an auction type thing. CommandZone will be doing something similar at 6pm CST. Basically in like 15 minutes.
|
# ? Dec 21, 2023 00:46 |
|
Was there a specific Cube thread or am I misremembering
|
# ? Dec 21, 2023 15:53 |
|
If I needed to print custom cards where do you all go? Like custom tokens for copy effects and weird split cards?
|
# ? Dec 21, 2023 15:55 |
|
HootTheOwl posted:If I needed to print custom cards where do you all go? MPC for sure
|
# ? Dec 21, 2023 15:55 |
|
Vintage cube is up on mtgo fyi
|
# ? Dec 21, 2023 16:02 |
|
|
# ? Jun 9, 2024 16:52 |
|
Johnny Truant posted:MPC for sure They look pretty nice, what's their turnaround time?
|
# ? Dec 21, 2023 16:16 |