With all the news and controversies surrounding AI, I wouldn't be surprised to see quite a bit of a negative response to this post. With that out of the way, Do you think it is possible that there are plans for a dedicated D&D Art AI where you can choose your settings? Class, Species, and the like, along with some text descriptions, and possibly a reference photo or two, for making character profile images to drop into your character sheets, and possibly use in other places. If there are, I suspect that it will be a few years before we see a beta of that announced at some point in the future.
It's "the reference photo or two (presumably user provided)" where things get very problematic, there are many digital tools where a digital avatar can be sculpted from video games to HeroForge. WotC seems willing to draw the line at "and I want it to emulate the style of this particular artist" and if you are talking about a true reference photo, another line at "potential dumbfakes of IRL people whom the user may not have the right to."
I mean, it's possible WotC could develop a controlled and closed referenced base (with all references knowingly produced and submitted to the AI's resources), but the AI industry to date hasn't proven itself able to be so for the most part (and really why go through all the teaching of the AI to basically mimic what existing customization tools seen in video games and sites like Heroforge already do?) and I don't see WotC becoming an "AI company" to do it on its own.
Rollback Post to RevisionRollBack
Jander Sunstar is the thinking person's Drizzt, fight me.
Generative AI, I would think, is expensive computationally, and filled with 'hallucinations' though it seems to have gotten better at preventing that. Not something I would release to a large audience to use adhoc without some sort of rails.
The problem I see and would rather not be party to:
We seem to not care about how bad our sets of training data/images are. With no real automated tool to make sure AI is not trained on bad data, I don't see what value it holds. Though for genAI images, its a little bit easier to avoid since art is subjective.
Without some sort of defined limits to what the genAI can produce, you're just asking for inappropriate content. WoTC isn't going to want to be on the hook for inappropriate content showing up in their systems which are available to minors. That's just a liability. Now a lot of companies have been dumb enough to just say 'Do AI, go vroom', so who knows where WoTC's heads are at.
Again computationally, AI is known to be expensive; I would assume even more so for images. Please check me on this if I'm wrong.
I personally think 'AI' is functionally a product in open beta, and really should still be in alpha. But hey lets just start an AI arms race cause $$$. How could that blow up in our face?
The most important one for me is tech companies seem to think they can get away with stealing other peoples art to train their models on, and have generally gotten a slap on the wrist when caught. I do not like how its hurting our creative communities. It's so short sighted, because the only thing this does is create less space for real artists. This means there's less good data to consume, so your AIs may eventually start devolving for lack of a better term.
I agree with one of the above posts in that if you want to have a genAI image, just do it yourself.
people today already "borrow" googled art that may or may not have been meant for sale. a free AI service would expedite that taking and in so doing further normalize not paying artists for their work. on top of that it would be changing that unattributed art without author consent. a service with only consensual and attributed training data could work but would be either limited or expensive. AI itself isn't bad, but the people selling it aren't often leveraging their own sweat and effort.
Rollback Post to RevisionRollBack
unhappy at the way in which we lost individual purchases for one-off subclasses, magic items, and monsters?
tell them you don't like features disappeared quietly in the night: providefeedback!
It certainly looks like it, but polls are usually pretty subjective by their nature
I mean the poll itself is biased. This is interviewer bias; the way you word your answer choices could influence the responses. You have assigned judgement to one of your options and then charged it with a moral component by saying that this option protects the sanctity of art.
Personally, I am not very fond of AI being used to uproot people in an industry where it is already cutthroat and where corporations already held most of the negotiating power before AI even was mainstream, but I also don't think this poll will provide accurate data that is a reflection of how people necessarily view the topic.
second there is already dozens ai programmes for portrait creation , while wotc couldn't finish encounter builder tab properly
from my group 1 person use ai art, 4 not ai but likely art they not paid for, so if you don't have extra unique taste you would get what you want for free, and if you need something extremely specific comission from artist is only way to go
The key thing here is: for the things genAI can do well (images, backstory for minor characters, etc) there is nothing RPG-specific about them, so there's no reason to have a D&D genAI. I mean, they could slap a D&D logo on some standard product, but I'm not sure what it gets them. To have an actually useful product-specific AI, you need a large data set that is unique to you that you can use to train the AI, and Wizards doesn't have that.
With all the news and controversies surrounding AI, I wouldn't be surprised to see quite a bit of a negative response to this post. With that out of the way, Do you think it is possible that there are plans for a dedicated D&D Art AI where you can choose your settings? Class, Species, and the like, along with some text descriptions, and possibly a reference photo or two, for making character profile images to drop into your character sheets, and possibly use in other places. If there are, I suspect that it will be a few years before we see a beta of that announced at some point in the future.
What are your thoughts on the matter?
There aren't plans. Wizards has been extremely open on the fact that whenever AI art was alleged, they've altered it or disproved the fact that it was AI.
AI Art as it is can shove it. It's not about being negative and it's about the ethics behind it. There are no current large scale AI models that have been purely sourced and ethically produced based on material that was willingly given to it by all users. Reddit selling its data to Google for its AI to learn from wasn't anything that the people who have used Reddit for years and years ever could conceive, yet here we are. DALL-E 3 is known to have scraped the internet, but where it did that scraping is largely unknown, leading to mass copyright concerns.
If Wizards were to have a generative image model that was sourced from all art assets that Wizards owned and then it created art based on those images? I wouldn't have an issue with it because it would be using art that it owns. I'm sure artists would have an issue with it but sadly when you create art for a company, they own the art asset typically forever. You could also solicit people to add art assets into it for a nominal fee to keep a source of art flowing in to allow the model to continually learn, and then have a system where people voted on AI generated art that was considered "good" to reinforce behaviors. That'd be the most ethical way to do it, but we're a long way from that and even longer for people to accept it.
I think that it would be a great idea. AI applications are great tools and the artwork, backgrounds etc we can get out of it today are amazing. Saves a lot of DM prep work when used as a co-pilot, not an auto-pilot.
AI-tools cannot replace human creativity, but they sure can help you visualize your creative ideas and help create framework for backgrounds, descriptions etc.
If I had a human artist that could draw my characters for me that would be sweet but I do not have access to that or at least no one who has that sort of extra time. I see this as a tool to create your character. Your are making the prompts and being specific and shaping the image to your vision so even though GenAI is doing the processing it is unique more or less. Sure a lot of images can look similar but you can reject those. My first crack at an older half-elf pirate was awful. Too beautiful and perfect. But subsequent tries looked much better. Also when designing weapons it is super helpful. Though again to get my longhammer correct took hours of attempts.
It might be good to see the DnD groups/sites create better tools for this.
What are the implication of 3-D VTT's, re: character art? Might AI art generators play a part in fulfilling a possible need (want), on that front?
Existing AI art generators clearly have no concept that they're even representing something that could be 3d, so they aren't really applicable to VTTs. It's far more likely that you wind up with something like a video game character builder. It's not like you can't use something like BG3 to build your character portraits already.
With all the news and controversies surrounding AI, I wouldn't be surprised to see quite a bit of a negative response to this post. With that out of the way, Do you think it is possible that there are plans for a dedicated D&D Art AI where you can choose your settings? Class, Species, and the like, along with some text descriptions, and possibly a reference photo or two, for making character profile images to drop into your character sheets, and possibly use in other places. If there are, I suspect that it will be a few years before we see a beta of that announced at some point in the future.
What are your thoughts on the matter?
It's "the reference photo or two (presumably user provided)" where things get very problematic, there are many digital tools where a digital avatar can be sculpted from video games to HeroForge. WotC seems willing to draw the line at "and I want it to emulate the style of this particular artist" and if you are talking about a true reference photo, another line at "potential dumbfakes of IRL people whom the user may not have the right to."
I mean, it's possible WotC could develop a controlled and closed referenced base (with all references knowingly produced and submitted to the AI's resources), but the AI industry to date hasn't proven itself able to be so for the most part (and really why go through all the teaching of the AI to basically mimic what existing customization tools seen in video games and sites like Heroforge already do?) and I don't see WotC becoming an "AI company" to do it on its own.
Jander Sunstar is the thinking person's Drizzt, fight me.
Honestly, there's no point; it's not anything that existing AIs aren't already competent at.
I think I get what you're saying. That said, it almost seems like an inevitability at this point, with how things seem to be going currently.
That's a pretty biased poll.
DM mostly, Player occasionally | Session 0 form | He/Him/They/Them
EXTENDED SIGNATURE!
Doctor/Published Scholar/Science and Healthcare Advocate/Critter/Trekkie/Gandalf with a Glock
Try DDB free: Free Rules (2024), premade PCs, adventures, one shots, encounters, SC, homebrew, more
Answers: physical books, purchases, and subbing.
Check out my life-changing
It certainly looks like it, but polls are usually pretty subjective by their nature
Generative AI, I would think, is expensive computationally, and filled with 'hallucinations' though it seems to have gotten better at preventing that. Not something I would release to a large audience to use adhoc without some sort of rails.
The problem I see and would rather not be party to:
I agree with one of the above posts in that if you want to have a genAI image, just do it yourself.
edit: some grammar, some clarification
people today already "borrow" googled art that may or may not have been meant for sale. a free AI service would expedite that taking and in so doing further normalize not paying artists for their work. on top of that it would be changing that unattributed art without author consent. a service with only consensual and attributed training data could work but would be either limited or expensive. AI itself isn't bad, but the people selling it aren't often leveraging their own sweat and effort.
unhappy at the way in which we lost individual purchases for one-off subclasses, magic items, and monsters?
tell them you don't like features disappeared quietly in the night: provide feedback!
I mean the poll itself is biased. This is interviewer bias; the way you word your answer choices could influence the responses. You have assigned judgement to one of your options and then charged it with a moral component by saying that this option protects the sanctity of art.
Personally, I am not very fond of AI being used to uproot people in an industry where it is already cutthroat and where corporations already held most of the negotiating power before AI even was mainstream, but I also don't think this poll will provide accurate data that is a reflection of how people necessarily view the topic.
DM mostly, Player occasionally | Session 0 form | He/Him/They/Them
EXTENDED SIGNATURE!
Doctor/Published Scholar/Science and Healthcare Advocate/Critter/Trekkie/Gandalf with a Glock
Try DDB free: Free Rules (2024), premade PCs, adventures, one shots, encounters, SC, homebrew, more
Answers: physical books, purchases, and subbing.
Check out my life-changing
first of all options wording is kind of loaded
second there is already dozens ai programmes for portrait creation , while wotc couldn't finish encounter builder tab properly
from my group 1 person use ai art, 4 not ai but likely art they not paid for, so if you don't have extra unique taste you would get what you want for free, and if you need something extremely specific comission from artist is only way to go
Nah, **** AI.
The key thing here is: for the things genAI can do well (images, backstory for minor characters, etc) there is nothing RPG-specific about them, so there's no reason to have a D&D genAI. I mean, they could slap a D&D logo on some standard product, but I'm not sure what it gets them. To have an actually useful product-specific AI, you need a large data set that is unique to you that you can use to train the AI, and Wizards doesn't have that.
There aren't plans. Wizards has been extremely open on the fact that whenever AI art was alleged, they've altered it or disproved the fact that it was AI.
AI Art as it is can shove it. It's not about being negative and it's about the ethics behind it. There are no current large scale AI models that have been purely sourced and ethically produced based on material that was willingly given to it by all users. Reddit selling its data to Google for its AI to learn from wasn't anything that the people who have used Reddit for years and years ever could conceive, yet here we are. DALL-E 3 is known to have scraped the internet, but where it did that scraping is largely unknown, leading to mass copyright concerns.
If Wizards were to have a generative image model that was sourced from all art assets that Wizards owned and then it created art based on those images? I wouldn't have an issue with it because it would be using art that it owns. I'm sure artists would have an issue with it but sadly when you create art for a company, they own the art asset typically forever. You could also solicit people to add art assets into it for a nominal fee to keep a source of art flowing in to allow the model to continually learn, and then have a system where people voted on AI generated art that was considered "good" to reinforce behaviors. That'd be the most ethical way to do it, but we're a long way from that and even longer for people to accept it.
Makes sense.
I think that it would be a great idea. AI applications are great tools and the artwork, backgrounds etc we can get out of it today are amazing. Saves a lot of DM prep work when used as a co-pilot, not an auto-pilot.
AI-tools cannot replace human creativity, but they sure can help you visualize your creative ideas and help create framework for backgrounds, descriptions etc.
And they can make for a decent substitute when the option for a human artist isn't a viable option, but a human artist is still the best option.
If I had a human artist that could draw my characters for me that would be sweet but I do not have access to that or at least no one who has that sort of extra time. I see this as a tool to create your character. Your are making the prompts and being specific and shaping the image to your vision so even though GenAI is doing the processing it is unique more or less. Sure a lot of images can look similar but you can reject those. My first crack at an older half-elf pirate was awful. Too beautiful and perfect. But subsequent tries looked much better. Also when designing weapons it is super helpful. Though again to get my longhammer correct took hours of attempts.
It might be good to see the DnD groups/sites create better tools for this.
What are the implication of 3-D VTT's, re: character art? Might AI art generators play a part in fulfilling a possible need (want), on that front?
Existing AI art generators clearly have no concept that they're even representing something that could be 3d, so they aren't really applicable to VTTs. It's far more likely that you wind up with something like a video game character builder. It's not like you can't use something like BG3 to build your character portraits already.
The closest solution I see would be going to hero forge and buying the STL that you make, which could then import into the vtt.