Using an AI art generator to make graphics

Hello everyone,

If you played my game Midsummer’s Eve, you might have seen in my notes that I used Midjourney, an AI art generator, to make the graphics. Here is a blog post explaining how I did that.

I’m mainly posting this to share the blog post for anyone interested, but feel free to use this post to discuss the upsides, downsides, ethics, etc. of using AI-generated art for interactive fiction and text adventures.


Well, I don’t think anyone should be doing it for now, for ethical reasons.

Put simply, the AI would not be able to function without the scraped images. They have scraped a zillion images they had no permission to scrape, vast amounts of copyrighted work by visual artists, presumably including myself, if not already, later. The scale of the data is probably beyond easy human comprehension (I mean, it’s beyond mine. I can’t picture or conceive of it all at once in my head.)

I think the AIs should be reset to go where it has permission, or there need to be new services that scrape only from where they have permission. You could view this as a similar solution to what Unsplash eventually provided for stock photos. I am kind of stunned by the thoughtlessness of the people who programmed these AI who didn’t think about what seems a pretty obvious issue, or who did and went ahead anyway.

I have said what I said above in a few different venues, and I often get a “AI’s here! Get on the wave or be crushed by it.” AI is indeed here, but this binary of actions is nonsense. Now is the time you can perform many other actions – like push for ethically functioning AI and reject the current unethical tools.

There are various court cases going and maybe concepts will be redefined (legally or in culture) by the time all the dust settles. But the world is how it is now, and so for now, I don’t think anyone should use these tools, as all I can think of is the exploitation of artists who, in any other context at all, would have had to give permission. None were asked, or had a chance, and the AI doesn’t (and can’t) do attribution.

Due to the complexity of all this, I won’t post again in the topic. I’ll just state my home position.



Thanks for sharing!

Never say never :grin:


The Secret of Darkwoods is another piece of IF that uses AI generated images: The Secret of Darkwoods on Steam

I don’t want to open the “AI ethics” can of worms all the way, but in short, my advice for artists who are bothered by the prospect of other artworks possibly containing traces of a zillionth of their own works would be to simply never show their art to anyone and just keep it locked in a drawer where nobody can ever “steal” it :wink:


I get concerned with the idea of artists, musicians, programmers, writers, animators losing their jobs. We’ll always need human originality and ingenuity, but employers willing to pay for that will be much less. AI medical assessments, psychiatric care, security surveillance, transportation, fake users on social media… it eventually will all be AI.

Before, we were worried about robots taking over our jobs, but now I worry about AI taking over our creativity.

Your game doesn’t bother me for having AI generated art, to be honest. Your topic just got me thinking, is all.

Oh, and thank you for sharing your blog. Great job on that!

I have to ask if you’ve ever thought about going all the way with the retro pixelated look? I noticed that your game jumps in rendering size (at even pixel increments of 360p, 720p, 1080p – 1x, 2x, 3x).

I’ve attached an image comparing what you have on the left and what I’m talking about on the right:

Edit: You’ll have to excuse the slight fuzziness of the JPG format.


This topic will probably haunt us for a while.

What would really interest me personally: To what extent can you actually call the use of artwork from the Internet as training data really theft? What are the AIs doing with it? Is the source material really recognizable?

Are there any good articles on this that show what Midjourney & Co. do with the training data and where a connection between training data and AI output becomes apparent?


I know this is barely related, but @Perry_Simm, @HAL9000… Both of you with names based off fictional AI systems next to each other in a chat about AI - seems funny to me!


Are humans also trained on copyrighted work? :stuck_out_tongue:


Thanks for posting this about your AI experiences. A lot of people have been telling me to use AI. Apparently, it’s the thing. Or soon will be!

Aside from ethics, my reservations have been mostly along the lines of the problems you describe in your blog posts. so thanks for writing that up.

My guess is that your consistency was largely down to asking for it all to be like “Monkey island”. Having a style anchor point like that could work, albeit with a lot of editing. But without a style anchor, you’re stuffed!

For info, I’m currently using 3D for my artwork. it’s a lot of work, but on the upside, once i have a scene built, i get a lot of art from it;

For example:

  • multiple viewpoints
  • easy to change or rearrange the content to reflect game change
  • character consistency.
  • re-render bits when i suddenly realise the game design has to change!

Once I’ve built a character, i can pose them in scenes how i wish with 100% consistency between views.

Recently, i tried making a promo video. It was a simple 360 cam flyby of an existing scene with a bit of character animation. I thought to myself: How would you do this with AI?

3D doesn’t give me that “drawing” look, which AI does do very well. Personally, i prefer the “realistic” look. But it’s entirely a matter of taste (and to some extent game genre).


To catch up with what people are saying about ethics, I’d like to point out something the other way around;

One of my concerns is AI generating a result that happens to have taken too much out of something copyright out there. How would I know, unless I’ve seen it before?

Let me give an example from the text based chat prompts:

I asked chatgpt to invent an original joke about pirates. It said:

Q: What do you call a pirate with two arms and two legs?

A: a rookie!

Actually quite witty. BUT, it’s not original, I’ve head it before (maybe with slightly different words), but it hasn’t been “invented” by AI.

How do i know that’s also not happening with “generated” AI art? Next up, someone says, “Hey that’s the castle from Disney Island”, or something. Then what?


Ethics aside, which isn’t something we will be able to come to any conclusions in this thread - there are arguments on both sides and there are better forums for discussing the wider issue of generative AI in creative sectors. What I think we can usefully discuss is the impact of using AI art in interactive fiction, and here are my thoughts:

  1. The bulk of interactive fiction is not commercial - we don’t have budget for art and there are probably only a small minority of us that have the artistic skills required to illustrate an amateur adventure. If we don’t use AI art, the majority of text adventures will be purely text (which I’m not saying is necessarily a bad thing - I happen to prefer ‘theatre of the mind’ adventures). Artists are not losing income by an amateur community using generative AI for non-commercial projects, nor is anyone profiting.

  2. How much more engaging are adventures that are accompanied by art? Does it make it more accessible to a younger audience? There is the potential for purely text-based interactive fiction to be overlooked by new people joining our community because they have an expectation of graphics/art. If multi-media adventures are more likely to hook a new audience but we don’t pander to their preferences, will our niche hobby become stale (or atrophy)?

  3. If text-only adventures might to be overlooked by people looking for a multi-media experience, the barrier to entry for new authors also becomes much higher. Use of AI art can be a way for a new author to create an eye-catching game that attracts attention/players and encourages emerging talent. This doesn’t guarantee that a game is good, but it probably does increase the likelihood that it gets played and that the author gets feedback, which should improve quality over time.

Generative AI has the same problem as taxes - if there are legal loopholes that some companies take advantage of but others don’t (for moral reasons, or whatever), the companies that don’t use those loopholes operate at a disadvantage. Just as authors that take advantage of generative AI will have an advantage over those that don’t. To compete in business, you have to take advantage of (legal) tax loopholes just as I foresee authors will need to take advantage of generative AI to compete for attention. To create a level playing field for everyone (w.r.t. taxes or automation), legislation is required (way beyond the scope of our community to lobby for - but please do join groups lobbying for this if this is something about which you feel passionately, assuming your country allows for you to do so). Authors can take a principled stand and not use generative AI to help them write code, plot an adventure, generate art or music, etc., which is entirely their choice - but then the question becomes, will that game get played as much or more than the richer multi-media game (even if it has a superior story)?


If you’re trying to make money, maybe that last bit is true (although then your first point no longer applies), but if not, “you have to do whatever you can to get ahead, even if it’s unethical, as long as it’s legal” is a pretty joyless way to think about a hobby. There are already dozens of things that would make my games more popular that I don’t do because I don’t want to. And I’m not sure your basic assumption that games without art are more niche and doomed to be less successful is even true—look at Choice of Games.


I do take issue with the assumption that generative AI is unethical in all cases (strongly, in fact), but even if that were true, that wasn’t really my point. Perhaps I could better articulate it as the personal choice we have regarding using generative AI. Is eating meat unethical? Arguably; and people have the choice to become vegetarian/vegan/whatever. Are computers, with all the electricity they consume, unethical? Again, arguably; and people have a choice to buy a CYOA book from a charity shop instead of playing computer games. If you have a moral compunction against generative AI, then don’t use it, but there also shouldn’t be any judgment of those people who do because they are motivated by having more people play their games (which calling someone ‘joyless’ would seem to be). I appreciate that I am blurring ‘ethics’ and ‘morals’ in that defence, which might be lazy of me, but I hope my position is clearer now.

I support this completely. In fact, I do this myself - choosing not to sensationalise a game based on what is fashionable but preferring that which is narratively and thematically consistent with the game I am choosing to make. And the best games I play/films I watch/etc. are the ones that are internally coherent. The best advice I think I ever heard about game design is to make the game that you want to play. It might not make it popular, but it will make you proud (and if I’m not creating text adventure games commercially, isn’t it more fulfilling to create something of which I’m proud than something with a broad audience?). That’s all separate from my point about whether art in an adventure game makes it more likely to be played, though.

It is just an assumption - one I’m not sure how to test (ideas, anyone?). Are people on this forum that are new to the hobby (say, within 6-12 months) who would like to comment on whether they were more or less likely to play a game based on whether it has graphics when they were getting into text adventure gaming? Old hands like us probably don’t care either way (and I actually prefer artless games because I prefer the theatre of the mind anyway), but I was brought up on pure text adventure games. For younger gamers that have been brought up on more visual games, is the lack of graphics going to be a factor in deciding which games they would play? Probably a topic for another thread. But I guess for my point to stand, it would only require just one gamer to be turned off by the lack of graphics, and I woud suggest there’s a high likelihood that of all the younger gamers out there curious about text adventure games, there is going to be at least one that would be swayed towards playing a game with graphics than one without.


I’m sorry, I should have said “even if you believe it is unethical.” Also, what I was calling “joyless” was specifically looking at a hobby as if it were a business scenario where you have to do whatever it takes to keep up with/get ahead of your competitors, not simply wanting an audience for your work. I hope that clarifies my position as well!


Then I completely agree wholeheartedly. If you’re compromising your personal ethics to court popularity, then I imagine that would be pretty dissatisfying. Again, a topic for another thread, but I wonder whether the monetisation of popularity via services like Instagram is creating a cultural shift towards shallower ethics?


The copyright laws we have today, while perhaps meant originally to protect authors, have turned out mostly to the benefit of the publishers instead. I am quite confident they will remain in force for you and me but of course waived for the internet giants.


AI already generates some pretty convincing fully generated videos… don’t get me started on deepfakes and voice mimicking. Google “AI video”.

The music industry will be churning out songs that are very catchy to our primitive minds, but ultimately it’s just an amalgamation of what we’ve learned and already discovered. There’s only a handful of people who are composing the majority of music in the North American pop scene today, but we’re not too concerned about that.

I should also say that I don’t think AI will be the ruin of society. It’s going to reinvent a lot of what we already know. Somehow, someway, we’ll still appreciate art. I was just bring up a concern about relying too heavily on AI and it passing for originality.

AI may actually push us to be more original. When someone says that “looks like AI art”, it probably means it looks like 90% of what’s out there, just mashed together. True artists will always find a way to push the boundaries beyond what AI can conjure.

We’ll also find people who really know how to get the most out of AI. They’ll be artistic engineers, so to speak. There are people doing this already with amazing results that go beyond what the average person can do. Even Tristin had to clean up the generated art for her game and it was definitely better for it.

My dream scenario is that AI will help us create unique experiences (not art, per say) with ease. Like when the the crew in Star Trek talk to the Holodeck computer to generate a world and characters. That’s a good application of AI, but we have to go down the path we’re on to get there.


I read a really interesting article about how AI will not replace the creative workforce, but it will replace the creative workforce that doesn’t use AI. Automation in manufacturing hasn’t replaced manufacturers of bespoke items any more than, as you suggest, AI art will replace custom and experimental art pieces, but it has replaced the manufacture of commodity items.

I am reminded of Bernard Suits’ thesis, The Grasshopper, in which he posits that Utopia is achieved when everyone plays games (I’m summarising A LOT) because when there is no work to be done, that is what it is in our natures to do (and his definition of “game” is pretty broad). I would add that because games do not spontaneously appear, they must be created; so Utopia will be defined when everyone in society is creative - whether they are aided by AI or otherwise.


A part of me wishes someone would create a more ethical option. An AI art generator that was specifically trained on public domain material is certainly plausible. I understand there’d be a trade-off in quality, but many would feel better about using it.

With that said, this reminds me of some of the arguments about pirating. Many proponents of piracy argue that much of the pirated material would not have been purchased in the first place, meaning the only reason you’re downloading, say, Korean language learning software is the fact that it was free. You would simply not purchase it at all at, I dunno, let’s say $400. If this is true, then the act represents no loss of income for the copyright holder, as you were never remotely a potential customer. There are plenty of counter-arguments to this.

Anyway, this ties into why the other part of me wonders if a more ethical AI art generator is still problematic at best. Even if you had a “more-ethical” AI art generator trained solely on public domain images, you could argue that folks would use this method in lieu of paying an artist for their services. This could potentially negatively impact folks who make a living this way. And this is where the old piracy argument comes into play. If you’re making a free hobbiest interactive fiction game, would you have paid for art otherwise, or would you have simply gone without? If the latter, does your use of a public domain trained AI art generator represent a loss for anyone? Also, are people being entirely honest with themselves? Meaning, will people who might have purchased art otherwise tell themselves they wouldn’t have, and use this as a comforting rationalization? I honestly do not know.

I won’t tar and feather anyone using it, but I’ll still sit on the sidelines myself.


As an artist, and someone who is friends with both hobbyist and professional artists, my stance is pretty simple and one I’ve also seen reflected by game developers in other spaces (such as the visual novel genre, where art is a huge factor in garnering both interest and sales.)

AI art is exploitative and hugely damaging from an environmental perspective. I’ve seen it damage my friends’ ability to make a livelihood, and how so often advocates of it don’t value the craft or artisans.

I refuse to support or promote any artwork that includes its use, and if I find out at a later date that I accidentally did, then I’ll rescind it. I won’t use it myself, or platform it.

It’s a major black mark against your reputation in most artistic circles to use it, especially when people attempt to pass it off as having been made entirely by themselves. It alone will put me off entirely from ever checking out a developer’s body of work, and I won’t recommend or review it knowingly.


I really love this idea and first came across it in “Lord of All Things” by Andreas Eschbach (great book!!). Not to add another controversial/political element to this thread :smirk:, but this idea is why I’m a proponent of UBI.