|
Hello! The time has come for the next generation for this thread! So what is it about? AI tools can be a great resource for generating new content for traditional games like Dungeons and Dragons and Magic the Gathering and other TTPRGs. These tools can help you come up with new spells, feats, magic cards, character art, backgrounds, battlemaps, Visual Novel scenes, DM for you! And more! Quickly and easily. But of course be warned that getting useable results can take a lot of time and effort, and in some cases even dedicated hardware; and sometimes it might be easier to get the outcome you want by commission a human creative for your specific project if you have the means. But what are AI tools? What is AI? Most of the answer is beyond the scope of this thread, but in general in my opinion AI driven content generation and similar tools can cover using Perlin Noise and Random Number Generators via deterministic algorithms to create interesting and useful content ad infinitum like infinite words like Minecraft to more specialized programs that rely on "learning" algorithms and extensive training data sets to produce output that might seem similar to something a human can produce. Your mileage may vary as to how similar it may or may not be. This thread is going to be more about discussion and sharing of such tools and ones results and experiences with these tools. While the broader ethical discussion and discourse of "AI Art", and how they acquire their training data, is important and absolutely deserving of focus and attention, that is not the focus of this thread. Below is a list of tools that might be useful, however to learn more about AI and the ethicacy surrounding their use, you can go to the AI Debate thread in D&D. And to learn more about the technology behind AI like Machine Learning you can go to the Caverns of Cobol thread here. Stable Diffusion: https://playgroundai.com/ (this is just one example, there are many others) Stable Diffusion is a diffusion-based image creator that's trained on a massive dataset of scraped and tagged art from the web. It presently comes in three versions: 1.5, 2.1, and SDXL. 1.5 is capable of running on fairly middling hardware and plenty of user-supported tools but produces smaller and less coherent images. 2.1 is generally considered a failed experiment and has very little support both officially and unofficially. SDXL was released in July 2023 so is fairly new at the time of writing and produces larger and more coherent images, but has less user-supported tools and requires upper-end hardware to be effective. Midjourney: https://www.midjourney.com/ This AI tool is designed to generate new images and graphics for use in games and other creative projects. It uses a deep learning model to generate original artwork based on a set of user-specified parameters. If you're looking for new magic cards or other graphics for your game, this tool might be worth checking out. ChatGPT: https://chat.openai.com/chat This AI tool uses a variant of the popular GPT language model to generate text-based content. You can use it to generate new descriptions for spells or magic items, or to come up with ideas for new quests or adventures. ChatGPT is also highly customizable, allowing you to fine-tune the type of content it generates based on your specific needs. Additional Resources: Here is a link to VTTRPG Resources, which seems to be a how-to/useful tips and tricks guide to prompts for getting useful content for TTRPGs. Overall, these AI tools can be a great resource for generating new content for traditional games like Dungeons and Dragons and Magic the Gathering. Whether you're looking for new spells, magic items, or quests, these tools can help you come up with fresh ideas and add some excitement to your game. This thread is for encouraging you to share your experiences with AI tools and how you've used them to generate new content for your roleplaying games. This can be a great way to inspire others and help them get the most out of these tools. For example, you might share: 1) Examples of new spells, feats, or magic items you've generated using Stable Diffusion or ChatGPT 2) Artwork or graphics you've created using Midjourney or Dall-E 3) Stories or adventures you've created using AI Dungeon or other interactive storytelling tools Examples, surrounding text/context trimmed for brevity: Battlemaps! Humbug Scoolbus posted:For the maps, midjourney kicks out really good battlemaps actually since v4. Humbug Scoolbus posted:Bonus MJ cyberpunk battlemaps... VN Style Cutscenes! (Sadly the links here seem to be broken, I will restore them as soon as I get working links) Fuzz posted:Just posting here so I can link said post in another thread and not get probated yet again: Portraits! Doctor Zero posted:
AI DM! feverish and oversexed posted:I'm using a combination of different AIs to run ironsworn. I've been obsessed with getting AI to be my DM since chatgpt4 was released, and I've had varying levels of success. All my testing led me to solo ttrpg which I hadn't heard of before, and I just wanted to talk to other peeps doing similar things. Magic the Gathering Cards! Custom D&D Content like Spells! quote:Spell Name: Castle Keeper's Touch And all sorts of styles! Skios posted:And some NPC illustrations: Skies the limit! However remember that there's probably no one size fits all solution to generating your content. Many outputs at first might be mediocre and require significant guidance and handholding. By sharing your experiences, you can help others see the potential of these tools and give them ideas for how they can use them in their own games; and you can learn how to further improve the output you might derive from these tools. You might also discover that others have found new and creative ways to use these tools that you hadn't thought of before. In addition to sharing your experiences with these specific tools, we also encourage you to share any other tools you know about that might be useful for generating new content for roleplaying games. There are always new tools and resources being developed, and by sharing your knowledge, you can help others discover new ways to enhance their games. So don't be shy! Share your experiences with AI tools and other resources for generating new content for roleplaying games. You never know who you might inspire or what new ideas you might discover. Guides & Tutorials (Expected to be WIP and continuously updated) KwegiboHB's Observations: Part 1 Part 2 Megaman's Jockstrap Stable Diffusion Guide: Link BrainDance's Guide on running local LLM's: Link Rules Section TBD Link to the Previous Thread This post originally generated by ChatGPT, modified Sept 6 2023 by a human. Raenir Salazar fucked around with this message at 23:01 on Sep 7, 2023 |
# ? Sep 6, 2023 23:54 |
|
|
# ? May 2, 2024 05:29 |
|
With one of the games I'm in I started very quickly generating our sessions right after and sharing pictures with the group. Once I started doing them in SDXL it went over really well and now it's just a thing we do, and I think it adds a lot to the game. They're very flawed because it's just about getting something cool out right after the session, but SDXL makes it easy. We've had a short hiatus because one of our DMs just had a kid, the other who's sorta taking things over just started a new job and has to get situation. So this is from one of our sessions in July. It's a horror themed campaign. This is from when we went into the sewers, and it's just rats on rats, but turns out a lot of these rats used to be people. We run into a rat king at the end because of course you do. Everything to do with rats turned out great. Our warlock My character, the professor, a clown fairy. This is what I mean by flawed, the wings are just wrong. If I spent more than a minute on this though I think I could actually get this to be great, I might go back and fix it when I have the time. I made a bunch of other ones that were not flawed, I just really, really liked this picture. Most of the "clowns" it gives me, especially with the horror theme, are IT, the professor isn't IT. The sewer itself, gets a little over the top for the sewer but, it's Barovia, who knows how their sewers are. The first rats, I think these are great. It's really hard to mess up a rat, because rats just are messed up. The humans turned rat. You can see a few little flaws, again mostly things that could probably be fixed without too much work, like how his elbow is really another hand (though I actually think I wouldn't want to fix that, even if it's an error it's kind of cool), the position of one of his feet, and little bits of disconnected rat here and there. But, just overall, this is kind of great. The rat king, I think this is just kind of flawless. This took me a bit to get though, it doesn't actually have the concept of a rat king or at least not enough to counter the concept of a rat that is an actual king. All generated with plain SDXL in ComfyUI, dpmpp_2s_ancestral It's a little bit cheating when you're doing this kind of horror because the flaws aren't so apparent but, still, it's not hard to get good results with SDXL. I think this is going to be really cool after we've finished the whole game and we have a complete record of everything this way. BrainDance fucked around with this message at 09:50 on Sep 7, 2023 |
# ? Sep 7, 2023 09:46 |
|
I've been using Midjourney for generating NPC/Enemy pictures for a modern day sci-fi urban fantasy campaign I'm running, and it makes some great ones...
|
# ? Sep 7, 2023 17:18 |
|
Short post here but Raenir Salazar could you please fix this description of Stable Diffusion? "This AI tool is designed specifically for generating new content for Dungeons and Dragons. It uses a machine learning model trained on a large dataset of existing spells and abilities to generate new, unique spells and abilities. You can use it to come up with new ideas for your character's abilities, or to add some excitement to your game by introducing new spells and abilities that your players have never seen before." This is just totally wrong and I'm assuming you had an AI write it. Please replace with this: "Stable Diffusion is a diffusion-based image creator that's trained on a massive dataset of scraped and tagged art from the web. It presently comes in three versions: 1.5, 2.1, and SDXL. 1.5 is capable of running on fairly middling hardware and plenty of user-supported tools but produces smaller and less coherent images. 2.1 is generally considered a failed experiment and has very little support both officially and unofficially. SDXL was released in July 2023 so is fairly new at the time of writing and produces larger and more coherent images, but has less user-supported tools and requires upper-end hardware to be effective."
|
# ? Sep 7, 2023 22:27 |
|
Megaman's Jockstrap posted:Short post here but Raenir Salazar could you please fix this description of Stable Diffusion? Some aspects of the OP I copied from Rutibex's OP, who presumably DID write chunks of it with AI. I'll be happy to make the change but I do wish to point out I posted WIP's OPs with that segment around 3 different times in the previous thread and no one pointed it out until now. So definitely not my fault e to add: So technically your description describes it on a technical level and presumably with context we know we can use it to generate images, can you flesh out more what you can use Stable Diffusion for? Raenir Salazar fucked around with this message at 23:00 on Sep 7, 2023 |
# ? Sep 7, 2023 22:57 |
|
Fair enough!
|
# ? Sep 7, 2023 22:58 |
|
Lol I made Chat GPT write a scene where this random ogre I wrote a little backstory for explains his life philosophy. Lug, the massive ogre with a brutish appearance, sits on a makeshift throne made of bones and torn cloth. The cave is littered with the remnants of his victims, creating an eerie atmosphere. Lug speaks with a deep, gravelly voice, his words punctuated by his heavy breaths. LUG (stroking his chin) Life, it's a simple thing, it is. Eat, survive, repeat. That's the way of Lug. Lug picks up a large femur bone and examines it, a sinister grin spreading across his face. LUG (savoring the thought) When Lug was but a wee ogre, the world showed him its true colors. The weak, they suffer, and the strong, they feast. He hurls the bone against the wall, causing it to shatter. LUG (growing animated) Humans, they think they rule the land. But Lug knows better. We are the true lords of this wilderness. He flexes his massive python arms, veins bulging beneath the green-hued skin. LUG (voice filled with conviction) Lug's strength, his might, it's what sets him apart. The humans, they fear me. They fear my power, my wrath. Lug rises from his throne, towering over the scattered remains of his victims. LUG (pointing to the bones) Look at them, pitiful and weak. They couldn't survive, couldn't adapt. They're food, nothing more. He walks towards the cave entrance, his eyes gleaming with primal intensity. LUG (facing the darkness) The world, it's a harsh place, a struggle for survival. Lug thrives because he embraces his nature, his ogre self. He turns back to his throne, a wicked grin spreading across his face. LUG (raising his voice) In this world, there is no room for weakness. Only the strong prevail, and Lug is the strongest! Lug lets out a mighty roar, echoing through the cave, reverberating with his savage philosophy.
|
# ? Sep 9, 2023 04:32 |
|
How heavily edited are the results you utilize in your games? When I use Chat GPT to create names, personalities, and general background details of NPCs in a tavern for instance, I usually end up having to rewrite at least half of it with details and specifics adjusted to my game and my intention with that NPC. I try adding more details in the prompt, but that tends to just make the edited needed in different ways. Are there better tools or is this a pretty common experience?
|
# ? Sep 9, 2023 18:11 |
|
I don't know about your specifics, but there are absolutely ways to control things better with better prompting. (Feel free to share a before/after example with prompt) It's often useful to think of the prompt as if you were giving a writing assignment to someone, but there are also ways to better structure things to get where you want. If you're into it, check out the documentation on best practices if you haven't yet: https://help.openai.com/en/articles/6654000-best-practices-for-prompt-engineering-with-openai-api It has loads of helpful stuff that might get you thinking in different and maybe helpful directions. Again, feel free to share examples of what you get vs what you actually find useful, and your prompt.
|
# ? Sep 9, 2023 18:20 |
|
Oh, I just noticed there was a new thread! Thank you for writing up, Raenir. I don't have any new projects to post, since I am not GMing anything new. But here are some very fancy Aaron Eckharts:
|
# ? Sep 9, 2023 20:58 |
|
Hi, I've been tinkering with a ttrpg rule system / settings to force upon my nerdy friends and here's a bunch of prototyping / concepts. It's generated in bing since it tackles my weird long prompts pretty spot on and I managed to make it mimic my own style kind of (it even fucks up the hands as I do!) if I ever have 100+ hours to churn out my own drawings for it instead. The setting is 80's fantasy put through a blender with disney, body horror and old nes 8-bit games. It's a purposely obscure world as I'm trying to trigger the players imagination in a similar fashion to how mine went off when I saw pixels trying to paint a whole world on a screen for the first time. The players won't have a world map to point out locations by, since the laws of physics are messed up and instead the specific spots of locations are there so that players hopefully imagine a blurry picture of the places between (which in itself part of the setting and how magic works). End of rant. Oh and everybody's an animal and the only humans are melted giant blobs of flesh. Here's some player character concepts Imaginary Friend fucked around with this message at 05:49 on Sep 16, 2023 |
# ? Sep 16, 2023 05:44 |
|
Those are trippy as hell. Rock on.
|
# ? Sep 16, 2023 06:46 |
|
Fed the rabbit into a research tool for making a short animation (but not directly animating the original) I get messed up results a lot, but they're occasionally kind of neat https://i.imgur.com/yNA2RZv.mp4
|
# ? Sep 16, 2023 08:16 |
|
Imaginary Friend posted:Hi, I've been tinkering with a ttrpg rule system / settings to force upon my nerdy friends and here's a bunch of prototyping / concepts. It's generated in bing since it tackles my weird long prompts pretty spot on and I managed to make it mimic my own style kind of (it even fucks up the hands as I do!) if I ever have 100+ hours to churn out my own drawings for it instead. That is a very dope aesthetic, yes.
|
# ? Sep 16, 2023 09:25 |
|
The Eyes Have It posted:Fed the rabbit into a research tool for making a short animation (but not directly animating the original) The Eyes Have It posted:Those are trippy as hell. Rock on. Here's the crude summary it churned out of the world: quote:Absolutely! Let's refine the summary to align with your preferences: And here's after I brainstormed one location: quote:The tragic fate of the swine king and his kingdom paints a vivid and haunting image. The melding of citizens with the gold and the agony experienced by the king add a layer of horror and sadness to the story of this lost kingdom. The gold fortress now stands as a grim monument to the twisted consequences of contact with the anomaly.
|
# ? Sep 17, 2023 19:53 |
|
This reminds me of a couple tips I have for people using ChatGPT to help with worldbuilding or scenario development or whatever. Both help cut down on the need for post-editing the results, by a) limiting length, and b) giving more structure. 1) ChatGPT responds very well to being given a length constraint. "A few sentences", "about 50 words", "one paragraph", and so on works very very well. Length is actually very useful to constrain. 2) Results can really benefit from being given a structure to follow. e.g. "Provide a conflict, some leads to follow, and a boss battle" can work wonders when asking it to create a scenario. Ditto for worldbuilding details, character backgrounds (e.g. "describe a personal crisis that challenged their value system, how they overcame (or failed to overcome) the crisis, and how the events changed them") especially useful to combine with #1 ("... in a few sentences") so it knows you want an outline, not a novel. 3) GPT3.5-turbo is very capable, but when it comes to more complex story type stuff GPT-4 can be noticeably better.
|
# ? Sep 18, 2023 04:26 |
|
Haha you should see those descriptions before they left my brain. I kind of use it backwards; I have a bunch of detailed but scattered ideas and want to connect them to a world for which I have some constraints and guidelines I don't want to stray away from.
|
# ? Sep 18, 2023 07:36 |
|
I'm designing an Arkham Horror the Card Game-esque fantasy-themed game and wanted to show this off: I used MJ to make the card art, icons, and background texture. However, the interesting part is the QR code. If you scan it, it'll take you to a website that has an animated gif version of the card art that was created using Gen-2. Please check it out! As a next step that I haven't done yet, I thought it would be fun to include in-game lore easter eggs into the last few frames of the gif animation. I think this could be a really interesting way to mix analog and digital. It's a cooperative game, so while your buddy is taking their turn, you could scan cards and enjoy looking for lore stuff in the animated digital elements.
|
# ? Sep 19, 2023 01:37 |
|
New audio tool from stablity.ai -- really easy to use. I've been mucking around with it https://stableaudio.com/ I got a pretty nice track right off the bat with a prompt of "Vocoder robots sucking the souls from human minds in cryo-sleep, deep space, inevitable procedural beat. Haunting string and synth melody in background."
|
# ? Sep 19, 2023 04:55 |
|
I've recently been playing around with inpainting via multi-image prompting over generations with MJ. Creating part of your scent, then selecting a portion to change, then swapping out description and adding an image prompt. I can kinda get whatever I want out of it regardless of complexity and number of scene elements. It's great!
|
# ? Sep 19, 2023 05:49 |
|
Yeah that was a major feature and differentiator between Stable Diffusion and Midjourney, and Midjourney has got it now. I consider it a key feature. Also SDXL's inpainting seems to have significantly worsened, although that could just be the added level of complexity that the creators of SDXL unwisely decided to introduce when they attempted to move everyone from a canvas model to a workflow model. I might write this up later, but the TLDR of SDXL is that they made a significant and useful improvement to their base model but then shot it in the foot with two terrible decisions: 1) brought it out in a stripped form that had less functionality for inpainting and controlnets than their previous model, and have been scattershot adding functionality over the last 3 months, and 2) decided to push their pet project, ComfyUI, on people. It's a node-based instead of canvas based image editor and unless you're a programmer or VFX artist or want to be one it's a SIGNIFICANT barrier to entry. Megaman's Jockstrap fucked around with this message at 16:42 on Sep 19, 2023 |
# ? Sep 19, 2023 16:29 |
|
Ceramic Shot posted:I'm designing an Arkham Horror the Card Game-esque fantasy-themed game and wanted to show this off: This is really cool but based on the QR, are you selling this commercially? Because yes the ethics have been discussed to death, but if you're using MJ for art in something you're actually selling, honestly that actually is really unethical.
|
# ? Sep 19, 2023 22:33 |
|
Fuzz posted:This is really cool but based on the QR, are you selling this commercially? Because yes the ethics have been discussed to death, but if you're using MJ for art in something you're actually selling, honestly that actually is really unethical. I think the ruleset I'm using is too similar to Arkham's to sell on that basis alone. I'm not prickly at all when it comes to discussing the ethics of hypothetically making a salable product with MJ though. In the "In the World" channel in the MJ Discord there are many examples of people who've already made traditional games products with it. Could you explain your position? I guess it has to do with the scraped images from the original artists?
|
# ? Sep 20, 2023 00:50 |
|
Why does selling matter? If I give it away or charge $20 for it, I’m still using whatever copyrighted content is deemed to have had its rights of ownership survive the training process. I can benefit from giving it away, and giving it away can plausibly substitute for the art of an artist whose works were in the training set. What if I ask for charitable donations? I think it’s fascinating to think about the edges of copyright law and how much the model is deemed a derivative of the training works, but “non-commercial use” is a very vague and fluid thing to make an underpinning of a moral stance.
|
# ? Sep 20, 2023 01:01 |
|
Leaving the ethics of the thing aside - because there's other places to discuss that, and this thread is a No Fighting (About AI Ethics) In The War Room kind of space - it's worth keeping an eye on the legal proceedings surrounding AI-generated material. At the moment, you're best off never mentioning that you're using generative models to make something you intend to sell, or you risk your copyright.Megaman's Jockstrap posted:Yeah that was a major feature and differentiator between Stable Diffusion and Midjourney, and Midjourney has got it now. I consider it a key feature. ComfyUI drives me crazy. I can generate things on it using other people's workflows, no problem, but it's some real cargo cult poo poo because the exposed guts of the thing are incomprehensible. Related, I simply cannot wrap my head around prompting SDXL ComfyUI: it behaves very, very differently from A1111 SD 1.5, and seems to do well if you give it broad concepts, but it hates specificity with a burning passion. Hopefully by the end of the year SDXL will be in the kind of shape where it will start to replace SD 1.5 for me, but boy is it taking its time.
|
# ? Sep 20, 2023 03:29 |
|
Going all in on making my homebrew look like an official ttrpg-rulebook. Here's a chapter about the available origins that the players can chose from. It's one of 11 and on top of that, there's four classes with 3 subclasses each.. so this might take me a while to finish. Also, sorry for the non-english language. edit I also have a growing concern on how this might turn my friends into furries and that I will have to find new ones after the first session. Imaginary Friend fucked around with this message at 17:12 on Oct 12, 2023 |
# ? Oct 12, 2023 17:09 |
|
The art is neat.
|
# ? Oct 12, 2023 17:25 |
|
Raenir Salazar posted:The art is neat. It pretty cool to be able to create a more ambitious project that would require a whole team before. While there is alot of value in lots of ideas coming together, being able to create one very focused project with the help of AI is an intriguing thought.
|
# ? Oct 12, 2023 18:07 |
|
Yeah there was this one account that was generating fantasy anime art and they captured a very interesting style with what I assume was some kind of custom Stable Diffusion thingy; where they had maintained a sort of hand drawn graphic novelesque shading and line art that maintained the right degree of "sketchy" (as in sketching) feeling to it. Wish I knew how they did it because it'd be something I'd consider using for making tokens.
|
# ? Oct 12, 2023 19:24 |
|
I am encouraging my girl friend to run our Sunday game for out group while I am in the middle of the semester from hell. I'm taking 16 hours of graduate school classes. Anyway she apparently used chatGPT to plan the whole adventure Sunday and while it wasn't the best, I am proud of her for even attempting to run a game. The theme of the game was being pirates in the Caribbean. It really helped her as a newbie GM. I always encourage plagiarism when running home games.
|
# ? Oct 17, 2023 08:24 |
|
Zoeb posted:I always encourage plagiarism when running home games. Yeah, I'm running something with people using mushrooms for power, and one of my players of course namedropped The Last of Us. "If your going to steal, steal from the best" or, Wilson Mizner (screenwriter) said, “If you steal from one author, it’s plagiarism, and if you steal from many, it’s research.”
|
# ? Oct 17, 2023 08:58 |
|
Zoeb posted:I always encourage plagiarism when running home games. This feels like why I enjoy using AI for tabletop. I don't have any problems plagiarizing for my games the old fashioned way, why should I object to using better technology to do it?
|
# ? Oct 28, 2023 14:42 |
|
I definitely think its interesting the ways the short comings of AI pop up very often in the context of running D&D games, it really makes it clear the importance of human curation for providing a quality experience to your players.
|
# ? Oct 28, 2023 16:58 |
|
AI GMs aren't going to be viable anytime in the short-term. That being said, I have had great success in using AI as a collaborative storytelling partner in a slightly less traditional RPG format. I have a little Jurassic Park '93 fan TTRPG I've been toying with for a while, which is mostly worldbuilding rather than game mechanics. I also have a list of random event cards in there, and general rules around those cards such as what should happen when a dinosaur attack occurs and such. I've had a lot of fun popping it in claude.ai (since it has the largest context window) and bouncing the narrative back and forth. Once "we" establish a starting location and a group of characters, I give it a handful of details such as character personality, give it a few vague bullet points for what the scene should be, then tell it to dramatize the action, then tell it to draw a random event card. It does a pretty darn good job of following what I give it, throwing in lots of fun details, then taking the story into an unexpected direction. Then I tell it how the characters react to the scenario, tell it to dramatize it, and have it draw another random event card. It's very fun! It's a little more like I'm the GM and it's the player than the other way around, but for me it's the same effect in the end. As long as I decide the majority of events, the limitations of AI don't really rear their heads much. I still occasionally run up against limitations, of course. It definitely doesn't "draw" a card at random, it instead picks an event that it thinks lines up with where the story has been going. To combat this, I made a random event roller and throw in those events here and there. It also gets a little confused about geography and such here and there, but that tends to be pretty easy to course-correct. I also had to remove a lot of the mechanics I had defined since it tends to ignore them. But I'm not a very crunchy player so I don't really mind so much. Plus, I wrote the game, so I can always corral it more in the direction I prefer. At the end of the day it's more like a fan fiction generator than an RPG session, but I'm having fun with that for now. The biggest limitation isn't in the AI itself, but that claude.ai doesn't allow you to go back and edit things like ChatGPT does. If you give it bad instructions, you're stuck with them and have to do a "ignore the last prompt and your last response..." which is annoying. e: This kinda stuff: feedmyleg fucked around with this message at 17:27 on Oct 28, 2023 |
# ? Oct 28, 2023 17:14 |
|
People go straight to "replacing the human" but AI right now works much better as "enhance the human". Text-based AI, for example, can be the next level of spell check. Instead of hard and brittle rules for spelling/grammar, AIs at the capability level of GPT-4 can check for semantic consistency or suggest ways to make rules explanations more succinct.
|
# ? Oct 28, 2023 19:09 |
|
A little short notice, but can anyone have a go at generating cartoony/graphic novel/stylized art of Pinky and the Brain but as Mind flayers?
|
# ? Nov 18, 2023 03:02 |
|
Raenir Salazar posted:A little short notice, but can anyone have a go at generating cartoony/graphic novel/stylized art of Pinky and the Brain but as Mind flayers? I can't get DALLE 3 to make anything that looks good for Pinky, but here's the best minimal-prompt-finagling Brain I got: "the characters Pinky and the Brain except as mindflayers in a graphic novel style"
|
# ? Nov 18, 2023 03:32 |
|
HOMOEROTIC JESUS posted:I can't get DALLE 3 to make anything that looks good for Pinky, but here's the best minimal-prompt-finagling Brain I got: Thanks for the effort, but sadly that's not quite what I was hoping for. Mind Flayers in D&D look more like this: Where basically I want one that has a bigger head (like Brain) and one that's more tall and lanky like Pinky.
|
# ? Nov 18, 2023 17:34 |
|
Raenir Salazar posted:Thanks for the effort, but sadly that's not quite what I was hoping for. That's too bad. "Elder Brain" was the best looking of the bunch. None of the traditional Ilithid style Pinky/Brains looked any good, sadly. If you want to give it a stab yourself, the Bing AI (Dalle 3) is free. I didn't put much more than 5 minutes into getting that image. You might do better with some more trying and playing around with keywords.
|
# ? Nov 18, 2023 17:51 |
|
|
# ? May 2, 2024 05:29 |
|
I gave it a try as well. Illithids are hard to get right, it seems. That said, I'd say the image you posted of the, lol, gnome illithid already works for Brain. Just outline select his head in an editor and make it even bigger if that's not big enough for ya. And for Pinky, the big noggin illithid already provides enough contrast for both of them, so grab a regular goofy-looking illithid off Google Image Search. EDIT: kekeke or Megazver fucked around with this message at 21:12 on Nov 18, 2023 |
# ? Nov 18, 2023 20:59 |