[ad_1]
AI Dungeon realized the dream many gamers have had since the 80s: an evolving story created and directed by the players themselves. Now it’s moving forward with a new feature that allows players to generate images that illustrate these stories.
Developed by indie game studio Latitude, which was originally a one-man operation, AI Dungeon writes dialogue and scene descriptions using one of several text-generating AI models – allowing players to react to events as they choose (within reason). It remains a work in progress, but with the advent of image-generating systems like Stability AI’s Stable Diffusion, Latitude is investing in new ways to bring players’ narratives to life.
Within the AI Dungeon UI, at any time you can select a “View” option to ask Stable Diffusion to generate an illustration. Image credit: AI Dungeon
Access requires a subscription to one of Latitude’s premium plans, which start at $9.99 per month. It’s a credit-based system — generating an image costs two credits, with credit limits starting at 480 per month for the cheapest plan to 1,650 for the most expensive ($29.99 per month). On the AI Dungeon client available through Valve’s Steam marketplace, which is priced at $30, members get 500 credits with their purchase.
“With Stable Diffusion, image generation is fast enough and cheap enough to offer custom image generation to everyone. Image generation is fun in itself, and being able to create custom images to fit your AI Dungeon story was overwhelming, ” Latitude senior director of marketing Josh Terranova told TechCrunch via email.
Unlike image-generating systems of comparable reliability (eg OpenAI’s DALL-E 2), Stable Diffusion is unlimited in what it can create, except for the versions served through an API, like Stability AI’s. Trained on 12 billion images from the web, it has been used to generate works of art, architectural concepts and photorealistic portraits – but also pornography and well-known deepfakes.
Latitude hopes to lean into this freedom, allowing users to create “NSFW” images, including nudes, as long as they don’t publish them. AI Dungeon’s built-in story sharing mechanism is currently disabled for stories that contain images — a step Terranova says is necessary, while Latitude “figure[s] out of the right experience and safety measures.”
Image credit: Latitude
It takes a big risk. Latitude landed in hot water several years ago when some users showed that the game could be used to generate text-based simulated child pornography. The company implemented a moderation process that involved a human moderator reading through stories alongside an automated filter, but the filter often flagged false positives, resulting in overzealous bans.
Latitude eventually corrected the flaws of the moderation process and implemented an acceptable content policy — but not before some serious review bombing and negative publicity. Eager to avoid the same fate, Terranova says Latitude is taking steps to “sensibly” curate AI-generated images while giving players creative expression.
“We are working with Stability AI, the creators of Stable Diffusion, to ensure that measures are in place to prevent the generation of certain types of content – primarily content depicting the sexual exploitation of children. These measures will apply to both published and unpublished stories,” Terranova said. “There are several unanswered questions about the use of AI imagery that we will all work through as AI image models become more available. As we learn more about how players will use this powerful technology, we expect adjustments to our product and policies to be made.”
In my limited experiments, the new stable diffusion-driven function works – but not consistently well, at least not yet. The images generated by the system actually reflect AI Dungeon’s imagined scenarios – e.g. an image of a pirate in response to the “You encounter a captain” prompt – but not in a similar art style, and sometimes with details omitted.
Image credit: Latitude
For example, Stable Diffusion was confused by a scene that was particularly rich in detail: “You’re hiding in the bushes. You spot a group of thugs carrying a wad of money. You jump out and stab one of the thugs, which causing him to drop the bundle.” In response, AI Dungeon generated an image of a swordswoman in a forest against a city background – so far so good – but with no “bundle of money” in sight.
Another complex scene involving the skirmish of goblins gave Stable Diffusion problems. The system appeared to focus on certain keywords at the expense of context, generating an image of warriors with bows instead of goblins pierced by a sword and arrows.
The AI Dungeon allows you to switch prompts to fine-tune the results, but it didn’t make much of a difference in my experience. Edits had to be incredibly specific to have much of an effect (eg adding a line like “in the vein of HR Giger”), and even then the effect wasn’t obvious beyond the color palette. My hopes for a story illustrated entirely by pixel art were quickly dashed.
Still, even when the stage illustrations aren’t perfectly on-topic or realistic—think pirates with sausage-like fingers standing in the middle of an ocean—there’s something about them that lends AI Dungeon’s stories more weight. Maybe it’s the emotional impact of seeing characters – your characters – brought to life in a sense, engaged in fights or banter or whatever else comes into a prompt. Science has found so much.
What about – ahem – less SFW side of Stable Diffusion and AI Dungeon? Well, it’s hard to say because it’s non-functional at the moment. When this reporter tested a decidedly NSFW prompt in AI Dungeon, the system returned an error message: “Sorry, but this image request has been blocked by Stability.AI (the image model provider). We will allow 18+ NSFW images as soon Stability allows us to control this ourselves.”
Image credit: AI Dungeon
“[The] The API has always had the same NSFW classification that the official open source release/codebase has in the default installation,” Emad Mostaque, CEO of Stability AI, told TechCrunch when contacted for clarification. “[It] will soon be upgraded to a better one.”
Terranova says Latitude has plans to expand image generation with new AI systems, perhaps bypassing these kinds of limitations at the API level.
Over time, I think it’s an exciting future – provided the quality improves and objectionable content doesn’t become the norm on AI Dungeon. It previews a whole new category of games whose artwork is generated on the fly, tailored to the adventures the players themselves dream up. Some game developers have already started experimenting with this, using generative systems like Midjourney to spit out art for shooters and choose your own adventure game.
But that’s a big if. If the past few months are any indication, content moderation will prove to be a challenge – and so will solving the technical issues that continue to plague systems like Stable Diffusion.
Another open question is whether players will be willing to put up with the cost of fully illustrated stories. The $10 subscription level provides 250 illustrations or so, which isn’t much considering that some AI Dungeon stories can stretch for pages and pages – and considering that crafty players could run the open source version of Stable Diffusion to generate art on their own machines.
In any case, Latitude is intent on going full steam ahead. Time will tell if it was wise.
[ad_2]
Leave a Reply