Review: LitRPG Adventures RPG Tabletop Generators and D&D Character Creator

Hacked multi-model heh…and work on dnd encounter generator

1 Like

Still tinkering on the Dnd backstory generator at LitrPG Adventures…


Quests now using 4o … just one of many fine AI RPG tools at LitRPG Adventures!


More updates to my D&D Character Creator with GPT-4o

1 Like

Free Random Event Generator for Fantasy Tabletop RPG Encounters with GPT-4o - YouTube

1 Like

Some updates to the City/Town generator… Working on making it multi-step like taverns and magic shops… ie you can add buildings/districts/people to the city… discussing it a bit on the comments at YouTube…

I really need to buckle down and just build a single WORLD/CAMPAIGN creator… I’ve got all the separate pieces now…

Updates to City and Town Generator + Request for Ideas! (Bug Appears at End) - YouTube

1 Like

Fantasy City Generator Update + Paul Bellow Face Reveal?!?! ( + explanation of locations in cities below!

I’m building out the multi-step aspect… basically a RAG-type system that takes bits and pieces of what’s there already and builds on it…

ETA: A little more on what I’ve done…

First, I generate the city/town…

Then you can generate a location within the city/town…What I tried doing here is make sure it’s always 2 paragraphs and the first is about the shop and the second about the NPC. This allows me to get everything in one LLM call… It’s 98% accurate or so with a zero-shot. I might have to give it an example to push it over the edge, but I’m using 3.5 here…

I can then just scroll the DOM to grab the paragraph I want and send a modified version of that to DALLE3 for the appropriate images…

1 Like

Another day, another update!

Fantasy City Generator Update: Stunning Visuals for Your Tabletop RPG with GPT-4o (

Basically, I’ve taken the Locations generated in the city and add the ability to add an image of the building or NPC who owns it… Last post explains the process a bit more…

1 Like

Another dev log update for unified fantasy world builder with OpenAI as the core and procedural generation helping for the win…

AI Dungeon Worldbuilder Hex Crawl Tabletop RPG Tool at LitRPG Adventures (

Each tile on the main map will open up into 7 hexagons in the pop-up… So, you can place a city on any of those seven… This is gonna be nuts for getting granular…

I’m thinking maybe 21 or 27 miles per main map tile then each of the smaller zoomed in tiles could be 7 or 9 miles respectively… or go even more granular? Hrm… Gotta keep it simple for now…


The idea is to have the ability to hit a button to have the LLM start generating more kingdom data… and “overland description” in addition to the cities, etc.

Still playing with the UI/UX as it’s going to be a lot of info… and thinking of storage too, of course. Looking promising, tho!

More progress… LLM hooked up… just need to figure out the RAG system I want to use…

AI + RPG: Creating Dynamic Hex Maps for Fantasy Worlds with GPT-4o - YouTube

every city needs a map…

1 Like

Slow and steady progress… still on maps, though the LLM is hooked up here and there…

Overland a bit better, but city maps are still a mess. I want to be able to store the building locations, dimensions, etc. to feed that to the GPT-4o to get interesting descriptions back… images coming soon too…

They’re not pretty (yet), but I’ve got rivers in that are named and tied to tiles, so I can send 'em to the LLM…

going from one hex to 7 hexes for zoomed in still a bit clunky, but progress…

1 Like

Adding resources to tile based on terrain, so I can feed that extra bit of info to the LLM…

It’s once a day and other people do it too.

Hello fellow developers,

As you might know, I’m Paul Bellow, creator of LitRPG Adventures. I’m excited to share the latest developments in my ongoing project, which I’ve been refining since April 2021.

Project Overview:

  • Continually evolving AI-powered worldbuilder for LitRPG and solo RPG experiences
  • Utilizing GPT-4o for advanced text generation and DALL-E 3 for visual content

Latest Update Focus:

Further improving consistency in LLM outputs for our hex crawl worldbuilding. Our refined approach involves:

  1. Pre-generating regions and other info based on map data
  2. Feeding this structured information to the LLM for more coherent output
  3. Ensuring all map locations are interconnected in a logical manner

Recent Challenges and Solutions:

  • LLM Consistency: Implemented a two-step process where initial region data acts as a “seed” for subsequent, more detailed generation
  • Visual-Textual Alignment: Developed more sophisticated prompts to create stronger synergy between DALL-E 3 outputs and GPT-4o descriptions

While I’ve made significant progress since my 2021 launch, I’m continuously refining my methods. I’m particularly interested in hearing from others who have experience with long-term AI integration projects.

Questions for the community:

  1. Has anyone else worked on similar long-term projects combining multiple AI models for creative applications, especially gaming?
  2. What strategies have you found effective for maintaining consistency in LLM outputs over large, interconnected datasets?
  3. How are you handling the evolution of AI models (like the transition to GPT-4o) in ongoing projects?

Looking forward to your insights and discussion!

Here’s the latest dev log at YouTube if you’re more visually inclined.

Thanks for your time, and happy coding!

Solo Hex Crawl RPG Map +: GPT-4: LitRPG Adventures’ AI Worldbuilding Leap (

ETA: Update… fleshed out the idea in the video for pre-generated regions, so the LLM doesn’t hallucinate facts about the region…