Skip to content

Dataset · 0xJustin

Images of D&D fantasy races and classes

0xJustin/Dungeons-and-Diffusion

A captioned image dataset for training D&D character generation: races and classes, roughly 30 images each.

What's inside

The dataset contains images and text captions for Dungeons & Dragons characters. In the new version, races and classes are manually captioned, with about 30 images per race or class. The underlying model was trained using EveryDream; this is a dataset, not a .ckpt model.

What you can do with it

  • Training LoRA adapters for fantasy characters
  • Fine-tuning diffusion models on D&D illustrations
  • Generating characters based on race and class
Size
1K–10K rows
Downloads this month
656
Likes
70

How to load it

from datasets import load_dataset

ds = load_dataset("0xJustin/Dungeons-and-Diffusion")