This LoRA is trained on a hand-curated and manually inspected dataset focused on consistent, high-quality DnD battle map artwork. The model specializes in generating top-down battle mats.
The style is optimized specifically for battle maps, not world maps, large-scale overworlds, or abstract cartography.
Expect clean terrain, readable topology, and minimal artifacts when prompting correctly. You're free to mix styles, combine with tilesets, or push creative variations—this LoRA responds very well to tuning CFG and sampling methods.
