First SDXL style LoRA version trained on 20 user-captured screenshots.

Focus:

- top-down / isometric dark fantasy pixel-art gameplay

- moody low-key lighting

- gothic black-red UI frames

- glowing spell effects

- skeletal enemy crowds

- dark forest / ruin / desert combat scenes

Dataset includes both gameplay and UI screenshots, so the model can reproduce scene mood, combat presentation, and interface framing. Best results come from combining the trigger word with dark fantasy, top-down, isometric, pixel art, moody lighting, and spell effect prompts.