A LoRA trained on 400+ images by the artist Geonjeonji. No trigger word needed. All sample images were generated with strength 1.

This was my first time training a pony LoRA and it honestly turned out a lot better than I expected. It has a tendency to produce black and white images so i would suggest putting "black and white, monochrome" in the negative prompt. I'd also recommend a lower ADetailer face denoising strength to preserve the style in the face. The sample images use 0.2. I might retrain this at a later date with a refined dataset but for now I want to try some other things.