Style: Celscape

作者：NerissG

Slightly painterly style anime character with 3/4 photo background. No trigger word needed. You can get fully photo background by prompting it. Or get stronger painterly style by turning down the strength a little and prompting it. This is not the most inspired LoRA out there, but it helps make the model very very consistent in quality for this style.

Trained on a few of my own Anima gen with something like 10 LoRAs stacked and decoded with PiD, so it has a few rough edges. Used an experimental training technique to collapse style really strongly even though they are close to base distribution, so the style should feel very sticky.