
This is the first Lora i ever trained. Its trained on Pixai.art where registered people can train Lora on their own without borrowing an high end PC from someone. Because i was not happy that the typical Model couldnt handle Fairy Wings well, i tried to train this lora to get constantly good Fairy Wings. I tested it on different weights but it seem 0.5 - 0.9 works the best without damaging your model type too much. As i mostly use Anime Models i only trained it for Anime Models, but you can try realistic, i wont make any promises for it.
Mark 2 of the Fairy Wing Lora was trained on around 50 Pictures of High Quality and Fidelity, the outcome of the Wings themselves is more stable and more color correct. You get even more Wings then less wings. It seems to work with more different Models too now, but i cant promise anything.
描述:
Trained on Anything V5, seems to have problem at higher weights. Use Weight ~0.5 - 0.8 to get good overall results.
训练词语: Wings, Fairy Wings, Big Fairy Wings, Fairy
名称: Fairy Wing Lora.safetensors
大小 (KB): 38163
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success