V 2.0 is use extension.
Dataset: 512*512
No trigger words
LoRa weight 0.3-0.7 for characters, weight > 0.7 for background/scene
2.5d and realistic checkpoint are suitable for this lora.
Please give me 5 stars and like ?, which is important to me.
?your reviews motivate me to update，your suggestion for updates are also helpful！
This lora is to generate mirror broken, glass broken, kaleidoscope.
weight suggested:0.6-0.8 (1 for larger size and realistic)
Do not use face restore，Hires.fix and negative prompt（deep fix faces).
mirror broken (face internal) is defined as "bad, worse, error" in AI painting, so that it hard to conduct desire effect of mirror broken.
?️This lora is experimental, view/comments are free to play it, if you could put your output and prompt will help me explore better solution to make mirror broken with face/body.?
?️宝贝们，这个lora是个试验性lora, 我还在探索中，因为人像在玻璃/镜面破碎的效果AI很难做到符合光学原理，目前我也是在找最优解，你们有兴趣可以放上你们的出图和参数，感谢?
Reference
描述:
trained dataset: 512*512
Epoch :10
image quantity: 33p
训练词语: mirror broken,glass broken
名称: Kaleidoscope-万花筒.safetensors
大小 (KB): 73842
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success
