该LoRA模型主要是模拟轴测图画法,将室内场景进行轴测2.5d模型风格化,此LoRA最适配一些2.5d风格的模型,当然在其他像2d风格和3d写实风格也能使用(可能需要适当提高模型权重),但是如果提高权重后LoRA仍然无法生效,建议还是换一个基础大模型吧。后期我还会继续尝试优化训练过程以及训练图集的质量,制作更高质量和泛化性的LoRA模型。
The LoRA model primarily simulates the axonometric drawing technique, stylizing indoor scenes into a 2.5D model. This LoRA model is most suitable for creating models in a 2.5D style, but it can also be used for other styles like 2D and realistic 3D (although it may require adjusting the model weights accordingly). However, if increasing the weights still doesn't yield satisfactory results, it is recommended to switch to a different base model. In the future, I will continue to work on optimizing the training process and improving the quality of the training dataset to create a higher quality and more generalized LoRA model.
Base models:revAnimated_v11、chilloutmix_NiPrunedFp32Fix、deliberate_v2、OldFish_v11、anything-v4.5-pruned...