No trigger word is needed, but adding "cute" to the object helps.1.0 strength is fine. It can be decreased for more freedom if necessary.For the lora training, no copyright-protected images were used, yet the model turned out very very well.The style of the lora is high-definition 3D rendering with an animated flair. The characters depicted feature exaggerated proportions and traits typical of animated films, such as large expressive eyes and broad smiles. The skin looks smooth and is lit in a way that mimics real-life lighting, with nuanced shading and defined highlights that give a sense of three-dimensionality.The colors are vibrant and well-saturated, with a clear use of contrast in both hues and shadows to accentuate details.It works well on SDXL base model and also without Refiner.

Files with SHA256: 5a6c859a99d6e10b8744edcf6e637ae72972c5770cd0c4e6b76de80f604208a7