
The training set comes from the artist DISH (@Dishwasher1910). This time, due to a commission, I also trained a version of Animagine XL V3.1 and included xy comparison charts. Indeed, the version trained with Animagine XL V3.1 has better limb articulation compared to my own model, JRU2. However, in terms of style and color tones, training with JRU2 is closer to the original style of the artist.
Additionally, if the lycoris trained on Animagine XL V3.1 is used on the JRU2 model, or vice versa, it will retain some features of the original model and create a unique style, which is quite interesting.
训练集来自画师DISH (@Dishwasher1910)这次的风格因为受人委托,所以我也训练了一份Animagine XL V3.1版本的,并且附上了xy对比图,使用Animagine XL V3.1训练出来的版本确实肢体要比我自己的模型,即JRU2要好,但是风格和色调上,使用JRU训练会更接近艺术家原本的风格,
另外如果吧在Animagine XL V3.1上训练的lycoris用在JRU2这个模型上,或者相反,会因为保留了一原本模型本身的一些特征而形成一个特别的风格,很有意思,
描述:
训练词语:
名称: dishXLlokr3fanimagine.safetensors
大小 (KB): 1094074
类型: Model
Pickle 扫描结果: Success
Pickle 扫描信息: No Pickle imports
病毒扫描结果: Success