I want to investigate why a certain aspect is working well in the Stable Diffusion 1.5 model.
According to the original paper, this method is not supposed to be version-specific.
However, why does it perform poorly in version 2.0?
Can anyone provide me with some direction?
I want to investigate why a certain aspect is working well in the Stable Diffusion 1.5 model.
According to the original paper, this method is not supposed to be version-specific.
However, why does it perform poorly in version 2.0?
Can anyone provide me with some direction?