Recently, there has been a surge in face personalization techniques,
benefiting from the advanced capabilities of pretrained text-to-image diffusion
models. Among these, a notable method is Textual Inversion, which generates
personalized images by inverting given images into textual embeddings. However,
methods based on Textual Inversion still struggle with balancing the trade-off
between reconstruction quality and editability. In this study, we examine this
issue through the lens of initialization. Upon closely examining traditional
initialization methods, we identified a significant disparity between the
initial and learned embeddings in terms of both scale and orientation. The
scale of the learned embedding can be up to 100 times greater than that of the
initial embedding. Such a significant change in the embedding could increase
the risk of overfitting, thereby compromising the editability. Driven by this
observation, we introduce a novel initialization method, termed Cross
Initialization, that significantly narrows the gap between the initial and
learned embeddings. This method not only improves both reconstruction and
editability but also reduces the optimization steps from 5000 to 320.
Furthermore, we apply a regularization term to keep the learned embedding close
to the initial embedding. We show that when combined with Cross Initialization,
this regularization term can effectively improve editability. We provide
comprehensive empirical evidence to demonstrate the superior performance of our
method compared to the baseline methods. Notably, in our experiments, Cross
Initialization is the only method that successfully edits an individual's
facial expression. Additionally, a fast version of our method allows for
capturing an input image in roughly 26 seconds, while surpassing the baseline
methods in terms of both reconstruction and editability. Code will be made
publicly available.

通过初始化方法的研究，我们提出了一种称为 Cross Initialization 的新型初始化方法，它明显缩小了初始嵌入和学习嵌入之间的差距，通过与正则化项相结合，有效提高了可编辑性，并在编辑人脸表情方面取得了成功，同时实现了较快的图像捕获速度和较好的重建和可编辑性。