Positional Encoding as Spatial Inductive Bias in GANs
This work addresses the problem of understanding and improving spatial inductive bias in convolutional GANs for researchers and practitioners working on image generation.
The paper investigates how translation-invariant convolutional GANs like SinGAN and StyleGAN2 capture global structure from i.i.d. input, revealing that implicit positional encoding from zero padding is crucial for high-fidelity image generation. They propose a new multi-scale training strategy based on explicit and more flexible positional encoding, which improves StyleGAN2 and enhances SinGAN for image manipulation.
SinGAN shows impressive capability in learning internal patch distribution despite its limited effective receptive field. We are interested in knowing how such a translation-invariant convolutional generator could capture the global structure with just a spatially i.i.d. input. In this work, taking SinGAN and StyleGAN2 as examples, we show that such capability, to a large extent, is brought by the implicit positional encoding when using zero padding in the generators. Such positional encoding is indispensable for generating images with high fidelity. The same phenomenon is observed in other generative architectures such as DCGAN and PGGAN. We further show that zero padding leads to an unbalanced spatial bias with a vague relation between locations. To offer a better spatial inductive bias, we investigate alternative positional encodings and analyze their effects. Based on a more flexible positional encoding explicitly, we propose a new multi-scale training strategy and demonstrate its effectiveness in the state-of-the-art unconditional generator StyleGAN2. Besides, the explicit spatial inductive bias substantially improve SinGAN for more versatile image manipulation.