InfinityGAN: Towards Infinite-Pixel Image Synthesis
arXiv:2104.03963
Abstract
We present a novel framework, InfinityGAN, for arbitrary-sized image generation. The task is associated with several key challenges. First, scaling existing models to an arbitrarily large image size is resource-constrained, in terms of both computation and availability of large-field-of-view training data. InfinityGAN trains and infers in a seamless patch-by-patch manner with low computational resources. Second, large images should be locally and globally consistent, avoid repetitive patterns, and look realistic. To address these, InfinityGAN disentangles global appearances, local structures, and textures. With this formulation, we can generate images with spatial size and level of details not attainable before. Experimental evaluation validates that InfinityGAN generates images with superior realism compared to baselines and features parallelizable inference. Finally, we show several applications unlocked by our approach, such as spatial style fusion, multi-modal outpainting, and image inbetweening. All applications can be operated with arbitrary input and output sizes. Please find the full version of the paper at https://openreview.net/forum?id=ufGMqIM0a4b .
arXiv commentAccepted to ICLR 2022. Full Paper: https://openreview.net/forum?id=ufGMqIM0a4b ; Project page: https://hubert0527.github.io/infinityGAN/
References in corpus (9)
- GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium
- Conditional Image Synthesis With Auxiliary Classifier GANs
- Which Training Methods for GANs do actually Converge?
- A Good Image Generator Is What You Need for High-Resolution Video Synthesis
- Painting Outside the Box: Image Outpainting with GANs
- Improving Inversion and Generation Diversity in StyleGAN using a Gaussianized Latent Space
- Understanding Overparameterization in Generative Adversarial Networks
- In&Out : Diverse Image Outpainting via GAN Inversion
- Infinite Nature: Perpetual View Generation of Natural Scenes from a Single Image