Towards Ultra-Resolution Neural Style Transfer via Thumbnail Instance Normalization

We present an extremely simple Ultra-Resolution Style Transfer framework, termed URST, to flexibly process arbitrary high-resolution images (e.g., 10000x10000 pixels) style transfer for the first time. Most of existing state-of-the-art methods would fall short due massive memory cost and small stroke size when processing ultra-high resolution images. URST completely avoids problem caused by (1) dividing image into patches (2) performing patch-wise with a novel Thumbnail Instance Normalization (TIN). Specifically, TIN can extract thumbnail features' normalization statistics apply them patches, ensuring consistency among different patches. Overall, framework has three merits compared prior arts. divide input adopt TIN, successfully transferring high-resolution. Experiments show that our surpasses SOTA on benefiting from effectiveness proposed perceptual loss in enlarging size. (3) Our be easily plugged most directly improve their performance even without training. Code is available at https://git.io/URST.