Abstract
Diffusion models have become a mainstream approach for high-resolution image synthesis. However, directly generating higher-resolution images from pretrained diffusion models will encounter unreasonable object duplication and exponentially increase the generation time. In this paper, we discover that object duplication arises from feature duplication in the deep blocks of the U-Net. Concurrently, We pinpoint the extended generation times to self-attention redundancy in U-Net’s top blocks. To address these issues, we propose a tuning-free higher-resolution framework named HiDiffusion. Specifically, HiDiffusion contains Resolution-Aware U-Net~(RAU-Net) that dynamically adjusts the feature map size to resolve object duplication and engages Modified Shifted Window Multi-head Self-Attention(MSW-MSA) that utilizes optimized window attention to reduce computations. we can integrate HiDiffusion into various pretrained diffusion models to scale image generation resolutions even to 4096×4096 at 1.5-6× the inference speed of previous methods. Extensive experiments demonstrate that our approach can address object duplication and heavy computation issues, achieving state-of-the-art performance on higher-resolution image synthesis tasks.
Paper: arxiv.org/abs/2311.17528
Code: github.com/megvii-model/HiDiffusion
Colab Demo: …google.com/…/1EiBn9lSnPZTU4cikRRaBBexs429M-qty?u…
Project Page: hidiffusion.github.io
wewbull@feddit.uk 7 months ago
Does this mean, that because you’re now liberated from the dimensions of the training data, that all training data will apply to all sizes? e.g. generated portrait images will be influenced by landscape training data.