Ultra-High-Resolution Image Synthesis with Pyramid Diffusion Model
This work provides an incremental reinforcement for existing image generative frameworks, addressing the problem of scalable high-resolution image generation for researchers and practitioners in computer vision.
The paper tackles ultra-high-resolution image synthesis by introducing the Pyramid Diffusion Model (PDM), which achieves the synthesis of 2K resolution images for the first time, as demonstrated on new datasets with sizes up to 2048x2048 pixels.
We introduce the Pyramid Diffusion Model (PDM), a novel architecture designed for ultra-high-resolution image synthesis. PDM utilizes a pyramid latent representation, providing a broader design space that enables more flexible, structured, and efficient perceptual compression which enable AutoEncoder and Network of Diffusion to equip branches and deeper layers. To enhance PDM's capabilities for generative tasks, we propose the integration of Spatial-Channel Attention and Res-Skip Connection, along with the utilization of Spectral Norm and Decreasing Dropout Strategy for the Diffusion Network and AutoEncoder. In summary, PDM achieves the synthesis of images with a 2K resolution for the first time, demonstrated on two new datasets comprising images of sizes 2048x2048 pixels and 2048x1024 pixels respectively. We believe that this work offers an alternative approach to designing scalable image generative models, while also providing incremental reinforcement for existing frameworks.