Dual-Domain Diffusion based Progressive Style Rendering Towards Semantic Structure Preservation
Mengjun Tao, Richard Jiang · 2023
In this paper, we propose a Dual-Domain Diffusion based Progressive Style Rendering (D3PSR) method to achieve style rendering from the semantic Domain A to the style Domain B. Although some existing models obtain style transfer in Image-to-Image (I2I) translation, little is known about how computers draw images due to the black box problem of classic generative model.Leveraging diffusion models for I2I style transfer, for the first time, we have implemented a progressive method to visualize the intermediate steps in image generation, which provides interpretability for style transfer.As far as we know, we are the first to innovatively use the diffusion model for dual-domain image style transfer.Numerous experimental results and comparisons with state-of-the-art methods in the same field show that our approach has extremely superior performance in stylization and extraordinary preservation of semantic structure.