Decoupled Diffusion Transformers: Accelerating High-Fidelity Image Generation via Semantic-Detail Separation and Encoder Sharing

Decoupled Diffusion Transformers: Accelerating High-Fidelity Image Generation via Semantic-Detail Separation and Encoder Sharing

Diffusion Transformers have demonstrated outstanding performance in image generation tasks, surpassing traditional models, including GANs and autoregressive architectures. They operate by gradually adding noise to images during a forward diffusion process and then learning to reverse this process through denoising, which helps the model approximate the underlying data distribution. Unlike the commonly used UNet-based diffusion models, Diffusion …