AddSR: Accelerating Diffusion-based Blind Super-Resolution with Adversarial Diffusion Distillation

TL;DR

AddSR significantly accelerates diffusion-based blind super-resolution by integrating adversarial diffusion distillation and ControlNet, resulting in faster processing and improved image restoration quality.

Contribution

The paper introduces AddSR, a novel method combining distillation and ControlNet to enhance efficiency and robustness in diffusion-based blind super-resolution.

Findings

Achieves 7x faster speed than previous state-of-the-art models.

Produces better image restoration results.

Demonstrates robustness with HR-based training constraints.

Abstract

Blind super-resolution methods based on stable diffusion showcase formidable generative capabilities in reconstructing clear high-resolution images with intricate details from low-resolution inputs. However, their practical applicability is often hampered by poor efficiency, stemming from the requirement of thousands or hundreds of sampling steps. Inspired by the efficient adversarial diffusion distillation (ADD), we design~\name~to address this issue by incorporating the ideas of both distillation and ControlNet. Specifically, we first propose a prediction-based self-refinement strategy to provide high-frequency information in the student model output with marginal additional time cost. Furthermore, we refine the training process by employing HR images, rather than LR images, to regulate the teacher model, providing a more robust constraint for distillation. Second, we introduce a timestep-adaptive ADD to address the perception-distortion imbalance problem introduced by original ADD. Extensive experiments demonstrate our~\name~generates better restoration results, while achieving faster speed than previous SD-based state-of-the-art models (e.g., $7$$\times$ faster than SeeSR).