We present a novel approach to leverage prior knowledge encapsulated in pre-trained text-to-image diffusion models for blind super-resolution (SR). Specifically, by employing our time-aware encoder, we can achieve promising restoration results without altering the pre-trained synthesis model, thereby preserving the generative prior and minimizing training cost. To remedy the loss of fidelity caused by the inherent stochasticity of diffusion models, we introduce a controllable feature wrapping module that allows users to balance quality and fidelity by simply adjusting a scalar value during the inference process. Moreover, we develop a progressive aggregation sampling strategy to overcome the fixed-size constraints of pre-trained diffusion models, enabling adaptation to resolutions of any size. A comprehensive evaluation of our method using both synthetic and real-world benchmarks demonstrates its superiority over current state-of-the-art approaches.

翻译：我们提出了一种新颖的方法，利用预训练文本到图像扩散模型中蕴含的先验知识，实现盲超分辨率（SR）。具体而言，通过采用时间感知编码器，我们无需改动预训练合成模型即可取得显著的重建效果，这既保留了生成先验，又降低了训练成本。为弥补扩散模型固有随机性导致的保真度损失，我们引入了可控特征包裹模块，该模块允许用户在推理过程中通过简单调整标量值来平衡质量与保真度。此外，我们开发了渐进式聚合采样策略，以克服预训练扩散模型的固定尺寸限制，从而适应任意分辨率的处理。在合成和真实基准上的全面评估表明，我们的方法优于当前最先进的技术。