BADiff: Bandwidth Adaptive Diffusion Model
This work addresses the practical issue of efficient image delivery in bandwidth-constrained environments, offering a domain-specific solution for cloud-to-device applications.
The paper tackles the problem of diffusion models generating high-fidelity images without considering network bandwidth constraints, which leads to wasted computation and quality loss in cloud-to-device scenarios. It proposes a framework that adapts generation quality based on real-time bandwidth, achieving significant improvements in visual fidelity compared to naive early-stopping methods.
In this work, we propose a novel framework to enable diffusion models to adapt their generation quality based on real-time network bandwidth constraints. Traditional diffusion models produce high-fidelity images by performing a fixed number of denoising steps, regardless of downstream transmission limitations. However, in practical cloud-to-device scenarios, limited bandwidth often necessitates heavy compression, leading to loss of fine textures and wasted computation. To address this, we introduce a joint end-to-end training strategy where the diffusion model is conditioned on a target quality level derived from the available bandwidth. During training, the model learns to adaptively modulate the denoising process, enabling early-stop sampling that maintains perceptual quality appropriate to the target transmission condition. Our method requires minimal architectural changes and leverages a lightweight quality embedding to guide the denoising trajectory. Experimental results demonstrate that our approach significantly improves the visual fidelity of bandwidth-adapted generations compared to naive early-stopping, offering a promising solution for efficient image delivery in bandwidth-constrained environments. Code is available at: https://github.com/xzhang9308/BADiff.