An Empirical Study of Sampling Hyperparameters in Diffusion-Based Super-Resolution
This is an incremental improvement for researchers and practitioners using diffusion models for image super-resolution.
The study tackled the problem of tuning hyperparameters in diffusion-based super-resolution methods, finding that conditioning step size has a greater impact than diffusion step count, with optimal step sizes in [2.0, 3.0] yielding the best performance.
Diffusion models have shown strong potential for solving inverse problems such as single-image super-resolution, where a high-resolution image is recovered from a low-resolution observation using a pretrained unconditional prior. Conditioning methods, including Diffusion Posterior Sampling (DPS) and Manifold Constrained Gradient (MCG), can substantially improve reconstruction quality, but they introduce additional hyperparameters that require careful tuning. In this work, we conduct an empirical ablation study on FFHQ super-resolution to identify the dominant factors affecting performance when applying conditioning to pretrained diffusion models, and show that the conditioning step size has a significantly greater impact than the diffusion step count, with step sizes in the range of [2.0, 3.0] yielding the best overall performance in our experiments.