An Empirical Study of Sampling Hyperparameters in Diffusion-Based Super-Resolution

TL;DR

This paper empirically investigates how hyperparameters, especially conditioning step size, affect the performance of diffusion models in super-resolution tasks, providing practical guidance for tuning these models.

Contribution

It presents an empirical ablation study identifying key hyperparameters influencing diffusion-based super-resolution, highlighting the importance of conditioning step size.

Findings

Conditioning step size has a greater impact than diffusion step count.

Optimal step sizes are in the range of 2.0 to 3.0.

Proper tuning of hyperparameters significantly improves reconstruction quality.

Abstract

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.