Search papers, labs, and topics across Lattice.
This paper addresses the problem of hallucination and spectral inconsistency in generative image super-resolution (SR) by proposing Adversarial Sobolev Alignment for Super-Resolution (ASASR). ASASR aligns the generative flow with the natural image manifold by coloring the noise transition kernel to match the spectral decay of natural images, effectively creating a Sobolev-induced Riemannian geometry. By using a parametric adversary grounded in the Riesz Representation Theorem to generate targeted negative samples, ASASR directs optimization along the tangent space of plausible structural failures, leading to improved spectral consistency and structural fidelity.
Hallucinated details in super-resolution are not just random noise; they reveal a fundamental spectral mismatch that can be corrected by shaping the generative process itself.
Generative priors in Image Super-Resolution (SR) often compromise faithful restoration, we attribute this limitation to a fundamental spectral misalignment between isotropic objectives and the intrinsic natural image manifold. While Direct Preference Optimization offers a path to alignment, its reliance on spectrally flat Gaussian noise fails to distinguish authentic high-frequency details from hallucinations. To bridge this geometric gap, we propose ASASR, a theoretically grounded framework that recasts the generative flow into a Sobolev-induced Riemannian geometry by explicitly coloring the noise transition kernel to mirror natural spectral decay. Driving this geometric alignment, we integrate a parametric adversary grounded in the Riesz Representation Theorem, which synthesizes targeted negative samples equivalent to worst-case Sobolev gradients to direct optimization along the tangent space of plausible structural failures. Extensive evaluations demonstrate that ASASR outperforms leading generative baselines, particularly in preserving spectral consistency and structural fidelity, offering a robust solution that effectively mitigates artifacts.
