Search papers, labs, and topics across Lattice.
The paper addresses the "prior collapse" problem in one-step real-world image super-resolution (Real-SR) when fine-tuning generative models on limited LR-HR pairs. They introduce Allo{SR}$^2$, a framework that uses allomorphic generative flows to rectify one-step SR trajectories, maintaining generative realism. Allo{SR}$^2$ incorporates SNR-guided trajectory initialization, flow-anchored trajectory consistency (FATC), and allomorphic trajectory matching (ATM) to achieve state-of-the-art performance in one-step Real-SR.
One-step image super-resolution can now achieve state-of-the-art realism and fidelity thanks to a novel framework that guides generative flows to avoid trajectory drift and prior collapse.
Real-world image super-resolution (Real-SR) has been revolutionized by leveraging the powerful generative priors of large-scale diffusion and flow-based models. However, fine-tuning these models on limited LR-HR pairs often precipitates"prior collapse"that the model sacrifices its inherent generative richness to overfit specific training degradations. This issue is further exacerbated in one-step generation, where the absence of multi-step refinement leads to significant trajectory drift and artifact generation. In this paper, we propose Allo{SR}$^2$, a novel framework that rectifies one-step SR trajectories via allomorphic generative flows to maintain high-fidelity generative realism. Specifically, we utilize Signal-to-Noise Ratio (SNR) Guided Trajectory Initialization to establish a physically grounded starting state by aligning the degradation level of LR latent features with the optimal anchoring timestep of the pre-trained flow. To ensure a stable, curvature-free path for one-step inference, we propose Flow-Anchored Trajectory Consistency (FATC), which enforces velocity-level supervision across intermediate states. Furthermore, we develop Allomorphic Trajectory Matching (ATM), a self-adversarial alignment strategy that minimizes the distributional discrepancy between the SR flow and the generative flow in a unified vector field. Extensive experiments on both synthetic and real-world benchmarks demonstrate that Allo{SR}$^2$ achieves state-of-the-art performance in one-step Real-SR, offering a superior balance between restoration fidelity and generative realism while maintaining extreme efficiency.
