FoA-SR: Faithful or Aesthetic? Profile-Aware Preference Optimization for Real-World Image Super-Resolution
Quick Answer
FoA-SR introduces a profile-aware optimization approach for real-world image super-resolution, distinguishing between Faithful and Aesthetic restoration objectives.
Quick Take
FoA-SR introduces a profile-aware optimization approach for real-world image super-resolution, distinguishing between Faithful and Aesthetic restoration objectives. Experiments demonstrate that the Faithful adapter enhances reference consistency, while the Aesthetic adapter improves perceptual quality metrics, showcasing the effectiveness of tailored restoration strategies.
Key Points
- FoA-SR utilizes a supervised FLUX.2-based SR adapter for enhanced image restoration.
- The model generates a stochastic candidate pool and ranks them based on profile-specific rewards.
- Faithful and Aesthetic rewards often select different optimal candidates for restoration.
- Hybrid-LoRA ablation shows that merging profiles leads to implicit compromises in restoration quality.
- Experiments on RealSR and DIV2K validate the distinct improvements in restoration metrics.
Paper Resources
Article Content
From source RSS / original summaryarXiv:2606. 10275v1 Announce Type: new Abstract: Real-world image super-resolution (SR) is often designed with a single restoration objective, despite the current capacity of generative models to produce multiple high-quality reconstructions for the same input.
In this paper, we argue that the best restoration strategy is subject to the specific restoration profile: a Faithful restoration prioritizes reference consistency, structure preservation, and hallucination suppression, whereas an Aesthetic restoration prioritizes visually pleasing and natural-looking details. We propose FoA-SR, a novel preference optimization approach to real-world SR based on profiles. To achieve this goal, FoA-SR starts with our supervised FLUX.
2-based SR adapter (Flux2SR) trained with LR latent conditioning, flow matching, and image-space reconstruction losses for paired LR-to-HR image super-resolution. Following the development of the shared supervised super-resolution adapter, FoA-SR generates a shared stochastic candidate pool for each input image and ranks the same candidates using profile-specific Faithful and Aesthetic rewards to mine winner-loser pairs. These pairs are used to fine-tune separate LoRA adapters while keeping the base model frozen.
Experiments on RealSR and DIV2K show that FoA-SR can steer the same SR adapter towards distinct restoration objectives: a Faithful adapter improves reference-consistent metrics while an Aesthetic adapter boosts metrics that measure perceptual quality without reference.
Our candidate-pool analysis shows that Faithful and Aesthetic rewards frequently select different winners, and a Hybrid-LoRA ablation shows that collapsing both profiles into one reward yields an implicit compromise rather than explicit profile control.
Reader Mode unavailable (could not extract clean content).
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →LLM-Guided ANN Index Optimization for Human-Object Interaction Retrieval
A phase-aware LLM agent optimizes human-object interaction retrieval, outperforming Optuna TPE by 33.3% and VDTuner by 34.2% on the HICO-DET benchmark. This method enhances throughput by 15.3x over UniIR and demonstrates strong transferability across vector database management systems.