Search papers, labs, and topics across Lattice.
Face2Scene is a two-stage image restoration framework that uses high-fidelity face restoration as an oracle to guide full-scene restoration. First, a reference-based face restoration model reconstructs facial details, and a face-derived degradation code is extracted from the restored-degraded face pair. This degradation code is then transformed into multi-scale degradation-aware tokens to condition a diffusion model for single-step full scene restoration.
By using restored faces to estimate scene-wide degradation, Face2Scene unlocks more realistic and artifact-free image restoration compared to methods that ignore degradation cues or focus solely on faces.
Recent advances in image restoration have enabled high-fidelity recovery of faces from degraded inputs using reference-based face restoration models (Ref-FR). However, such methods focus solely on facial regions, neglecting degradation across the full scene, including body and background, which limits practical usability. Meanwhile, full-scene restorers often ignore degradation cues entirely, leading to underdetermined predictions and visual artifacts. In this work, we propose Face2Scene, a two-stage restoration framework that leverages the face as a perceptual oracle to estimate degradation and guide the restoration of the entire image. Given a degraded image and one or more identity references, we first apply a Ref-FR model to reconstruct high-quality facial details. From the restored-degraded face pair, we extract a face-derived degradation code that captures degradation attributes (e.g., noise, blur, compression), which is then transformed into multi-scale degradation-aware tokens. These tokens condition a diffusion model to restore the full scene in a single step, including the body and background. Extensive experiments demonstrate the superior effectiveness of the proposed method compared to state-of-the-art methods.
