Übersicht
GFPGAN reconstructs a clearer-looking face from degraded imagery using learned priors. This can invent details, so perceptual improvement and historical fidelity must be judged separately. Compare model variants for identity preservation and artifacts.
Wichtige Funktionen
- Facial priors
- Model variants
- Inference script
- Face helpers
- Optional background enhancement
Voraussetzungen, Installation und Schnellstart
Nutzung
How it works
Faces are detected and aligned, restored using a generative prior, and placed back into the image. Real-ESRGAN can separately enhance the background.
Audience and requirements
Restoration developers and researchers. Python, PyTorch, dependencies, weights, optional GPU.
Practical use cases
Old-photo previews; restored archive copies; restoration research.
Limitations and selection
Generated details can alter identity and are not forensic evidence. Tiny or occluded faces may fail.
Related projects and selection
photoprism/photoprism：Complement: archive restored copies alongside originals.
gradio-app/gradio：Complement: upstream links a Gradio demo for trying restoration.
Source review
Editorial analysis of upstream sources, without runtime or benchmark testing. Proposed workflows are editorial suggestions.
Modellkompatibilität und Anwendungsfälle
Match GFPGAN inference options to the weight version; background enhancement is a separate optional component.
Lizenz- und Risikohinweise
The repository page identifies no standard SPDX license. Read LICENSE; model weights and datasets may have separate terms.
Editorial source review 2026-09-09T05:00:00.950Z. README and live repository page verified; current stars/forks from GitHub HTML. Last-push metadata retained from 2026-09-05 discovery snapshot. No runtime benchmark. Integration proposals are editorial analysis.
Veröffentlichung und Wartung
Reviewed 2026-09-09. Counters come from repository pages; features are based on upstream documentation. See Releases in the source links. Editorial analysis of upstream sources, without runtime or benchmark testing. Proposed workflows are editorial suggestions.