face restoration
CodeFormer is a blind face restoration transformer that uses a learned codebook to restore realistic facial details in degraded photos and videos.
Solves the same problem with a different architecture or at a different layer. Expect to rewrite the integration.
Parsing-guided video face restoration with temporal coherence; choose it when you need temporally stable video face enhancement.
[IJCAI'24] Beyond Alignment: Blind Video Face Restoration via Parsing-Guided Temporal-Coherent Transformer
Variational codebook video face restoration with temporal coherence; choose it for video rather than stills if temporal stability matters.
[ICCV2025 Highlight] DicFace: Dirichlet-Constrained Variational Codebook Learning for Temporally Coherent Video Face Restoration
Diffusion-based blind face restoration; choose it when you prefer a diffusion prior over a codebook lookup approach.
DifFace: Blind Face Restoration with Diffused Error Contraction (TPAMI, 2024)
An earlier or more limited way to do the job. Still the right call when you need something small, proven, or CPU-only.
Older GAN-based face restorer; choose it for legacy pipeline compatibility and smooth background blending, but it is unmaintained.
GFPGAN aims at developing Practical Algorithms for Real-world Face Restoration.
Earlier codebook-transformer face restorer with an Apache-2.0 license; choose it when you need permissive licensing despite lack of maintenance.
[CVPR 2022] RestoreFormer: High-Quality Blind Face Restoration from Undegraded Key-Value Pairs
Older semantic-parsing-guided face restoration method (CVPR2021); choose for legacy research comparisons, but it is not maintained.
PyTorch codes for "Progressive Semantic-Aware Style Transformation for Blind Face Restoration", CVPR2021
These projects were analysed and named CodeFormer among their alternatives. The relationship is not symmetric — how CodeFormer rates them is a separate judgement, made when CodeFormer is analysed in its own right.
calls CodeFormer “choose for face restoration with codebook transformer prior; handles very degraded faces and offers fidelity control”
GFPGAN aims at developing Practical Algorithms for Real-world Face Restoration.
calls CodeFormer “Dedicated blind face-restoration model; choose when face identity matters more than general scenes.”
Real-ESRGAN aims at developing Practical Algorithms for General Image/Video Restoration.