Independent documentation team
GFPGAN.com is an independent documentation project. We are not affiliated with Tencent ARC, who built and maintain GFPGAN, and we do not sell face restoration software.
Everything we publish about GFPGAN comes from one of three places: the original CVPR 2021 paper and its supplementary material, the reference implementation in the TencentARC/GFPGAN repository, or our own hands-on testing of local installs, Replicate endpoints, and Stable Diffusion integrations. Where a claim comes from our own testing, we say so and describe the setup.
We do not publish benchmark numbers we cannot attribute. If you find a figure on this site without a source, that is a bug — please report it and we will correct or remove it.
GFPGAN restores faces, Real-ESRGAN enlarges whole images. Here is the real difference, which one your photo needs, and why the order you run them matters.
Read more
Fix the errors that actually stop GFPGAN: the torchvision functional_tensor ImportError, missing basicsr, CUDA out of memory, and failed model loads.
Read more
Restore old scanned photos, blurry portraits, and damaged prints with AI — scanning tips, realistic expectations, and how to pair it with an upscaler.
Read more
Enable GFPGAN and CodeFormer face restoration in Automatic1111, Forge, and ComfyUI — with fixes for the 'model not found' errors people hit most.
Read more
A practical comparison of AI face restoration tools — browser tools, open-source models, and paid desktop software — so you can pick the right one.
Read more
Who built GFPGAN, where to download the model weights, and how to run it on Replicate without installing Python or owning a GPU.
Read more
What GFPGAN is, how the generative facial prior rebuilds detail, which photos it fixes, whether it changes how people look, and how to run it free.
Read more
GFPGAN, CodeFormer, and GPEN compared on identity preservation, speed, free access, and real-world restoration quality. Which model suits which photo.
Read more
Real GFPGAN before and after examples across old prints, blurry portraits, compressed screenshots, and low-light shots — and what it cannot recover.
Read more
CodeFormer vs GFPGAN compared on identity preservation, photo type, speed, and free access — with a clear answer on which to use for your photos.
Read more
Learn how GFPGAN AI face restoration fixes blurry, damaged, and old photos instantly. Free, browser-based, no app needed. Step-by-step guide inside.
Read more
Learn exactly how to use GFPGAN to restore blurry, old, and damaged photos. Expert tips, model settings, advanced workflows, and a free browser tool inside.
Read more
Step-by-step guide to install GFPGAN on Windows, Mac, and Linux. Covers Python setup, model download, common errors, and how to run your first restoration.
Read more