CodeFormer

What CodeFormer is

CodeFormer is a free, open-source research method for restoring faces in old photos and AI-generated images. It uses a learned discrete codebook and Transformer-based code prediction to reconstruct faces from degraded inputs. A controllable weight lets users trade image quality against fidelity: lower values tend toward higher quality, while higher values tend toward greater fidelity. The project supports face restoration and whole-image enhancement, with optional Real-ESRGAN upsampling for faces and backgrounds. Separate tools provide colorization for black-and-white or faded cropped faces and inpainting for cropped faces marked with a white brush. Its inference script accepts .mp4, .mov, and .avi clips; Windows and Mac users are instructed to install ffmpeg first. Setup requirements include PyTorch 1.7.1 or later, CUDA 10.1 or later, Python 3.8 in the setup example, and other packages in requirements.txt. Integrations include Stable Diffusion WebUI, ChaiNNer, and ComfyUI. The project uses the NTU S-Lab License 1.0, and the Replicate model page says its API cannot be used commercially. The project presents CodeFormer as a research method and asks users who use its dataset or paper to cite the work.