sczhou/codeformer

Robust face restoration algorithm for old photos / AI-generated faces

Capabilities

Reference Images

Community model (estimated from hardware time)

Name	Type	Description	Default	Constraints
`image`*	string(uri)	Input image	`—`	—
`background_enhance`	boolean	Enhance background image with Real-ESRGAN	`true`	—
`codeformer_fidelity`	number	Balance the quality (lower number) and fidelity (higher number).	`0.5`	min: 0, max: 1
`face_upsample`	boolean	Upsample restored faces for high-resolution AI-created images	`true`	—
`upscale`	integer	The final upsampling scale of the image	`2`	—

imagerequiredstring

Input image

background_enhanceboolean

Enhance background image with Real-ESRGAN

Default: true

codeformer_fidelitynumber

Balance the quality (lower number) and fidelity (higher number).

Default: 0.5min: 0, max: 1

face_upsampleboolean

Upsample restored faces for high-resolution AI-created images

Default: true

upscaleinteger

The final upsampling scale of the image

Default: 2

Version: cc4956dd26faUpdated: 7/25/202650.4M runs