Improving AI-Generated Faces

GFPGAN (Generative Facial Prior — Generative Adversarial Network) is a free-to-use open AI project incorporating algorithms for human face restoration.

John Iovine

--

This GAN (generative adversarial network) improves the quality of human faces, it has both a restorative effect and upscaling (increasing the resolution) on the image quality. If you have blurry facial images, this AI can impressively sharpen the image.

Developed by researchers at Chinese company Tencent, this tool is free to use and may be downloaded.

Practical Uses

It’s a great tool for restoring old and/or damaged photographs. My interest was fixing AI-generated portraits that sometimes create facial images that are a little wonky.

Pretty Little Wrapper

This website has placed the GFPGAN style AI in a pretty simple-to-use wrapper. Go to;

https://www.restorephotos.io/restore

to access the tool, upload your picture, wait, and viola’ one sharpened image,

This free tool is provided without restrictions on use, but I can’t say how long it will remain free and available. Therefore, I am providing another option.

Google Colab

You can access GFPGAN using your browser directed to Google Colab. While not as simple to use as the wrapper mentioned above, you can download a copy of the AI to your google drive and run it from there. Think, keep it for your own. And it is easy to use.

https://colab.research.google.com/drive/1sVsoBd9AjckIXThgtZhGrHRfFI6UUYOo#scrollTo=ZwH2ifWEYEfJ

Google Colab is an online tool for using Python in your web browser.

Fortunately for us (well, me), you don’t have to be knowledgeable about Google Colab or python coding to get this software to run. When you first land on the GFPGAN Colab page, you may think, “Yikes, it’s too complicated, I did.” Fear not; you don’t have to change anything or write a line of…

--

--

John Iovine

Science writer, thinker, self-experimenter, focusing on personal development and health — www.john-iovine.com