Here's a video describing an AI program that restores blurry and damaged old portraits. The technique is presented in a paper titled "Towards Real-World Blind Face Restoration with Generative Facial Prior."
As Louis Bouchard says in his piece about the software, the results are just best guesses of what the person might look like:
They do not represent the actual image. It's important to understand that these results are just guesses from the model- guesses that seem pretty damn close. To human eyes, it seems like the same image representing the person. We couldn't guess that a model created more pixels without knowing anything else about the person.
So the model tries its best to understand what's in the picture, fill in the gaps, or add pixels if the image is of low resolution.
You can try it out here.
I tried it out on a photo of my Belarusian ancestors on my mother's side, and the results were impressive. The blurriness is gone and details are very clear. The main problem are the eyes, which look unnatural.
