Programs that can paint old black and white images, quite a lot. But they are all far from perfect, says Gizmodo. According to the publication, developers from the University of California in the United States developed a neural network application that studied millions of color images and figured out what colors most often color specific details. Thus, the application has learned to apply algorithms to images that have never seen.

Most often the result is impressive, even if the application sometimes makes mistakes in the color details.

To avoid these mistakes, a user interface has been added to the application, through which neural networks can be set the desired direction and give a hint.

The app itself suggests choosing from a palette of colors that will fit in a particular area of ??the image. For example, it can be a certain shade of green. As a result, the dyeing process can be completed in a couple of minutes as opposed to the hours that can be spent in Photoshop.

Try the application on the pictures from your own family archive, if you download it from this link.