RePaint works by the beginning of ten different transparent pieces in thin layers, with the AI predicting the ideal pile needed to produce the intended colors. The feeding team also helped the AI to determine which colors were needed in certain areas of painting, adapting for changes in lighting conditions to ensure consistency. The technology also depends on half foot (where you create images using ink dots) to match colors more closely.
The technology can not, at the moment, reconstruct texture and surface reflection, so you do not have a matte or glossy look of some paintings. And surely you will not get Louvre's master of work at any time soon – the current output is the size of a postcard. When and when it's a scale, however, RePaint could have a wide appeal. You could get a more valid Monet or Van Gogh copy in your home, while museums could protect valuable roots by showing reproductions whenever possible.