A recent improvement to the software used to create this image was made using mid-journey it's called and version five. One part of that is that these systems are trained on a lot of images scraped from the internet especially high quality images from stock sites like Getty images and Reuters. So there is this correspondence between the training data and the images that they are good at producing. And I think this is something that is going to be very important when we keep in mind the future media environment.