This is not going to taste good for Nikon or Canon: after Snapchat has already interfered and announced its intention to become the next big camera company, Google is now also involved in the field of photography. The search engine giant recently introduced an experimental system that uses Streetview images and Machine Learning (ML) to create realistic panoramas - and even impresses photographers with the results.
The algorithm developed by Google mimics a photographer's workflow in creating panoramas by searching through countless Google Streetview images, selecting the best elements and then cutting them together. For example, Google's virtual photographer searched through some 40,000 panoramas for the images shown below. The images were then processed in three steps: they were formatted correctly, manipulated with an HDR filter and then colored with a color mask.
The image quality amazes even professional photographers
Google uses a second algorithm (GAN) that evaluates the previously generated panoramas individually to ensure that the image quality is appropriate. Only then, at the end of the process, the “Dramatic Mask” lighting effect ensures that the images look as realistic as possible. One result that is quite impressive: Google has had the results rated by professional photographers on a scale of 1 (beginner) to 4 (pro) with a majority of the panoramas rated at 2.5 (semi).
But what do you think about the way of algorithm photography? Is that the future? Where can Google's virtual photographer help the user to take better photos or panoramas? We look forward to your feedback in the comments.