Thanks to Google Street View, we can see many corners of the world we may never even visit. Some artists even use it to create photos of remote places from their own home. Now, Google itself finds a way to produce professional-looking photos from their Street View shots. They have created Creatism, a deep-learning system that analyzes Street View scenes searching for a beautiful composition. The algorithm finds the scenes which it’s supposed to turn into shots worthy of professional photographers.
Hui Fang and Meng Zhang have published the study describing the process and the goals of the algorithm. Using their system, they try to mimic the workflow of landscape photographers “from framing for the best composition to carrying out various post-processing operations.” The algorithm also features dramatic mask, a feature that improves dramatic lighting for a photo.
According to the developers, it was a challenge measuring aesthetic value in a photograph, as it’s a really subjective concept. They have created an experiment that’s supposed to objectively measure the quality of the photos. In this experiment, they asked professional photographers to blindly rate the photos from different sources. Apparently, a part of the AI-created images was confused with professional work. You can check out the entire gallery here and see for yourself.
I must admit, for some of these photos I would never guess they were chosen, taken and processed by an algorithm. As someone who enjoys taking landscape photos, I believe this could be a useful tool for finding locations and getting ideas how to photograph them. On the other hand, it makes me feel a bit strange, considering that an algorithm can “take” a photo similar to the one I would. I’m very interested to hear your thoughts about this. Could this be game changing for landscape photographers? In a good or in a bad way?