We use artificial intelligence. The algorithm is designed to most faithfully reproduce specified source image.
It uses specially extracted painting/drawing style from the second specified image or from the set of images.
No, this technique is not simple or even complicated image filtering. For each new photo it is necessary to run the end to end learning process. The process is similar to the learning by the human, but it is done on artificial neural networks.
This is done by gradually adjusting large number of the weights of specially designed artificial "deep" neural network. The network tries to paint a certain picture, the process is very similar to many attempts to paint a certain picture by a human, but here we do it automatically.
It uses artificial neural networks, and latest achievements of deep-learning, so it is able to adapt a large number of parameters of the neural network. The whole process requires significant amount of processing power, equivalent of large number of personal computers.
It takes quite a long time, because it is necessary to iterate. During each iteration, the machine attempts to paint and checks the results. After each iteration results are better, the final result is presented to the recipient.
The first possibility is to use the form, to upload all necessary input pictures and data to submit the work.
The second possibility is to use one of the predefined styles, available on the gallery subpage, by clicking button near expected style.
As a result, we will produce a new, unique image. Results are often unpredictable, sometimes excellent, interesting, but sometimes they need next attempts. There is no guarantee, that every attempt will produce good results. That's why we give additional coupons.