On September 12, the first Workshop: “Kandinsky: image generation” was held at the virtual platform of the Russian House in Brussels.
Denis Dimitrov, executive director for data research, Sber AI, scientific consultant at the AIRI Institute of Artificial Intelligence, presented a new version of the Kandinsky 2.2 neural network and how to work with it. According to the developers, using the generative model, you can create high-resolution photorealistic images and change the aspect ratio during generation.
Tatyana Nikulina, industrial engineer, spoke in detail about how to find an approach, a common language with a neural network, how to correctly come up with and set industrial engineering for such a model as Kandinsky, how to stylize visual content and images.
More details in the video tutorial:
In November we invite you to a master class on promtam for video, keep an eye on our poster.