Generation time reduced by almost 10 times
The SberBank team shared the latest news about the development of the proprietary neural network project — the free Kandinsky generative model, which creates images based on text descriptions in Russian and English. The updated version of Kandinsky 3.1 has been additionally trained on a larger set of images, which has improved the quality of the generated content. A limited number of users were the first to gain access to Kandinsky 3.1: artists, designers and bloggers.
First Deputy Chairman of the Board of Sberbank Alexander Vedyakhin said:
Today is exactly one year since Kandinsky 2.1 was released. During this time, we have constantly developed our neural network, which helps people create new images and gives absolutely everyone phenomenal opportunities for creativity. Compared to the previous model, Kandinsky 3.1 has become even faster, more convenient and more realistic. Kandinsky 3.1 — it is a flexible, feature-rich and completely free tool that will turn anyone into an artist and creator. Soon everyone will be able to test the new capabilities of the neural network. Like previous versions, the model will be free and available on different surfaces.
One of the key features of the version is a higher speed of image generation: generation time has been reduced by almost 10 times, and the resolution can be increased to 4K. It is also possible to improve the text query using a language model. Users have access to the functions of creating various variations of images, mixing pictures and text, creating packs of stickers and the ability to make local changes to the picture without changing the entire composition of the scene (ControlNet).
Also, a new Kandinsky Video 1.1 model for generating videos from text descriptions will appear in the near future. It will improve the quality of generation and also double the video resolution compared to Kandinsky Video 1.0.
