От концепции до видео за несколько минут: Сбер представил бета-версию Kandinsky Video 1.1

The new version of the neural network has become significantly smarter

Sber has presented a free test version of Kandinsky Video 1.1. The neural network was developed by Sber AI in collaboration with scientists from the Artificial Intelligence Institute AIRI on a combined dataset from Sber AI and SberDevices. The updated version can create 6-second videos based on text and images.

The improved Kandinsky Video 1.1 neural network opens up new possibilities:

  • video footage lasting up to six seconds with a frame rate of 8 and 32 per second;
  • selection of the required video format at the output — 16:9, 9:16 or 1:1 (screen aspect ratio); 
  • control of video dynamics is due to the motion score parameter.

As a reminder, the first version of Kandinsky Video was presented in November 2023.

Less than a year has passed since the launch of the first version of the neural network, and during this time our team has significantly improved indicators such as the quality and speed of generating full-fledged videos, thereby opening up limitless horizons for creativity.
Andrey Belevtsev, Senior Vice President, Head of the Technological Development unit of Sberbank

The opportunity to test the neural network is open to all users. This can be done on the fusionbrain.ai platform or using the Telegram bot.

Read materials on the topic:

«Spoiler»: Sber's neural networks Kandinsky, SymFormer, Salyut and GigaChat wrote a track and shot a video

Sber expands access to the GigaChat chatbot in Telegram for all users

The first real Russian competitor to ChatGPT? Sber introduced GigaChat

Sources
Sberbank

Now on home