Microsoft indicated that it paid special attention to the careful selection of the data on which the neural model was trained, with the aim of obtaining realistic scenarios in image making, taking into account the observations and recommendations of artificial intelligence image generation experts.

The company’s experts explained that their project aims to generate very realistic images that display lighting elements and reflections very close to reality.

Microsoft also noted that one of the main advantages of its new model is the speed of response to the user and the design of an image that accurately and professionally reflects his requirements.

“MAI-Image-1” is currently available on the “LMArena” platform, where users can compare the performance of different artificial intelligence models. It is planned that this model will be integrated in the future with the Copilot virtual assistant that works with computers and smart mobile devices.

Microsoft is seeking to keep pace with software development companies such as Google, Apple, and OpenAI in the field of neural network software, and also develop its own models for generating videos, especially since OpenAI had announced the launch of the second generation of the “Sora” tool, which creates videos with artificial intelligence.

Source: mail.ru