Microsoft has announced the release of MAI-Image-2, its newest in-house text-to-image model. Designed to create more realistic visuals and streamline creative workflows, the new model builds on MAI-Image-1 and is already being integrated into Microsoft products such as Copilot and Bing Image Creator.
According to Microsoft, MAI-Image-2 focuses on generating images with more natural lighting, accurate skin tones, and environments that look more lifelike. The company aims to reduce the need for extensive post-editing of generated images.
How MAI-Image-2 Improves Realism And Usability
Microsoft positions MAI-Image-2 as a tool for practical creative work rather than experimental image generation. In a blog post, Microsoft states that the model is designed for creators who need visuals that feel like they exist in the real world.
The company also notes that improvements in realism can help reduce editing time during post-production. Developers worked with photographers, designers, and visual creators who pointed out issues related to realism and consistency in earlier systems.
Better Text Rendering In MAI-Image-2
Microsoft explains that MAI-Image-2 addresses a longstanding issue in image generation models: creating readable and consistent text. The company states that the model can produce visuals with text elements such as posters, infographics, and slides more accurately.
This enhancement aims to support scenarios where clear text is important, including marketing and design workflows. According to Microsoft, this improvement results in less disconnect between guidance and output when generating visuals that include text.
Where MAI-Image-2 Ranks Versus Google And OpenAI
Microsoft reports that MAI-Image-2 has advanced its MAI model family into the top three on Arena.ai rankings. Despite this progress, the model still lags behind competitors from Google and OpenAI, according to the same rankings.
Microsoft considers the improvement from MAI-Image-1 a notable step forward in its internal AI development. Mustafa Suleyman, CEO of Microsoft AI, noted that the company has made significant progress on its models and products, with more releases anticipated.
How To Access MAI-Image-2 In Copilot And Bing Image Creator
MAI-Image-2 is now available through the MAI Playground and is gradually being rolled out to Copilot and Bing Image Creator. Microsoft states that some enterprise customers can access the model via an API, with broader developer access expected through Foundry.
The company has not provided a timeline for when the model will be available globally across all services.
Thank you for being a Ghacks reader. The post Microsoft Launches MAI-Image-2, Its Most Realistic Image Model Yet appeared first on gHacks.
0 Commentaires