Meitu Wu Xinhong: I hope more people will use the imaging productivity tools in the AI ​​era

On June 19, at the second Meitu Image Festival, Wu Xinhong, founder, chairman and CEO of Meitu, shared the story behind Meitu's large visual model MiracleVision (Qixiang Intelligence), and focused on "AI The theme of "Image Productivity Tool of the Times" reviewed the journey of leading Meitu to explore AI in the past year.

Wu Xinhong said that the value of tools will explode in the AI ​​era. The close integration of AI and imaging productivity tools has continuously lowered the threshold for high-quality content creation while improving the production efficiency of the imaging industry. We hope that more people can share the benefits brought by AI technology and jointly use image productivity tools in the AI ​​era.

Founder, Chairman and CEO of Meitu Company Wu Xinhong

AI has become the energy driving the transformation of tool efficiency, and Meitu AIGC products are ushering in an explosion

At the Image Festival, Wu Xinhong interpreted the theme of "Imaging Productivity Tools in the AI ​​Era": AI is becoming the energy source to promote the transformation of tool efficiency. "The significance of AI to tools is like the significance of electricity to electrical appliances. Discovery drove the industrial revolution, and the development of AI brought us into the era of intelligence."

Founder, Chairman and CEO of Meitu Company Wu Xinhong

Wu Xinhong said that the tool attributes of Meitu products have changed today. In the past, Meitu's main service targets were users' life scenes, such as taking pictures, retouching pictures, trimming videos and social sharing. In the AI ​​era, in addition to life scenes, Meitu will provide services for more productivity scenes, such as visual creation, commercial photography, professional video editing, commercial design, etc. ”

Since the establishment of MT Lab (Meitu Imaging Research Institute) in 2010, Meitu has begun to make plans in the field of artificial intelligence, focusing on the field of computer vision. Since 2013, it has focused on deep learning. In 2016, the company launched the prototype of AIGC product "Hand-drawn Selfie", which was widely welcomed by users around the world. Since November 2022, Meitu AIGC products have entered an explosive period.

The "AI Painting" launched in November 2022 has put Meitu XiuXiu at the top of the application list in many countries and regions around the world; the "AI Animation" launched in March this year has popularized animation video productivity tools; 5 In August, it launched "AI Product Pictures", "AI Fitting" and other functions based on business scenarios to reduce costs and increase efficiency in commercial design, and helped users train personal AI models through "AI Photos"... Wu Xinhong reviewed a series of Meitu combines the actual needs of users and has achieved great results in the field of AI.

Driven by AI, as of June 18, the number of Meitu’s global VIP members has increased from 4.5 million announced at last year’s Image Festival to 7.19 million. With the rapid growth in the number of VIP members, our subscription revenue last year exceeded advertising revenue for the first time, and we entered the second stage of Meitu commercialization. "Wu Xinhong said.

The core of making products is to return to user needs, 7 new products make the AI ​​ecosystem more complete

In the first Meitu Image Festival, Meitu released 6 products. After a year of continuous iteration, many products have achieved outstanding results. For example, the number of monthly active users of Wink has exceeded 10 million, and Meitu ID Photo has become the industry champion of online ID photos. .

At the 2nd Meitu Video Festival, Meitu released AI visual creation tool WHEE, AI spoken video tool for shooting, desktop AI video editing tool WinkStudio, Meitu Design Studio 2.0 focusing on AI business design, and AI digital human generation There are 7 products including tool DreamAvatar, Meitu AI assistant RoboNeo, and Meitu visual model MiracleVision.

Wu Xinhong emphasized that after the release of seven products, Meitu has advanced from life scenarios to productivity scenarios, and gradually formed an AI product ecosystem built from the bottom layer, middle layer and application layer.

In the application layer, Meitu innovates and explores in the fields of AI pictures, AI videos, AI design, and AI digital people, and realizes AI image service aggregation by Meitu Xiuxiu; in the middle layer, Meitu provides the Meitu AI open platform, Meitu It is a combination of the Meitu creative platform and Meitu AI assistant RoboNeo; the bottom layer is the MiracleVision visual model, which provides AI model capabilities for all Meitu products.

“Through this ecological structure, we will continue to bring users a better imaging experience.” Wu Xinhong said.

Integrate aesthetic cognition into large models and reverse the evolution of technology from visual creation scenes

At the Image Festival, Meitu released MiracleVision, China’s first large-scale AI visual model that understands aesthetics.

"MiracleVision combines many of our thoughts on aesthetics and creative scenarios.". "Wu Xinhong said that Meitu has integrated its long-term accumulated aesthetic cognition into the MiracleVision large visual model, and built an aesthetic evaluation system based on machine learning to score "aesthetic scores" for the model generation results, thereby continuously improving the model's understanding of aesthetics. . In addition, MiracleVision visual model has a unique aesthetic creator ecology, with artists, designers and other professionals with profound aesthetic backgrounds participating in the construction of visual models.

Different from other large models on the market, MiracleVision large visual models have strong visual expression and creativity, and are good at inferring technological evolution from visual creation scenarios such as painting, design, film and television, photography, games, 3D, and animation. Generating directions such as Asian portrait photography, national style and trend, commercial design, etc.

According to Wu Xinhong, successful large-scale visual models must receive continuous participation and support from creators and developers to form a good "model ecology." Meitu will systematically build a model ecosystem through model training, distribution and commercialization.

Wu Xinhong said that with the close integration of AI and imaging productivity tools, the threshold for creating high-quality content will continue to be lowered, and the production efficiency of many industries will also continue to improve. Meitu will uphold its mission of "bringing technology and art to a beautiful intersection" and continue to bring better products and services to users.

The above is the detailed content of Meitu Wu Xinhong: I hope more people will use the imaging productivity tools in the AI ​​era. For more information, please follow other related articles on the PHP Chinese website!