Meitu's self-developed AI visual model 3.0 is released and will be fully used in imaging and design products

Recently at Meitu’s 15th anniversary birthday party, Wu Xinhong, founder, chairman and CEO of Meitu, released version 3.0 of the self-developed AI vision model MiracleVision, which will be fully used in Meitu’s imaging and Design products.

Wu Xinhong said that three months ago, MiracleVision's painting level was still in its infancy, but now the 3.0 version is able to depict real and delicate picture details. Currently, users can visit the official website of the AI ​​visual creation tool "WHEE" to experience

According to Meitu, the core capabilities of the self-developed AI visual model MiracleVision are divided into two major features: "wonderful ideas" and "intelligent creation".

"Wonderful Ideas" Level

MiracleVision lowers the threshold of use for the public through the "intelligent association of word prompts" function. When users enter keywords, MiracleVision can automatically add relevant expressions, such as light and shadow effects, texture, style, picture quality, etc., to promote equal rights in creation. In addition, MiracleVision uses the "prompt word precise control" function to meet more professional design requirements, such as using "close shot", "far shot", "forward light", "backlight" and other descriptions to control the final generated effect.

In the field of "intelligent creation"

In the field of "intelligent creation", MiracleVision can further enrich the details of the work and improve the expressiveness through the "deepening creation" function. Through the "AI screen expansion" function, the size of the work is larger and the details are richer. Through the "local modification" function, precise modifications and adjustments can be made to some pictures. The "resolution enhancement" function generates high-definition large images, making detail expression, color display, and object recognition more accurate and vivid.

Wu Xinhong said that AI visual large models not only bring better visual presentation, but also improve the efficiency of workflow. However, the combination of large models and workflow is still in the exploratory stage, and it is necessary to seize the advantages of application scenarios. Core needs are more important. MiracleVision will be implemented in the five major industries of e-commerce, advertising, games, animation, and film and television, hoping to help the five major industries "improve workflow efficiency."

Wu Xinhong also said that MiracleVision will give full play to its capabilities of whimsy and intelligent creation, delve into the above-mentioned five major industries, and continue to help more design scenarios. It is expected that by 2026-2030, the popularity of visual large model applications will enter a mature stage

The above is the detailed content of Meitu's self-developed AI visual model 3.0 is released and will be fully used in imaging and design products. For more information, please follow other related articles on the PHP Chinese website!