MiracleVision, the self-developed AI visual model by Meitu, has been upgraded to version 4.0, introducing two new capabilities: AI Design and AI Video. AI Design includes four major functions: vector graphics, text effects, intelligent layering, and intelligent typesetting. Users can quickly generate vector graphics that match their design style, customize text effects, separate objects into independent layers through object recognition, and use natural language interaction to typeset text and images. MiracleVision 4.0 also features a visual model store, where users can choose from different visual models to create various novel visual effects. AI Video capabilities include four major functions: text-to-video, image-to-video, video framing, and video synthesis. Users can generate smooth videos using prompts, animate images, simulate professional camera movements, or apply different artistic styles to existing videos. The video generation capabilities of MiracleVision 4.0 have been integrated into industry workflows and are suitable for e-commerce and advertising industries. In e-commerce, MiracleVision can quickly produce a large number of product videos to make product displays more vivid. In advertising, MiracleVision can create creative videos that align with product characteristics, break through conventions, and deeply resonate with the brand concept. It is reported that MiracleVision 4.0 will be gradually applied to Meitu's products starting from January 2024. Additionally, commercial APIs will be opened, and multiple packages will be provided for users to choose from.