Kunlun Wanwei (kunlun.com) has launched the SkyReels-A3 model, an advanced AI-driven digital human video generation system based on a Diffusion Transformer (DiT) video diffusion architecture, integrated with frame interpolation, reinforcement learning, and camera control modules. It enables audio-driven animation of static images or portraits, producing lifelike mouth movements, natural body motion, and stable facial expressions for uses such as livestream e-commerce, music videos, film clips, and speeches. The release is part of Kunlun Wanwei’s “SkyWork AI Technology Release Week” (August 11–15, 2025), during which it will unveil a new AI model each day across five technology domains. The SkyReels-A3 model has been specifically optimized for online livestream scenarios and high-aesthetic content creation, supporting frame-level precision in camera movement via ControlNet-based control modules. This development reflects Kunlun Wanwei’s broader strategy to strengthen its position in AI video generation, digital humans, and large model commercialization, extending its influence in cultural, entertainment, and commercial applications.

SkyReels (skyreels.ai) is an AI-powered, all-in-one video creation platform that enables users to transform text, images, and animations into cinematic-quality videos with minimal effort. It integrates tools for scriptwriting, storyboarding, character customization, voiceover generation, lip-syncing, and film synthesis, supporting both short and long-form productions, including infinite-length videos. Built on open-source models like SkyReels-V1 and V2, it offers human-centric video generation with features such as multiple expressions, realistic movements, and advanced audio-driven portrait animation. The platform caters to filmmakers, marketers, educators, and content creators, and is accessible via web and mobile, with free and paid plans. Its technology supports applications in digital humans, drama production, advertising, and immersive storytelling, emphasizing ease of use, creative flexibility, and integration with platforms like ComfyUI and Hugging Face.

We generate AI fashion models for eCommerce brands, enabling brands to showcase collections on diverse models which actually look like their customers.

League of Legends is a team-based game with over 140 champions to make epic plays with.

LearnBrite bridges the experiential training gap by offering a solution that can create immersive and impactful self-paced experiences.

Leyard Optoelectronic Co., Ltd., founded in 1995 and listed on the Shenzhen Stock Exchange (300296.SZ) since 2012, is a leading Chinese company specializing in advanced visual technology. Headquartered in Beijing with operations worldwide, Leyard focuses on LED display solutions including micro LED, mini LED, and COB (chip-on-board) technologies. The company’s product portfolio spans indoor and outdoor displays, rental and staging solutions, transparent displays, and large-format LCD screens. It serves a wide range of sectors such as culture and tourism, broadcasting, education, public security, and the metaverse. Leyard operates multiple international subsidiaries including Leyard Vteam and Planar, with key production and service facilities in Shenzhen and global support networks across Asia, Europe, and the Americas. With over 4,600 employees, Leyard positions itself as a global pioneer in immersive, high-resolution visual experiences and continues to drive innovation in smart displays, cultural technology, and spatial computing.