HANGZHOU, China, Dec. 10, 2025 /PRNewswire/ -- At the 2025 Cool+ Conference, Manycore Tech officially launched LuxReal, its innovative 3D AI content creation product designed to significantly improve the "spatial consistency" of AI-generated videos. LuxReal enables rapid generation of videos with spatially consistent subjects and controllable scenes from multimodal inputs like images or 3D models, greatly enhancing the practicality of AI video in sectors such as e-commerce, short-form video content, and advertising.

Built upon Manycore's proprietary AI 3D generation model, Lux3D, and integrated with image and video generation models, LuxReal introduces the industry's first 3D multi-agent system. This system efficiently produces creative video content that meets specific requirements while maintaining spatial consistency, striking an optimal balance between stability and flexibility in AI video creation. It addresses the more stringent requirements for reliable content generation of professionals including e-commerce practitioners, film and TV directors, advertising creatives, and product managers.

The lack of "spatial consistency" is a widespread challenge in current AI video generation, often manifesting as object displacement, spatial reasoning errors, and incorrect occlusion relationships during perspective shifts. This issue stems from most models being trained on 2D image or video data, lacking a deep understanding of and reasoning capability for 3D spatial structures and physical laws.

"While current AI creation tools have significantly lowered the barriers and costs for users in casual scenarios like entertainment and social video production, most AI videos still fall short of true physical controllability," said Long Tianze, AI Product Director at Manycore. He said that LuxReal tackles this core challenge through its two technological pillars: the AI 3D model Lux3D and a Multi-Agents system. This combination creates an integrated generation pipeline from 3D understanding and rendering to video enhancement, aiming to enable AI to truly 'comprehend space.' This leads to a substantial improvement in the "spatial consistency" of the output, significantly boosting the practicality and deployment value of AI video in serious applications like e-commerce, industrial design, gaming, and advertising.

Notably, to achieve optimal video generation results, LuxReal integrates multiple video models tailored to various scenarios—leveraging third-party models for creative support while LuxReal itself provides enhanced spatial consistency. "A practical AI, whether it's from a black box or white box approach, is ultimately what matters," Long Tianze added metaphorically.

An AI practitioner who has applied to join the beta program commented, "Previously, the focus was primarily on the creativity of AI video generation. However, future AI video must prioritize usability as much as novelty. The emergence of LuxReal offers a new solution for applying AI video in fields like film and TV production and e-commerce marketing."

LuxReal has started inviting users to its global beta test, with the official beta phase scheduled to commence in late December 2025.