Reallusion Partners with ASUS to Upgrade its Computing Platform


Content Creation Reallusion Partners with ASUS to Upgrade its Computing Platform Print Page As global digital-content creation enters the AI-driven era, Taiwan-based animation software leader Reallusion is accelerating its R&D innovation through both technological and computing-power upgrades. By deploying ASUS ESC4000A servers equipped with NVIDIA L40S GPUs, Reallusion has transformed previously hardware-limited AI model training and digital twin applications into a high-efficiency, high-quality, multi-threaded workflow—unlocking new possibilities for the film, gaming, and industrial simulation markets. The ASUS ESC4000A-E12 is a server built for enterprise AI infrastructure, featuring industry-leading GPUs, faster GPU interconnects, and higher-bandwidth architecture for unprecedented performance. Driving AI-powered Animation Innovation Founded in 2000, Reallusion has long specialized in developing digital animation software and content platforms for global markets. According to Sheng-Yan Huang, Director of the Advanced Innovation Department, the company’s three flagship products—iClone (3D animation and motion software), Character Creator(3D character-creation tools), and Cartoon Animator (2D animation software)—are widely adopted across film production, game development, and education. In the early stages of integrating AI into its animation pipeline, Reallusion relied on desktop systems with NVIDIA GeForce or NVIDIA RTX™ graphics cards for model training. However, limited GPU memory often caused interruptions during training. While the company tried renting third-party cloud computing services, long-term costs were high and operational flexibility was limited. As project scale and complexity increased, Reallusion urgently needed a local-computing platform that offered both large GPU memory capacity and expansion flexibility. The company adopted ASUS ESC4000A servers powered by NVIDIA L40S GPUs as the core infrastructure for its AI R&D. Powerful Compute Platform Boosts R&D Efficiency Using ActorGen AI, Reallusion generates editable 3D virtual characters from a single photo—including head modeling, body pose, and automatic clothing modules—while accelerating training and generation through the ASUS compute platform. The selected ASUS ESC4000A system features four NVIDIA L40S GPUs, providing a total GPU memory capacity of 196GB per server. Huang highlighted three key advantages: exceptional computational performance for training large AI models, significantly improved output quality, and system stability that supports continuous 24/7 operation. Through NVIDIA NVLink™, multiple GPUs are connected into a unified memory pool—breaking single-machine limitations and enabling scalable upgrades. This architecture allows Reallusion to run multiple compute-intensive projects simultaneously, minimizing idle time and improving both R&D efficiency and final output quality. Six AI Applications Accelerated by ASUS Infrastructure The hardware upgrade has directly accelerated and optimized six of Reallusion’s AI R&D initiatives: AI-Generated 3D Models: Create lifelike virtual human heads from a single photo. The training dataset now includes 30,000 cross-ethnic samples across six major ethnic groups, with generation time reduced to just 2–3 seconds. AI-Generated Textures: Automatically produce high-quality materials such as leather or knitwear, supporting both text-to-image and image-to-image generation for faster content creation. AI Motion Generation: With only a few key poses, the system can automatically fill in all intermediate frames, producing smooth and natural animations that lower the production threshold for complex scenes. AI Rendering (AI Render): Convert cartoon or game-style visuals into realistic or alternative styles with precise rendering control. AI RAG: Combine large-language-model reasoning with virtual character animation to enable real-time interactive experiences—ideal for virtual assistants, exhibitions, and smart venues. Digital Twin: Integrate characters, props, and interaction logic into virtual environments for factory simulation, crowd management, and intelligent transportation use cases. Real Results and Continuous Collaboration The enhanced computing power has drastically shortened AI model training time while preventing task interruptions caused by resource limitations. With NVLink interconnects, Reallusion’s team can now run multiple R&D projects concurrently on a single platform, achieving higher collaboration efficiency and improved rendering stability. Moreover, the system’s reliability enables 24/7 continuous operation, supporting both high-load R&D and commercial virtual-interaction platforms without downtime. During deployment, the ASUS team provided end-to-end support, from configuration recommendations and installation to NVLink setup and system tuning. Reallusion initially tested the servers through a marketing partnership, validating performance in real projects before expanding adoption. Both sides continue to explore upgrades and new collaboration opportunities to ensure long-term R&D success. Huang stated that Reallusion will continue to refine its six AI applications and develop additional commercial modules across film, gaming, industrial simulation, and smart-space markets. The company plans to expand computing resources, separate R&D and production servers, and advance digital twin and AI agent natural-language interaction applications—continuing its deep collaboration with ASUS to bring Taiwan’s animation and AI innovation to the global stage.

Content Creation

Reallusion Partners with ASUS to Upgrade its Computing Platform

Print Page

As global digital-content creation enters the AI-driven era, Taiwan-based animation software leader Reallusion is accelerating its R&D innovation through both technological and computing-power upgrades. By deploying ASUS ESC4000A servers equipped with NVIDIA L40S GPUs, Reallusion has transformed previously hardware-limited AI model training and digital twin applications into a high-efficiency, high-quality, multi-threaded workflow—unlocking new possibilities for the film, gaming, and industrial simulation markets.

The ASUS ESC4000A-E12 is a server built for enterprise AI infrastructure, featuring industry-leading GPUs, faster GPU interconnects, and higher-bandwidth architecture for unprecedented performance.

Driving AI-powered Animation Innovation

Founded in 2000, Reallusion has long specialized in developing digital animation software and content platforms for global markets. According to Sheng-Yan Huang, Director of the Advanced Innovation Department, the company’s three flagship products—iClone (3D animation and motion software), Character Creator(3D character-creation tools), and Cartoon Animator (2D animation software)—are widely adopted across film production, game development, and education.

In the early stages of integrating AI into its animation pipeline, Reallusion relied on desktop systems with NVIDIA GeForce or NVIDIA RTX™ graphics cards for model training. However, limited GPU memory often caused interruptions during training. While the company tried renting third-party cloud computing services, long-term costs were high and operational flexibility was limited.

As project scale and complexity increased, Reallusion urgently needed a local-computing platform that offered both large GPU memory capacity and expansion flexibility. The company adopted ASUS ESC4000A servers powered by NVIDIA L40S GPUs as the core infrastructure for its AI R&D.

Powerful Compute Platform Boosts R&D Efficiency

Using ActorGen AI, Reallusion generates editable 3D virtual characters from a single photo—including head modeling, body pose, and automatic clothing modules—while accelerating training and generation through the ASUS compute platform.

The selected ASUS ESC4000A system features four NVIDIA L40S GPUs, providing a total GPU memory capacity of 196GB per server. Huang highlighted three key advantages: exceptional computational performance for training large AI models, significantly improved output quality, and system stability that supports continuous 24/7 operation.

Through NVIDIA NVLink™, multiple GPUs are connected into a unified memory pool—breaking single-machine limitations and enabling scalable upgrades. This architecture allows Reallusion to run multiple compute-intensive projects simultaneously, minimizing idle time and improving both R&D efficiency and final output quality.

Six AI Applications Accelerated by ASUS Infrastructure

The hardware upgrade has directly accelerated and optimized six of Reallusion’s AI R&D initiatives:

AI-Generated 3D Models: Create lifelike virtual human heads from a single photo. The training dataset now includes 30,000 cross-ethnic samples across six major ethnic groups, with generation time reduced to just 2–3 seconds.
AI-Generated Textures: Automatically produce high-quality materials such as leather or knitwear, supporting both text-to-image and image-to-image generation for faster content creation.
AI Motion Generation: With only a few key poses, the system can automatically fill in all intermediate frames, producing smooth and natural animations that lower the production threshold for complex scenes.
AI Rendering (AI Render): Convert cartoon or game-style visuals into realistic or alternative styles with precise rendering control.
AI RAG: Combine large-language-model reasoning with virtual character animation to enable real-time interactive experiences—ideal for virtual assistants, exhibitions, and smart venues.
Digital Twin: Integrate characters, props, and interaction logic into virtual environments for factory simulation, crowd management, and intelligent transportation use cases.

Real Results and Continuous Collaboration

The enhanced computing power has drastically shortened AI model training time while preventing task interruptions caused by resource limitations. With NVLink interconnects, Reallusion’s team can now run multiple R&D projects concurrently on a single platform, achieving higher collaboration efficiency and improved rendering stability.

Moreover, the system’s reliability enables 24/7 continuous operation, supporting both high-load R&D and commercial virtual-interaction platforms without downtime.

During deployment, the ASUS team provided end-to-end support, from configuration recommendations and installation to NVLink setup and system tuning. Reallusion initially tested the servers through a marketing partnership, validating performance in real projects before expanding adoption. Both sides continue to explore upgrades and new collaboration opportunities to ensure long-term R&D success.

Huang stated that Reallusion will continue to refine its six AI applications and develop additional commercial modules across film, gaming, industrial simulation, and smart-space markets. The company plans to expand computing resources, separate R&D and production servers, and advance digital twin and AI agent natural-language interaction applications—continuing its deep collaboration with ASUS to bring Taiwan’s animation and AI innovation to the global stage.