Today at Computex, Hour One’s groundbreaking AI Avatar technology was showcased as part of NVIDIA’s keynote. For the address, Hour One created a lifelike avatar of NVIDIA founder and CEO Jensen Huang that is capable of creating messages in multiple languages with a cinematic level of realism. This demonstrates the incredible potential of combining generative AI technologies, built on NVIDIA accelerated computing, to push the boundaries of what’s possible with digital humans.
As leaders in creating professional and cinematic videos using photorealistic avatars, Hour One is on a mission to enable enterprises to effortlessly create high-quality cinematic videos and photo-realistic avatars. Providing developers with access to its video cloud as a service makes it easier than ever to create rich media pipelines.
To bring these cinematic avatars to life, we built a sophisticated AI video cloud infrastructure parallelizing inference and 3D video rendering, enabling rich-media pipelines at an infinite scale, all powered by NVIDIA’s accelerated computing stack – including NVIDIA GPUs and AI acceleration libraries.
To enable parallelized Gen-AI video inference, we introduced a new GPU micro-orchestration technology that enables complex inner flows of audio-visual sequences. This innovative approach allows us to manage and optimize multi-modal Gen-AI processes, simultaneously, ensuring seamless and efficient operations.
To maximize cloud utilization for NVIDIA GPUs, we have introduced a multi-cloud adaptive GPU autoscaling system. This system is capable of scaling across multiple compute pipelines, including rendering, graphics, and inference, managing GPU allocation on all major clouds, on top of Kubernetes, to maximize resource utilization, performance, and cost-efficiency.
At Hour One, we rely on NVIDIA’s powerful technology at every stage of our operations. From data processing and model training to video rendering and encoding, leveraging the NVIDIA AI platform, which includes NVIDIA H100 and A100 Tensor Core GPUs, along with CUDA, Container Toolkit, and Video Codec SDK, we deliver high-quality, scalable solutions to our clients.
Our collaboration with NVIDIA empowers companies across all industries to harness the power of generative AI, advanced graphics, and rendering technology. By providing tools that simplify the creation of rich media content, we enable businesses to innovate and communicate in new and engaging ways.
Developers can find more information on NVIDIA ACE and access NVIDIA NIM microservices, including open-source sample workflows to help developers get started with ACE NIM.
