At the Open Data Science Conference 2023 in Boston from May 9–10, 2023, Oracle Cloud Infrastructure (OCI) and NVIDIA are collaborating to present an exciting product demo that combines the prowess of Stable Diffusion and NVIDIA Live Portrait technology. This demo showcases the seamless generation and animation of a person’s face using the open source Stable Diffusion image generator and NVIDIA Live Portrait, offering a powerful solution for data scientists, developers, and creators alike.
Introduced in 2022, Stable Diffusion has primarily been used to generate detailed images based on text descriptions. Its capabilities extend beyond image generation to cover things as diverse as game development, including textures, 3D models, depth maps, skins, and more. Diffusion models can also be utilized for other modalities, such as video and music generation. OCI’s demo of Stable Diffusion is powered NVIDIA A10 Tensor Core GPUs in OCI.
NVIDIA Live Portrait is a state-of-the-art AI technology that animates a 2D photo or stylized portrait from a person’s live video feed or recorded video clips. As part of the NVIDIA Omniverse Avatar Cloud Engine (ACE) suite of AI microservices, Live Portrait accelerates workflows for developers and creators by providing AI-powered character animation.
At the Open Data Science Conference 2023, OCI and NVIDIA are collaborating to showcase the powerful synergy of Stable Diffusion and Live Portrait. With OCI’s high-performance, scalable, and secure NVIDIA-powered platform, developers can efficiently run generative AI workloads to create and animate high-quality images.
The product demo highlights the ease and efficiency with which developers and creators can generate a person’s face using Stable Diffusion on OCI and animate it using NVIDIA Live Portrait. You can train Stable Diffusion on existing datasets and customized to generate unique content quickly and easily. This combination allows creators to maintain control over the creative process while benefiting from faster iterations and greater creative freedom.
Creators have the opportunity to experiment with the latest and greatest of features in the Stable Diffusion ecosystem, such as inpainting, outpainting, img2img, controlnet, prompt fusion, and text2video. These techniques unlock many unique capabilities for creators who want to use generative AI in innovative ways.
The demo also shows the scalability of Stable Diffusion and its ability to handle projects of various sizes. With the option to build Stable Diffusion clusters, performance scales almost linearly as compute resources increase. Developers and creators can use Stable Diffusion and Live Portrait to generate and animate high-quality content rapidly and efficiently, regardless of their project size.
OCI and NVIDIA’s product demo of Stable Diffusion and Live Portrait at ODSC 2023 is a groundbreaking advancement in the realm of generative AI. By combining the capabilities of OCI’s Stable Diffusion and NVIDIA Live Portrait, developers and creators have access to powerful tools that enable them to create and animate high-quality content quickly and efficiently.
Visit our booth at ODSC 2023 to learn more about Oracle Cloud Infrastructure and NVIDIA’s innovative technologies and solutions.
For more information, see the following resources:
NVIDIA resources for data science and AI development
Early-access application for Live Portrait
NVIDIA Riva for speech AI
Allen Roush is a Principal Machine Learning Architect who focuses on NLP Research within OCI.
Next Post