Getting Started with NVIDIA Maxine

NVIDIA Maxine™ features GPU-accelerated SDKs with state-of-the-art AI features for developers to build virtual collaboration and content creation solutions, including video conferencing and streaming applications. Maxine consists of three SDKs: Video Effects, Audio Effects, and Augmented Reality (AR).

Download the Latest Video Effects SDK

Apply incredible AI effects for video quality enhancement. The Video Effects SDK is free to use within any application on both client-side or server/data center. To download the SDK, you must be a member of the NVIDIA Developer Program.

Key Features	Virtual background - Beta Super resolution (up to 4X scaling factor) - Beta Upscaler (up to 4X scaling factor) - Beta Artifact reduction - Beta Video noise removal - Beta GPU-accelerated image format conversion Performance optimized on NVIDIA platforms (PC, data center, server) Containerized deployment available
Latest Release Highlights	Virtual Background Quality mode enhanced for higher quality Added CUDA graph optimization option for improving overall latency under a GPU-intensive workload Super Resolution 4K video input support for 1.33x, 1.5x and 2x super resolution Support for A100 MIG for Linux SDKs Update to TensorRT 8.0.1.6, CUDA 11.3u1 & cuDNN 8.2.1 (Linux)
Operating Systems	Windows 10 64-bit, Ubuntu 18.04, CentOS 7
Dependencies	Windows SDK: NVIDIA Display Driver 465.89 or more recent, CMake 3.12+ Server SDKs (Linux): CUDA® 11.3.1, NVIDIA TensorRT® (TRT) 8.0.1.6, cuDNN 8.2.1, CMake 3.12+ For Containers: Docker (19.03+) and the up to date version of NVIDIA-Docker
Supported Hardware	Windows SDK: NVIDIA® GeForce RTX® 20XX and 30XX Series, Quadro RTX™ 3000, TITAN RTX™, or higher (any NVIDIA GPUs with Tensor Cores) Server SDK: V100, T4, A10, A30, A100 (with MIG support)

Download Documentation Tutorial DevBlog

Select Target Platform

Click on the green buttons that describe your target platform. Only supported platforms will be shown.

Windows Docker Container Ubuntu Centos

Download Installer for Windows 0.6.5.2

Download SDK

Installation Instructions:

Download and install the latest drivers (465.89+)
Download and unzip the SDK

Download Installer for Docker Container

Pull From NGC

Installation Instructions:

Download and install the NVIDIA Driver 465.19.01+
Install Docker (19.03+) and the up to date version of NVIDIA-Docker
Create an NGC account and generate an API key
Pull the container!

Download Installer for Ubuntu 18.04 local(tar) 0.6.5.0

Download SDK

Installation Instructions:

Install CUDA 11.3.1 (System reboot recommended after installation)
Download TensorRT 8.0.1.6 and cuDNN 8.2.1
sudo tar -xvf <TensorRT>.tar.gz -C /usr/local
sudo tar -xvf <cuDNN>.tar.gz -C /usr/local
sudo tar -xvf <VideoEffectsSDK>.tar.gz -C /usr/local

Download Installer for CentOS 7 local(tar) 0.6.5.0

Download SDK

Installation Instructions:

Install CUDA 11.3.1 (System reboot recommended after installation)
Download TensorRT 8.0.1.6 and cuDNN 8.2.1
sudo tar -xvf <TensorRT>.tar.gz -C /usr/local
sudo tar -xvf <cuDNN>.tar.gz -C /usr/local
sudo tar -xvf <VideoEffectsSDK>.tar.gz -C /usr/local

Download the Latest Audio Effects SDK

Improve audio quality from your incoming and/or outgoing audio feeds, making conversation easier to understand. To download the SDK, you must be a member of the NVIDIA Developer Program.

Key Features	Noise removal (Added Emotive Voice!) Room echo removal - Beta Audio Super Resolution - Apply for Early Access Acoustic Echo Cancellation - Apply for Early Access Supports wideband (16kHz) and full band (48kHz) speech 25+ audio profiles supported Performance optimized for NVIDIA platforms (PC, data center, server) Containerized deployment available
Latest Release Highlights	Acoustic Echo Cancellation feature available in Early Access Audio Super Resolution feature available in Early Access Emotive voice support Quality enhancements for Room Echo Cancellation & Noise Removal Support for A100 MIG for Linux SDKs Update to TensorRT 8.0.1.6, CUDA 11.3u1 & cuDNN 8.2.1 (Linux)
Operating Systems	Windows 10 64-bit, Ubuntu 18.04, CentOS 7, CentOS 8 RHEL7, RHEL8, Debian 10+
Supported Hardware	Windows SDK: NVIDIA GeForce RTX 20XX and 30XX Series, Quadro RTX 3000, TITAN RTX, or higher (any NVIDIA GPUs with Tensor Cores) Server SDK: V100, T4, A10, A30, A100 (with MIG support)
Software Dependencies	Windows SDK: NVIDIA Display Driver 465.89+ or more recent, CMake 3.12+ Server SDKs (Linux): CUDA 11.3.1, CMake 3.12+, (TRT 8.0.1.6 and CuDNN 8.2 are packaged) For Containers: Docker (19.03+) and the up to date version of NVIDIA-Docker

Download Documentation Tutorial DevBlog Apply for Early Access

Select Target Platform

Click on the green buttons that describe your target platform. Only supported platforms will be shown.

Windows Linux Docker Container

Download Installer for Windows 1.0.5.11

Download SDK

Apply for Early Access to get the new EA features! Installation Instructions:

Download and install the latest drivers (465.89+)
Download and unzip the SDK

Download Installer for Linux 1.0.5.1

Download SDK

The SDK is supported on multiple Linux distributions, please select the correct Distro while installing CUDA. Apply for Early Access to get the new EA features!

Install CUDA 11.3.1 (System reboot recommended after installation)
sudo tar -xvf <AudioEffectsSDK>.tar.gz -C /usr/local

Download Installer for Docker Container

Pull From NGC

Apply for Early Access to get the new EA features! Installation Instructions:

Download and install the NVIDIA Driver 465.19.01+
Install Docker (19.03+) and the up to date version of NVIDIA-Docker
Create an NGC account and generate an API key
Pull the container!

Download the Latest Augmented Reality SDK

Enable real-time 3D tracking of a person’s face using a standard web camera. Enjoy fun, engaging AR effects, such as overlaying 3D content on your face, controlling 3D characters, or adding immersive effects. To download the SDK, you must be a member of the NVIDIA Developer Program, and for early access to the AR SDK for datacenter/cloud deployment, you must apply for early-access and be preapproved.

Key Features	Face tracking - Windows in Beta, Linux in Alpha Face landmark tracking - Windows in Beta, Linux in Alpha Face mesh - Windows in Beta, Linux in Alpha Body pose estimation - Windows in Beta, Linux in Alpha Eye contact - Apply for early access Audio2Face - Coming Soon Performance optimized on NVIDIA platforms (PC, data center, cloud) (New!) Added MIG Support on A100
Latest Release Highlights	Eye Contact feature available in Early Access Quality improvements in accuracy & stability for 3D Body Pose keypoint tracking Support for A100 MIG for Linux SDKs Update to TensorRT 8.0.1.6, CUDA 11.3u1 & cuDNN 8.2.1 (Linux)
Operating Systems	Windows 10 64-bit, Ubuntu 18.04
Supported Hardware	Windows SDK: NVIDIA GeForce RTX 20XX and 30XX Series, Quadro RTX 3000, TITAN RTX, or higher (any NVIDIA GPUs with Tensor Cores) Server SDK: V100, T4, A10, A30, A100 (with MIG support)
Software Dependencies	Windows SDK: NVIDIA Display Driver 465.89+ or more recent, CMake 3.12+ Server SDKs (Linux): CUDA 11.3.1, TRT 8.0.1.6, CuDNN 8.2.1, CMake 3.12+ For Containers: Docker (19.03+) and the up to date version of NVIDIA-Docker

Download Documentation Apply for Early Access

Select Target Platform

Click on the green buttons that describe your target platform. Only supported platforms will be shown.

Windows

Download Installer for Windows 0.7.6.2

Download SDK

Installation Instructions:

Download and install the latest drivers (465.89+)
Download and unzip the SDK

DeepStream Maxine Container

Maxine AI features are now available with NVIDIA’s intelligent video analytics toolkit called DeepStream. Develop video conferencing solutions with reference or new pipelines and add on speech recognition capabilities using NVIDIA Riva from a single DeepStream Maxine container. To download the container, you must be a member of the NVIDIA Developer Program, and for early access to the DeepStream Maxine containers for datacenter/cloud deployment, you must apply for early-access and be pre approved.

Key Features	Reference video conferencing application supporting two input streams All Maxine Video Effects supported All Maxine Audio Effects supported Gstreamer plugins Riva Speech recognition capabilities
Operating Systems	Ubuntu 18.04, CentOS 7
Supported Hardware	Server SDK: V100, T4, A10, A100 , A30
Dependencies	Linux: NVIDIA Driver 460.32+, Docker (19.03+) and the up to date version of NVIDIA-Docker

Apply for early access

Resources

Introductory Resources

Developer Blogs

SoftBank Solves Key Mobile Edge Computing Challenges Using NVIDIA Maxine

Softbank is using NVIDIA Maxine to improve their user's communication experience and overcome 5G capacity and coverage issues. Read this technical blog to learn how they used features such as noise removal and virtual background to deliver the best video conferencing solution.

Read Blog

Transform Noisy Low-Resolution into High-Quality Videos for Captivating End-User Experiences

NVIDIA Maxine’s Video Effects SDK offers AI-based visual features that transform noisy, low-resolution video streams into pleasant user experiences. This blog demonstrates how developers can run these effects with standard webcam input and easily integrate them into video conference and content creation pipelines.

Read Blog

Achieve Noise Free Audio for Virtual Collaboration and Content Creation Applications

Maxine’s Audio Effects SDK empowers developers to build applications that integrate features such as Noise Removal and Room Echo Removal into their applications to improve audio quality. This blog showcases these effects and how to build applications that provide high audio quality.

Read Blog

Video

Create Your Own AI-Powered Virtual Background

Learn how to create a virtual background of your choice in 3 simple steps with NVIDIA Maxine’s Video Effects SDK’s API. You can choose process media, define camera resolution and codec, select the AI Green Screen (AIGS) configuration you like, and more. Try it yourself, and let us know your feedback!

WATCH VIDEO

Remove Background Noise with Maxine!

Learn how to remove background noise from audio streams in three simple steps with NVIDIA Maxine’s Audio Effects SDK’s Noise Removal. You can use both 16 khz and 48 khz audio. Try it yourself, and let us know your feedback!

WATCH VIDEO

Documentation

Video Effects SDK Documentation

Browse the latest NVIDIA Video Effects SDK documentation, which includes how to get started, software requirements, sample applications, setup instructions, and more.

VIEW DOCUMENTATION

Audio Effects SDK Documentation

Browse the latest NVIDIA Video Effects SDK documentation, which includes how to get started, software requirements, sample applications, setup instructions, and more.

VIEW DOCUMENTATION

Augmented Reality SDK Documentation

Browse the latest NVIDIA Augmented Reality SDK documentation, which includes how to get started, software requirements, sample applications, setup instructions, and more.

VIEW DOCUMENTATION

Redistributables

Video Effects

Looking to build and ship an application built on top of the Video Effects SDK? You can make use of the redistributables we provide!

DOWNLOAD FROM GITHUB

Audio Effects

Looking to build and ship an application built on top of the Audio Effects SDK? You can make use of the redistributables we provide!

DOWNLOAD FROM GITHUB

Augmented Reality

Looking to build and ship an application built on top of the Augmented Reality SDK? You can make use of the redistributables we provide!

DOWNLOAD FROM GITHUB

Support

Forums

Engage with NVIDIA directly and discuss the latest Maxine technology with a global developer community.

Learn More

Getting Started with NVIDIA Maxine