The Khronos® Group continues to drive innovation in 3D rendering with its latest updates to the ANARI™ SDK and a successful Hackathon showcasing the power of the ANARI API in real-world applications. Supercomputing (SC24) also serves as the backdrop for exciting new developments that solidify ANARI’s role in high-performance scientific visualization.

What is ANARI?

ANARI™ (ANonymous Application Rendering Interface) is an open standard 3D rendering engine API developed by the Khronos Group, offering portable access to advanced 3D rendering techniques, such as ray tracing and global illumination. ANARI’s cross-platform capabilities make it widely adopted in scientific visualization, delivering performance and flexibility through implementations like AMD’s RadeonProRender, Intel’s OSPRay, and NVIDIA’s VisRTX.

ANARI has been shaped by public feedback, resulting in key improvements such as:

• Enhanced object interfaces for better developer experience
• Guaranteed stream robustness for error handling
• Refined runtime feature queries
• Directly mapped array parameters
• Better volume shading
• Full compatibility with Khronos glTF™ PBR (Physically-Based Rendering) materials

These features continue to make ANARI a go-to tool for developers seeking sophisticated 3D rendering functionality across multiple platforms.

ANARI SDK Enhancements for Developers

Khronos is pleased to introduce significant updates to the open-source ANARI Software Development Kit (SDK) in the latest v0.11.0 release on GitHub, ensuring an even more powerful toolkit for developers working on 3D rendering applications. These updates include:

• OpenUSD Hydra Plugin “hdAnari” included in latest release: hdAnari enables the use of ANARI-based renderers within Hydra-enabled applications. This plugin was initially made public in the SDK repository earlier this year, but a number of improvements have already been made with its inclusion in the latest tagged SDK release. While still a prototype, the plugin is already usable in OpenUSD applications such as USDView and will continue to improve in future releases.
• Better debug layer usability: Developers can now enable the debug layer using environment variables, removing the need for explicit setup code in the application. The debug layer gives application developers the chance to see whether their application is correctly using the ANARI API and will check its usage against whatever ANARI backend it currently is using at runtime.
• Conformance Test Suite (CTS) improvements: ANARI’s CTS continues to evolve, with better support for newer versions of python and additional test description information added to generated reports to add clarity for adopters and developers. Additionally, the CTS now covers more features in existing extensions and adds coverage of the new ANARI_KHR_GEOMETRY_ISOSURFACE extension.

These enhancements empower developers to deliver high-performance 3D visualization and rendering while leveraging ANARI’s open and accessible API framework.

Hackathon Results

At the ANARI Virtual Hackathon 2024, developers from around the globe gathered to showcase their skills in pushing the boundaries of 3D rendering with ANARI. The hackathon united creative minds, moving forward solutions in scientific visualization and simulation.

30 registered participants combined in teams to tackle complex 3D rendering challenges and showcase the flexibility and power of the ANARI API. Represented institutions included Argonne National Laboratory, Lawrence Livermore National Laboratory, Oak Ridge National Laboratory, University of Cologne, Intel, Kitware, NIST, NVIDIA, and UIUC.

In three days of the hackathon, the following contributions were made to the ANARI ecosystem:

• Overhaul to VTK-ANARI integration: The existing ANARI integration into VTK has already improved over the last year. However, ANARI’s ability to dynamically query the features of an implementation had not yet been fully utilized. During the hackathon, this integration was overhauled to better set up mainstream VTK applications, such as ParaView and VisIt, to be able to generate their renderer settings UI dynamically. This will let developers use any ANARI device at runtime and use an interface tailored to the implementation without needing to change application code.
• New VTK-m implementation: The VTK-m accelerated visualization library (not to be confused with VTK) has basic ray casting code for simple, portable rendering for VTK-m users. During the Hackathon, a team created an initial prototype of an ANARI implementation that leverages VTK-m’s existing ray-casting code to render images. This effort helps close the loop in VTK-m’s push toward eventually using ANARI for all rendering operations. VTK-m already has an integration that can use ANARI for rendering. In contrast, the new implementation developed during the hackathon will let VTK-m be one of the many ANARI implementations users can choose from at runtime.
• Volume rendering with the Cycles renderer: The Cycles renderer found in Blender can be used as a standalone rendering engine apart from Blender itself. During the hackathon, a team took a prototype implementation of Cycles behind the ANARI API and explored how to implement ANARI’s relevant volume rendering extensions from the v1.0 specification using Cycles. The team demonstrated a “hello world” prototype using Cycles and set up future efforts to build on their work to bring the ANARI-Cycles implementation into a more mature state.

This event demonstrated the potential of ANARI to enable state-of-the-art advancements in 3D rendering, primarily in scientific visualization. The event's success and positive feedback received from participants will positively impact the greater ANARI ecosystem and lead to future hackathon events. Be on the lookout for future ANARI hackathon events!

Feedback Welcome!

We would love your feedback on how ANARI and the ANARI SDK work. We encourage any contributions or issues to be reported at the ANARI SDK GitHub repo. Connect with us on the ANARI-Registry GitHub project anytime!  

Get More ANARI at SC24

Supercomputing 2024 offers several opportunities for developers and visualization experts to learn more about the latest developments in ANARI. The ANARI Birds of a Feather (BoF) session, which takes place on November 19 from 12:15-1:15 PM, will feature presentations from industry leaders and developers, focusing on use cases and best practices for distributed rendering and scientific visualization. Attendees will gain insights into how ANARI continues transforming the landscape of real-time rendering in high-performance computing (HPC). If you are attending SC24, you can add the ANARI BOF to your schedule here. No additional conference pass is needed to attend; an Exhibits Only, Exhibitor, or Technical Program pass will get you into this session.

For the supercomputing community, the recently published paper, Standardized Distributed Rendering Using ANARI, will provide valuable insights into using ANARI for large-scale distributed rendering. This paper, first presented at the LDAV 2024 conference, explores the standardization of distributed rendering techniques, a critical topic for supercomputing and HPC applications. With ANARI, distributed rendering becomes more accessible and streamlined, making it an essential solution for researchers and professionals tackling large, complex datasets.

Stay tuned for more updates on ANARI from SC24 and beyond as the Khronos Group continues to support and evolve this essential 3D rendering API.

About ANARI

ANARI is an open, cross-platform, 3D rendering engine standard that provides a high-level, cross-vendor API to significantly simplify the development of visualization applications that leverage the power of accelerated 3D rendering. The ANARI 1.0 API specification and SDK were released in August 2023. ANARI enables developers to build an in-memory scene representation that can be rendered without programming to low-level graphics APIs or libraries. ANARI provides rendering engines the semantics to expose innovation through extensions, access asynchronous scene updates and zero-copy data arrays for low frame latency, and ultimately create beautifully rendered imagery without proprietary APIs, all while enabling the interactivity necessary for a wide range of applications.

Khronos welcomes feedback from the graphics development community at the ANARI GitHub.

Learn more at: www.khronos.org/anari.

Banner Image: An example of how traditional scientific visualization tools, like VisIt, can be augmented with ANARI back-ends to export datasets to USD. The USD files can then be imported into applications like NVIDIA’s Omniverse to take advantage of its advanced rendering capabilities. This workflow will be demonstrated at SC24 during the ANARI BoF.