PIE-NeRF Serves Up a New Slice: Physics-Based NeRFs

Just when we thought the week couldn't get more exciting in the world of physics-based radiance fields, we're introduced to another groundbreaking development: PIE-NeRF (Physics-based Interactive Elastodynamics with Neural Radiance Fields).

Just when we thought the week couldn't get more exciting in the world of physics-based radiance fields, we're introduced to another groundbreaking development: PIE-NeRF (Physics-based Interactive Elastodynamics with Neural Radiance Fields).

Michael Rubloff

Michael Rubloff

Michael is the Founder and Managing Editor of Radiancefields.com

PIE-NeRF

Just when we thought the week couldn’t get more exciting in the world of physics-based radiance fields, we’re introduced to another groundbreaking development: PIE-NeRF (Physics-based Interactive Elastodynamics with Neural Radiance Fields). Coming hot on the heels of the remarkable PhysGaussian paper, PIE-NeRF represents another leap forward, this time in the domain of Neural Radiance Fields (NeRFs).

PIE-NeRF emerges as a standout innovation, shifting the paradigm from traditional mesh-based structures to a pioneering meshless technique for simulating physical dynamics. This leap is particularly noteworthy, considering the intensive computational demands typically associated with 3D modeling.

It’s been a busy time for Yutao Feng, Xuan Li, Yin Yang, and Chenfanfu Jiang, all of whom are authors on both papers!

PIE-NeRF integrates physics-based simulations with the NeRF scene. This integration is akin to infusing the static scene with the laws of physics - gravity, force, and motion - allowing for the realistic simulation of dynamic properties. Whether it’s a swaying plant or a flowing cloth, PIE-NeRF calculates how these objects would move and interact in a real-world setting.

PIE-NeRF starts its journey by reimagining how we capture 3D scenes. Unlike conventional methods that rely on rigid mesh structures, NeRF uses a fluid, continuous field, encoded within a neural network, to understand a scene’s color, texture, and geometry from 2D images. This neural encoding of space lays the groundwork for the dynamic animations PIE-NeRF aims to achieve and is the canvas on which PIE-NeRF will paint its dynamic animations. Like many other papers, PIE-NeRF employs NVIDIA’s Instant NGP to build on top off.

Transitioning to 3D, PIE-NeRF transcends static models to foster interactive, dynamic scenes. Here, the innovation truly shines: the framework calculates physical properties and movements within NeRF’s continuous space, sidestepping the need for meshes. This meshless method is not just a technical novelty but a strategic advantage, enabling simulations to occur independently of sampling resolution, which dramatically lowers computational demands. PIE-NeRF enables physical properties and movements to be calculated within this continuous space, without the need for a predefined mesh.

Central to PIE-NeRF’s operation is Quadratic Generalized Moving Least Squares (Q-GMLS), the engine that powers the framework. This advanced method interprets and simulates physical behaviors like elasticity and deformation within NeRF’s space. Its ability to handle complex, non-linear deformations—where traditional linear methods stumble—marks a significant advancement, especially in avoiding ‘locking artifacts’, a common issue in linear simulations under extreme deformation scenarios.

In processing complex simulations, PIE-NeRF smartly employs spatial reduction via Voronoi partitioning. This approach is akin to strategically positioning sensors across a field, each overseeing a specific region. It ensures that computational resources are concentrated where dynamic interactions are most pronounced, thus fostering real-time interactions and simulations, without compromising the quality of the final output.

PIE-NeRF’s true prowess is displayed when it integrates physics-based simulations with NeRF scenes. This integration is like breathing life into static models, as the laws of physics—gravity, force, motion—are applied, enabling realistic simulation of dynamic properties. Whether it’s a swaying plant or flowing cloth, PIE-NeRF accurately predicts their real-world movements and interactions.

PIE-NeRF renders the dynamically altered scenes. Using deformation information, it precisely maps changes back to the NeRF model, ensuring that rendered images maintain high fidelity, even under significant deformations. Here, the interactive element comes into play: users can manipulate the scene, apply forces, and witness the immediate impact of their actions, all in real-time.

As PIE-NeRF continues to evolve, we eagerly await the release of its code and video overviews. These resources will undoubtedly provide deeper insights into its capabilities and applications. Powered by a 3090 GPU, PIE-NeRF is not just a testament to current technological advancements but a beacon for future innovations in digital representation.

[Editor’s Note: This article will be updated with additional resources and demonstrations of PIE-NeRF as they become available.]

Thumbnail for Gaussian Splatting News episode 13

13:12

Watch · New every Monday

The Biggest Week Yet for Gaussian Splatting: Blender + Unreal

Episode 13 · Nov 16, 2026

Blender 5.3 is out, and Gaussian splats are now a native part of it: PLY and SPZ files import as their own point cloud type and render in EEVEE, Cycles and Workbench, with no add-on.

Featured

Recents

Platforms

SuperSplat 3.9 Imports GSAF 4D Splats as Timeline Sequences

SuperSplat v3.9.0 imports GSAF 4D Gaussian splat files as timeline sequences, plays several sequences at once and adds a review table for multi-file imports.

Michael Rubloff

Nov 17, 2026

Platforms

SuperSplat 3.9 Imports GSAF 4D Splats as Timeline Sequences

SuperSplat v3.9.0 imports GSAF 4D Gaussian splat files as timeline sequences, plays several sequences at once and adds a review table for multi-file imports.

Michael Rubloff

Platforms

CARLA 1.0 Adds NVIDIA NuRec 26.04 Support

CARLA 1.0 adds NVIDIA NuRec 26.04 support, turning recorded real-world drives into simulated scenes with neural lidar, asset swapping and actor compositing.

Michael Rubloff

Nov 17, 2026

Platforms

CARLA 1.0 Adds NVIDIA NuRec 26.04 Support

CARLA 1.0 adds NVIDIA NuRec 26.04 support, turning recorded real-world drives into simulated scenes with neural lidar, asset swapping and actor compositing.

Michael Rubloff

News

Volumetric Media Town Hall Turns to 4D Gaussian Splats

The Metaverse Standards Forum holds its third Volumetric Media Town Hall on November 24, with 4D Gaussian splatting talks by Dymensium, Chronospace and Gracia.

Michael Rubloff

Nov 17, 2026

News

Volumetric Media Town Hall Turns to 4D Gaussian Splats

The Metaverse Standards Forum holds its third Volumetric Media Town Hall on November 24, with 4D Gaussian splatting talks by Dymensium, Chronospace and Gracia.

Michael Rubloff

Platforms

ReactVision Advanced VPS Turns Phone Scans Into Splats

ReactVision Advanced VPS in ViroReact 3.1.0 turns a one-minute phone scan into a map, mesh, GLB model, object map and Gaussian splat in one coordinate frame.

Michael Rubloff

Nov 16, 2026

Platforms

ReactVision Advanced VPS Turns Phone Scans Into Splats

ReactVision Advanced VPS in ViroReact 3.1.0 turns a one-minute phone scan into a map, mesh, GLB model, object map and Gaussian splat in one coordinate frame.

Michael Rubloff

Trending Articles
  1. TRENDING
    Loading...
  2. TRENDING
    Loading...
  3. TRENDING
    Loading...
Michael Rubloff

Written by Michael Rubloff

Michael is the Founder and Managing Editor of Radiancefields.com

More from Michael Rubloff

More from Michael Rubloff

Platforms

SuperSplat 3.9 Imports GSAF 4D Splats as Timeline Sequences

SuperSplat v3.9.0 imports GSAF 4D Gaussian splat files as timeline sequences, plays several sequences at once and adds a review table for multi-file imports.

Michael Rubloff

Nov 17, 2026

The SuperSplat editor in a browser window, editing a Gaussian splat of a snow-covered cabin in a pine forest

Platforms

SuperSplat 3.9 Imports GSAF 4D Splats as Timeline Sequences

Michael Rubloff

Nov 17, 2026

The SuperSplat editor in a browser window, editing a Gaussian splat of a snow-covered cabin in a pine forest

Platforms

SuperSplat 3.9 Imports GSAF 4D Splats as Timeline Sequences

SuperSplat v3.9.0 imports GSAF 4D Gaussian splat files as timeline sequences, plays several sequences at once and adds a review table for multi-file imports.

Michael Rubloff

Nov 17, 2026

The SuperSplat editor in a browser window, editing a Gaussian splat of a snow-covered cabin in a pine forest

Platforms

CARLA 1.0 Adds NVIDIA NuRec 26.04 Support

CARLA 1.0 adds NVIDIA NuRec 26.04 support, turning recorded real-world drives into simulated scenes with neural lidar, asset swapping and actor compositing.

Michael Rubloff

Nov 17, 2026

CARLA

Platforms

CARLA 1.0 Adds NVIDIA NuRec 26.04 Support

Michael Rubloff

Nov 17, 2026

CARLA

Platforms

CARLA 1.0 Adds NVIDIA NuRec 26.04 Support

CARLA 1.0 adds NVIDIA NuRec 26.04 support, turning recorded real-world drives into simulated scenes with neural lidar, asset swapping and actor compositing.

Michael Rubloff

Nov 17, 2026

CARLA

News

Volumetric Media Town Hall Turns to 4D Gaussian Splats

The Metaverse Standards Forum holds its third Volumetric Media Town Hall on November 24, with 4D Gaussian splatting talks by Dymensium, Chronospace and Gracia.

Michael Rubloff

Nov 17, 2026

Metaverse Standards Forum

News

Volumetric Media Town Hall Turns to 4D Gaussian Splats

Michael Rubloff

Nov 17, 2026

Metaverse Standards Forum

News

Volumetric Media Town Hall Turns to 4D Gaussian Splats

The Metaverse Standards Forum holds its third Volumetric Media Town Hall on November 24, with 4D Gaussian splatting talks by Dymensium, Chronospace and Gracia.

Michael Rubloff

Nov 17, 2026

Metaverse Standards Forum

New to Gaussian Splatting? Start here

Radiance fields, explained from the ground up: what a Gaussian splat is, how to capture one, and which tools and gear to start with.

Newsletter

Keep up with the field.

New research, tools and captures from the Radiance Fields newsroom, in your inbox. Free, no spam.

Gaussian Splatting News on YouTube

The week in radiance fields, every Monday.

Gaussian Splatting News on YouTube

The week in radiance fields, every Monday.

Gaussian Splatting News on YouTube

The week in radiance fields, every Monday.

Gaussian Splatting News on YouTube

The week in radiance fields, every Monday.