Cost-Efficient Volumetric Capture: From Neural Rendering to Gaussian Splatting Rig - 25.45
Title:
Cost-Efficient Volumetric Capture: From Neural Rendering to Gaussian Splatting Rig
Authors:
Sergei ELISEEV 1, Georgii MOLODTSOV 2
1 Yandex, Almaty, Kazakhstan;
2 Yandex, Belgrade, Serbia
Keywords:
Volumetric capture, 3D Gaussian splatting, NeRF, mesh retargeting, low-cost scanning, portable rig
Abstract:
Volumetric content enables realistic, spatial storytelling across VR, AR, and cinematic applications. We present two complementary technologies that democratize high-quality 3D human capture using affordable hardware and novel software pipelines.
Our first approach replaces expensive volumetric studio rigs with a portable setup based on
commodity cameras, software synchronization, neural rendering, and dynamic background modeling.
Using neural implicit representations (e.g., NeRF and SDFs) and body pose priors, we achieve
smooth geometry and fine textures without green screens or hard camera sync. We refine textures
via differentiable rendering and correct camera pose drift through mid-sequence calibration, enabling clean reconstructions even in uncontrolled environments.
Building on this vision of accessibility, our second approach introduces a highly mobile Gaussian
Splatting Rig. The rig features 8 GoPro cameras mounted on a rotating handheld structure that can
be moved freely around the subject. This configuration enables high-fidelity, full-body scans by
capturing dense multi-view video data from all angles. The footage is processed using state-of-the-art 3D Gaussian Splatting algorithms to create photorealistic volumetric representations. These splat-based models exhibit exceptional surface detail, lighting consistency, and view-dependent effects.
To increase downstream utility, we convert the resulting point-based splats into watertight meshes
with minimal quality loss. This enables compatibility with standard animation pipelines. We further
apply off-the-shelf retargeting methods to map motion to the avatar mesh, making it suitable for game engines or real-time avatars in XR applications.
Together, these two methods offer a practical continuum: a garage-built studio setup with neural
pipelines for multi-person scenes, and a compact scanning rig for single-person high-fidelity avatars. Both approaches significantly reduce production costs while maintaining or exceeding visual quality compared to traditional systems. We believe these tools will empower creators, researchers, and developers to bring volumetric capture to new contexts and audiences.
Abstract:
Presentation:
VIDEO will be available here in Q3.2026.
VIDEO availble in proceedings (purchase order)
How to Cite (MLA):
S. Eliseev and G. Molodtsov, "Cost-Efficient Volumetric Capture: From Neural Rendering to Gaussian Splatting Rig", Proceedings of 3DBODY.TECH 2025 - 16th International Conference and Expo on 3D/4D Body Scanning, Data and Processing Technologies, Lugano, Switzerland, 21-22 Oct. 2025, #45
Details:
Proceedings: 3DBODY.TECH 2025, 21-22 Oct. 2025, Lugano, Switzerland
Paper/Presentation: #45
DOI: -
License/Copyright notice
Proceedings: © Hometrica Consulting - Dr. Nicola D'Apuzzo, Switzerland, hometrica.ch.
Authors retain all rights to individual papers, which are licensed under a Creative Commons Attribution 4.0 International License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
The papers appearing in the proceedings reflect the author's opinions. Their inclusion in the proceedings does not necessary constitute endorsement by the editor or by the publisher.
Note: click the + on the top left of the page to open/close the menu.