Splat2Real: Novel-view Scaling for Physical AI with 3D Gaussian Splatting

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Lim, Hansol, Choi, Jongseong Brad
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866912961348501504
author Lim, Hansol
Choi, Jongseong Brad
author_facet Lim, Hansol
Choi, Jongseong Brad
contents Physical AI faces viewpoint shift between training and deployment, and novel-view robustness is essential for monocular RGB-to-3D perception. We cast Real2Render2Real monocular depth pretraining as imitation-learning-style supervision from a digital twin oracle: a student depth network imitates expert metric depth/visibility rendered from a scene mesh, while 3DGS supplies scalable novel-view observations. We present Splat2Real, centered on novel-view scaling: performance depends more on which views are added than on raw view count. We introduce CN-Coverage, a coverage+novelty curriculum that greedily selects views by geometry gain and an extrapolation penalty, plus a quality-aware guardrail fallback for low-reliability teachers. Across 20 TUM RGB-D sequences with step-matched budgets (N=0 to 2000 additional rendered views, with N unique <= 500 and resampling for larger budgets), naive scaling is unstable; CN-Coverage mitigates worst-case regressions relative to Robot/Coverage policies, and GOL-Gated CN-Coverage provides the strongest medium-high-budget stability with the lowest high-novelty tail error. Downstream control-proxy results versus N provides embodied-relevance evidence by shifting safety/progress trade-offs under viewpoint shift.
format Preprint
id arxiv_https___arxiv_org_abs_2603_10638
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Splat2Real: Novel-view Scaling for Physical AI with 3D Gaussian Splatting
Lim, Hansol
Choi, Jongseong Brad
Computer Vision and Pattern Recognition
Physical AI faces viewpoint shift between training and deployment, and novel-view robustness is essential for monocular RGB-to-3D perception. We cast Real2Render2Real monocular depth pretraining as imitation-learning-style supervision from a digital twin oracle: a student depth network imitates expert metric depth/visibility rendered from a scene mesh, while 3DGS supplies scalable novel-view observations. We present Splat2Real, centered on novel-view scaling: performance depends more on which views are added than on raw view count. We introduce CN-Coverage, a coverage+novelty curriculum that greedily selects views by geometry gain and an extrapolation penalty, plus a quality-aware guardrail fallback for low-reliability teachers. Across 20 TUM RGB-D sequences with step-matched budgets (N=0 to 2000 additional rendered views, with N unique <= 500 and resampling for larger budgets), naive scaling is unstable; CN-Coverage mitigates worst-case regressions relative to Robot/Coverage policies, and GOL-Gated CN-Coverage provides the strongest medium-high-budget stability with the lowest high-novelty tail error. Downstream control-proxy results versus N provides embodied-relevance evidence by shifting safety/progress trade-offs under viewpoint shift.
title Splat2Real: Novel-view Scaling for Physical AI with 3D Gaussian Splatting
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2603.10638