PPISP-Local: Local Tone Mapping for Physically Plausible 3D Gaussian Splatting of Heritage Ceilings

Wong, Anthony and Wang, Cheng ORCID: https://orcid.org/0000-0003-4711-4534 (2026) PPISP-Local: Local Tone Mapping for Physically Plausible 3D Gaussian Splatting of Heritage Ceilings. In: London Imaging Meeting (LIM) 2026. UNSPECIFIED. (In Press)

[thumbnail of LIM26_PPISP_Local_CamReady] PDF (LIM26_PPISP_Local_CamReady) - Accepted Version
Restricted to Repository staff only until 31 December 2099.
Available under License Creative Commons Attribution.

Request a copy

Abstract

Mobile-phone-based 3D Gaussian Splatting (3DGS) is increasingly attractive for heritage digitization because it supports rapid, non-invasive acquisition in difficult environments. However, large indoor heritage scenes such as decorated ceilings often exhibit strongly unbalanced illumination, with side-window lighting, long capture trajectories, and automatic mobile camera processing causing view-dependent appearance inconsistencies. These effects are only partially handled by existing global photometric correction methods. We present PPISP-Local, a compact extension of a physically plausible image signal processing (ISP) correction model for radiance-field reconstruction. The method retains global exposure, vignetting, colour correction, and camera response modelling, and adds a low-capacity, luminance-only local tone-mapping module to better approximate mobile-phone ISP behaviour in unevenly lit indoor scenes. We evaluate the method on a high-ceiling heritage dataset and compare vanilla 3DGS, PPISP, and PPISP-Local on held-out ceiling views. PPISP-Local achieves the best reconstruction quality on this challenging smartphone dataset, suggesting that weak local tone adaptation is beneficial for large indoor heritage reconstruction when global photometric models are insufficient.

Item Type: Book Section
Uncontrolled Keywords: das to be completed,2026 ,/dt/atira/pure/keyword/DAStatement/das_to_be_completed
Faculty \ School: Faculty of Science > School of Computing Sciences
Depositing User: LivePure Connector
Date Deposited: 06 Nov 2026 10:40
Last Modified: 06 Nov 2026 10:40
URI: https://ueaeprints.uea.ac.uk/id/eprint/104991
DOI:

Actions (login required)

View Item View Item