Fidelity‐preserved Visual Proxy Construction using Architectural Semantics

Abstract City‐scale scenes are increasingly prevalent in games and visualizations, making efficient rendering of mid‐ and far‐range buildings crucial. It is desirable to have low‐polygon‐count proxies, connected bounding polygons derived from the originals. However, such proxies are still typically created manually, which is time and labor intensive. Architects frequently conceptualize the design as an assembly of primitives, known as building mass; for example, office buildings are often assembled with cuboids. Simultaneously, the volumetric faces and corner edges, known as architectural features, encode the physical boundaries and dominate the perceived shape. However, polygon simplification, point cloud reconstruction, and plane fitting primarily focus on local geometry while view‐based clamping approximates the mass with selected silhouettes derived from geometric aspects. None considers architectural semantics resulting in failing to preserve architectural features and inducing perceptual distortion or excessive polygon counts. Therefore, we derive an automatic proxy construction framework from architectural semantics. Due to gravity, architects vertically align floors' functional sectors and accessories, such as elevator/balcony shafts, to form the volumetric basis, denoted as roof‐top volumes, of the building mass while each has its own separate top. In other words, the assembly of these roof‐top volumes can be a good approximation. Thus, we identify functional sections via top‐down ray casting and spatial segmentation with locations and orientations, fit their boundaries and heights with primitives, extrude them to form roof‐top volumes and refine their sides laterally with a similar mechanism for mass approximation. Finally, quantitative comparisons against three baselines on 40 buildings under similar face budgets along with a 25‐ participant user study demonstrate that ours has higher perceptual fidelity, better feature preservation, and competitive efficiency.

Authors

Institutions

Publication Details

Journal
Computer Graphics Forum
Published
2026-11-13
DOI
https://doi.org/10.1111/cgf.70608
Primary Topic
Computer Graphics and Visualization Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Fidelity‐preserved Visual Proxy Construction using Architectural Semantics

Zhongqi Cai, Wen‐Kai Tai, Yu‐Chi Lai, Shih-Syun Lin et al.
2 links
Computer Graphics Forum
Computer Graphics and Visualization Techniques
article

Fidelity‐preserved Visual Proxy Construction using Architectural Semantics

Zhongqi Cai, Wen‐Kai Tai, Yu‐Chi Lai, Shih-Syun Lin, Yan‐Lin Chen, Jiang Bo‐Yuan
article en

Abstract

Abstract City‐scale scenes are increasingly prevalent in games and visualizations, making efficient rendering of mid‐ and far‐range buildings crucial. It is desirable to have low‐polygon‐count proxies, connected bounding polygons derived from the originals. However, such proxies are still typically created manually, which is time and labor intensive. Architects frequently conceptualize the design as an assembly of primitives, known as building mass; for example, office buildings are often assembled with cuboids. Simultaneously, the volumetric faces and corner edges, known as architectural features, encode the physical boundaries and dominate the perceived shape. However, polygon simplification, point cloud reconstruction, and plane fitting primarily focus on local geometry while view‐based clamping approximates the mass with selected silhouettes derived from geometric aspects. None considers architectural semantics resulting in failing to preserve architectural features and inducing perceptual distortion or excessive polygon counts. Therefore, we derive an automatic proxy construction framework from architectural semantics. Due to gravity, architects vertically align floors' functional sectors and accessories, such as elevator/balcony shafts, to form the volumetric basis, denoted as roof‐top volumes, of the building mass while each has its own separate top. In other words, the assembly of these roof‐top volumes can be a good approximation. Thus, we identify functional sections via top‐down ray casting and spatial segmentation with locations and orientations, fit their boundaries and heights with primitives, extrude them to form roof‐top volumes and refine their sides laterally with a similar mechanism for mass approximation. Finally, quantitative comparisons against three baselines on 40 buildings under similar face budgets along with a 25‐ participant user study demonstrate that ours has higher perceptual fidelity, better feature preservation, and competitive efficiency.

Computer Graphics Forum
National Taiwan University of Science and Technology (TW)
Sustainable cities and communities
Computer Graphics and Visualization Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon