figshare
Browse
wumengjz_mscd_architecture_2018_r.pdf (13.28 MB)

Make Models Great Again: Interactively generating simplified editable models from 3D-scanned dense meshes

Download (13.28 MB)
thesis
posted on 2018-05-10, 00:00 authored by Wumengjian ZhuWumengjian Zhu
3D scanning is a technique that collects data from real-world objects to construct digital three-dimensional models. Following various cutting-edge applications in manufacturing and various industries, this reverse-engineering technique was creatively and famously adopted by architect Frank Gehry to digitize physical models to facilitate his further design.
Although integrating these techniques into the earlier phases of design has been a tempting topic in design practice, achieving it is not easy. In addition to generating a mesh using either professional scanners or easy-to-access photogrammetry software, the generated mesh must be simplified into a editable form that is friendly to human designers. While the first part is a relatively mature process, current methods for mesh simplification are either difficult to control during simplification, or generate hard-to-manipulate triangle meshes.
The purpose of this thesis is to present the world of design with a newly developed 3D reconstruction oriented design workflow. It’s core contribution is implementing, optimizing, and evaluating a new interactive mesh simplification workflow based on Variational Shape Approximation, which generates easier-to-control designer-friendly simplifications. High performance parallel computing techniques (such as CUDA) were explored and used to improve efficiency of this algorithm. A new modelling plug-in, iSimp for Autodesk Maya, is developed and evaluated by a group of designers in a workshop. It is found that the proposed mesh simplification workflow will greatly facilitate model digitization for creative designers, and more importantly, inspire new approaches in designing from physical world. Digital models, which are once great in their physical forms, will be made great again.

History

Date

2018-05-10

Degree Type

  • Master's Thesis

Department

  • Architecture

Degree Name

  • Master of Science in Computational Design (MSCD)

Advisor(s)

Daniel Cardoso Llach

Usage metrics

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC