3D Model Editing from Contour Drawings on Orthographic Projection Views

Yuhui Hu, Xuliang Guo, Baoquan Zhao, Shujin Lin, Xiaonan LuoICISP 2014 · OpenReview

Abstract

3D modeling and mesh editing for objects from orthographic projections of engineering drawings have become common and convenient. Designers are accustomed to using simple sketching to express visualized thoughts. This paper presents a method for controlling the deformation of surface meshes by sketching on orthographic projections. The system allows users to select a region of interest (ROI) to constrain the deformation area and then sketch the desired contour. For a given new sketch, the system automatically deforms the 3D model using Laplacian deformation, fitting the model's contours to the user's sketch. Various examples validate the effectiveness of this approach as an intuitive method for 3D model editing.

Introduction

In traditional CAD and engineering workflows, designers frequently communicate ideas through 2D orthographic projections—front, top, and side views. However, editing existing 3D meshes often involves complex tools, vertex manipulation, or dense control cages that can be unintuitive for those used to 2D drafting.

This research bridges the gap between 2D drafting and 3D mesh editing by allowing users to draw 2D strokes directly over flat projections. The primary challenge addressed is translating a single 2D stroke into a smooth, detail-preserving 3D deformation that respects the original geometry while achieving the desired silhouette change.

Methodology

The proposed pipeline consists of four primary stages designed to translate 2D intent into 3D geometry:

  1. Orthographic Projection & ROI Selection: The 3D mesh is projected onto a 2D plane (xy, xz, or yz). The user defines a rectangular Region of Interest (ROI) to isolate the specific area of the model intended for deformation.
  2. Sketching the Contour: Within the ROI, the user draws a target contour using freehand strokes or uniform quadratic B-splines.
  3. Determining Correspondences: The system maps the dense 2D sketch to the 3D mesh vertices by sampling feature points along the sketch based on a distance threshold ($d$). It identifies the nearest projected 3D vertices within the ROI to serve as corresponding constraint points.
  4. Laplacian Deformation: To preserve local geometric features like surface texture or muscle definition, the system employs Laplacian coordinates. It solves a linear system where junction points between the ROI and the rest of the mesh act as fixed boundary constraints, while mapped vertices from the sketch act as moving constraints. This ensures the global shape matches the sketch while local differential properties are preserved.
The proposed editing pipeline: (a) original model, (b) orthographic projections, (c-e) ROI selection and contour sketching, (f) the final deformed 3D mesh.

Key Insights and Implementation

A critical parameter in this process is the distance threshold ($d$) used for sampling the 2D sketch. This threshold must align with the local grid density of the mesh; if $d$ is too large, the mesh fails to fit the new contour accurately, leading to a loss of shape fidelity.

Furthermore, the use of Laplacian coordinates allows the algorithm to treat the mesh like a flexible sheet. While the global shape stretches to meet the 2D sketch, the local relationships between adjacent vertices—the "details"—are preserved in a least-squares sense. The ROI boundary serves as an implicit blending mechanism, ensuring smooth transitions into the unedited parts of the model without requiring manual weight painting or transition zone definition.

Results and Discussion

The method was evaluated qualitatively using various animal models, such as deer and canines. As shown in the results, users can alter complex postures, such as a deer's neck or a wolf's stance, with simple profile lines. The system successfully updates the 3D mesh to match the new silhouette while maintaining surface smoothness and preserving unedited regions.

Orthographic projection views (front, top, side) of a 3D wolf model, providing the necessary multi-view context for sketch-based editing.

While the visual results are promising, performance bottlenecks currently exist. Calculating Laplacian deformations and refreshing projection views in real-time remains a challenge for very dense meshes, which is a key area for optimization.

Conclusion and Future Work

This paper demonstrates an intuitive approach to 3D mesh editing via 2D sketching on orthographic views. Future research will focus on improving computational efficiency for real-time interaction and incorporating depth and occlusion context from user sketches to handle more complex geometry. Additionally, extending the technique to allow direct sketching on 3D surfaces in perspective views could further enhance the flexibility of the designer's workflow.

Summary updated

Abstract

3D modeling and mesh editing for objects from orthographic projections of engineering drawing has been very common and convenient. People has been accustomed to using such simple and fast sketching way to express their visualized thought. In this paper we present a method for model designers by sketching on the orthographic projection to control the deformation of surface meshes. Our system lets the user easily choose the region of interest for constraining the deformation area, and then sketch the desired contour. For a given new sketch on the orthographic projection,the system automatically deform the 3D model by using Laplacian deformation, which makes the contours of 3D model fit the contour line of the users sketching. Various examples have validated the effectiveness of our proposed method, which can be regarded as a effective method in 3D model editing.
BibTeX
@inproceedings{hu2014d,
  author = {Yuhui Hu and Xuliang Guo and Baoquan Zhao and Shujin Lin and Xiaonan Luo},
  title = {{3D Model Editing from Contour Drawings on Orthographic Projection Views}},
  booktitle = {ICISP},
  year = {2014},
  url = {https://openreview.net/forum?id=4XwJGkhtgG},
}
Figure