Skip to content

Latest commit

 

History

History
51 lines (39 loc) · 2.4 KB

README.md

File metadata and controls

51 lines (39 loc) · 2.4 KB
Raw

Architectural Sketch To 3D Printing

Take a single architectural sketch as input and generate a ready-to-print 3D model.

Dataset Preparation

Simple House Dataset Generator

  • simple_house_generator.gh: A grasshopper script for generating the "Simple House" dataset. After running this script, you can use postprocess_sketch.py and gen_voxel_from_obj.py to convert the sketch into the same style as the shapenet-synthetic dataset from sketch2model and generate the corresponding voxel.

  • For training the model and evaluation, please refer to sketch2model.

Simple House Dataset

In this project, we proposed the Simple House Dataset. We define 5 categories of houses – distinguished by roof type and other geometric properties – and generated 1000 models for each category with parameters chosen randomly. Each model includes a mesh and 20 perspectival line drawings taken from different angles, where:

  • 0 ~ 1000: hip-flat (mansard)
  • 1001 ~ 2000: hip
  • 2001 ~ 3000: pyramid
  • 3001 ~ 4000: random roof (including M-shape, butterfly, gambrel)
  • 4000 ~ 5000: random shape

The dataset can be downloaded from TBD.

Mesh Generation

Here's a demostration of the mesh deformation approach.

Hip Mansard flat

Mesh Optimization

PolyFit is used here to perform the mesh optimization.

Demo Website

A demo website can be found in /demo. You can launch the sever with:

cd demo
bash ./run_flask.sh

Utilities scripts

  • angle_evaluation.py: A grasshopper script which takes azimuth and elevation as input and orient Rhino's view.
  • object_baker.gh: Dynamically bake a object generated from grasshopper script.
  • photo_taker_single_obj.gh: A grasshopper script that can take photo of 3D model in specific angle.

References

  1. Building Dataset Generator: https://github.com/CDInstitute/Building-Dataset-Generator
  2. sketch2model: https://github.com/bennyguo/sketch2model