Triangle-to-Quadrilateral Conversion Operator

1. Introduction

I developed a high-performance triangle-to-quad-dominant conversion operator and it reaches SOTA effect with rich geometric filter control and globally post-merge consistent normals. Come and try it !!!

The technical details and source code will be released upon my SIGGRAPH 2026 paper accepted!

TODO

• Release Python API for researchers.
• Release Blender Addon for 3D artists.
  Upon my SIGGRAPH 2026 paper accepted,
• Release Technical details.
• Release Source Code.

Pic 1. Quality Comparison: Left: Self-developed operator; Middel: pymeshlab API; Right: GT Tri Mesh

My operator indicates much higher topological & geometrical quality comparing to pymeshlab/blender build-in operator (e.g. meshing_tri_to_quad_dominant(level=2), 2:'Better quad shape').

Pic 2. Normal Consistent: Left: Other operators; Right: Self-developed operator
(Red faces indicate inward-pointing normals, while white faces indicate outward-pointing normals.)

My operator indicates globally normal consistent comparing to some other tools.

2. Installation

conda create -n quad python=3.11 -y
conda activate quad 
pip install quad_converter-1.0.0-cp311-cp311-linux_x86_64.whl

3. Test

Instructions

# Test Python API
python -c "import quad_converter; print(quad_converter.__version__)"

# Test Command Line Tools, we provide high control of geometric attributes, check it here!
quad-convert --help

# Test conversion
# Basic Usage
quad-convert --input test.obj --output test_quad.obj --verbose

# Usage with parameter
quad-convert \
  --input model.obj \
  --output model_quad.obj \
  --angle_threshold 150 \
  --method blossom \
  --verbose

# test.py for tri-obj input and ouput quad-output
python test.py

Python API

import quad_converter
import numpy as np

# Prepare Mesh Data
vertices = np.array([...], dtype=np.float64)  # (N, 3)
triangles = np.array([...], dtype=np.int32)   # (M, 3)

# Convert to quad dominant mesh
result = quad_converter.convert_tri_to_quad(
    vertices=vertices,
    triangles=triangles,
    angle_threshold_deg=150.0,
    matching_method="blossom"
)

# Results
new_vertices = result["new_vertices"]
quad_faces = result["quads"]
tri_faces = result["tris"]
stats = result["merge_stats"]

print(f"Generate {len(quad_faces)} quads")
print(f"Left {len(tri_faces)} tris")

4. More Results:

Pic 3. Quality Comparison: Left: Self-developed operator; Middel: pymeshlab API; Right: GT Tri Mesh

Pic 4. More results from my operator to indicate great topological & geometrical quality.

Pic 5. More results from my operator to indicate globally normal consistent.

License

The license is based on the modified, 3-clause BSD-License.

An informal summary is: do whatever you want, but include this operator's license text within your product.