Interactive and Controllable 3D Shape Detailization
Qimin Chen
PhD thesis, Simon Fraser University, 2026
pdf

This thesis introduces several feed-forward approaches for generating high-quality 3D shapes, which aim to address the challenges of structure-aware 3D content creation and provide interactive and controllable 3D modeling experience.

ART-DECO: Arbitrary Text Guidance for 3D Detailizer Construction
Qimin Chen, Yuezhi Yang, Wang Yifan, Vladimir G. Kim, Siddhartha Chaudhuri, Hao Zhang Zhiqin Chen
ACM SIGGRAPH Asia 2025 (Conference)
pdf / code / project page

ART-DECO is a 3D detailizer that instantly transforms coarse 3D shape proxies into high-quality, textured 3D assets guided by text prompts. Trained via SDS from MVDream, ART-DECO enables interactive modeling, style-consistent details, and creative structure control.

GenVDM: Generating Vector Displacement Maps From a Single Image
Yuezhi Yang, Qimin Chen, Vladimir G. Kim, Siddhartha Chaudhuri, Qixing Huang, Zhiqin Chen
CVPR 2025 (highlight)
pdf / code / project page

GenVDM is the first method for generating Vector Displacement Maps (VDMs): parameterized geometric stamps commonly used in 3D modeling. It generates multi-view normal maps from a single input image and then reconstructs a VDM via a novel reconstruction pipeline.

DECOLLAGE: 3D Detailization by Controllable, Localized, and Learned Geometry Enhancement
Qimin Chen, Zhiqin Chen, Vladimir G. Kim, Noam Aigerman, Hao Zhang, Siddhartha Chaudhuri
ECCV 2024
pdf / code / project page

DECOLLAGE is a learning-based method that enables novice users to add geometric details to a coarse 3D shape by selecting regions on it and assigning them the styles of exemplar shapes with compelling geometric details.

DAE-Net: Deforming Auto-Encoder for fine-grained shape co-segmentation
Zhiqin Chen, Qimin Chen, Hang Zhou, Hao Zhang
ACM SIGGRAPH 2024 (Conference)
pdf / code

DAE-Net is an unsupervised 3D shape co-segmentation method that learns a set of deformable part templates from a shape collection, which yields high-quality, consistent, and fine-grained 3D shape co-segmentation.

ShaDDR: Interactive Example-Based Geometry and Texture Generation via 3D Shape Detailization and Differentiable Rendering
Qimin Chen, Zhiqin Chen, Hang Zhou, Hao Zhang
ACM SIGGRAPH Asia 2023 (Conference)
pdf / code / project page

The first example-based deep generative neural network for generating a high-resolution textured 3D shape through geometry detailization and conditional texture generation applied to an input coarse voxel shape.

D²CSG: Unsupervised Learning of Compact CSG Trees with Dual Complements and Dropouts
Fenggen Yu, Qimin Chen, Maham Tanveer, Ali Mahdavi-Amiri, Hao Zhang
NeurIPS 2023
pdf

D2CSG is a neural model composed of two dual and complementary network branches, with dropouts, for unsupervised learning of compact constructive solid geometry (CSG) representations of 3D CAD shapes.

UNIST: Unpaired Neural Implicit Shape Translation Network
Qimin Chen, Johannes Merz, Aditya Sanghi, Hooman Shayani, Ali Mahdavi-Amiri, Hao Zhang
CVPR 2022
pdf / supplementary / code / project page

The first deep neural implicit model for general-purpose, unpaired shape-to-shape translation, in both 2D and 3D domains. UNIST can learn both style-preserving content alteration and content-preserving style transfer.

A New Deep Learning Engine for CoralNet
Qimin Chen, Oscar Beijbom, Stephen Chan, Jessica Bouwmeester, David Kriegman
ICCV Workshop on Computer Vision in the Ocean (ICCVW) 2021
pdf / bibtex / coralnet / code

CoralNet is a cloud-based website and platform for manual, semi-automatic and automatic analysis of coral reef images. Users access CoralNet through optimized web-based workflows for common tasks, other systems can interface through API.

Topology-Aware Single-Image 3D Shape Reconstruction
Qimin Chen, Vincent Nguyen, Feng Han, Raimondas Kiveris, Zhuowen Tu
CVPR Workshop on Learning 3D Generative Models (CVPRW) 2020
pdf / poster / bibtex

Composing volumetric-based generative model with topology-awareness auto-encoder allows them to learn high-level topological properties such as genus and connectivity for 3D shape reconstruction.