DreamCoser: Controllable layered 3D character generation and editing

Wang, Yi, Ma, Jian, Su, Zhuo, Wang, Guidong, Lai, Yu Kun ORCID: https://orcid.org/0000-0002-2094-5680 and Li, Kun 2025. DreamCoser: Controllable layered 3D character generation and editing. Presented at: SA Technical Communications '25, Hong Kong, 15-18 December 2025. Proceedings of the SIGGRAPH Asia 2025 Technical Communications. New York, NY: ACM, 10.1145/3757376.3771404 Item availability restricted.

PDF - Accepted Post-Print Version Restricted to Repository staff only until 15 January 2026 due to copyright restrictions. Download (2MB)

Abstract

This paper aims to controllably generate and edit layered 3D characters based on hand-drawing. Existing methods rely on global optimization or entangled representations, limiting fine-grained local editing and clothing replacement. To address this, we propose an innovative sketch-based method for layered 3D character generation and part-level editing. Our approach introduces a sketch-to-3D decoupled generation network for fine-grained layered control and a progressive upsampling module that enhances texture quality and complex geometric structures. Extensive experiments on public datasets and in-the-wild data demonstrate our method effectively generates high-quality layered 3D characters while supporting precise local editing through hand-drawn sketches. The code will be available at http://cic.tju.edu.cn/faculty/likun/projects/DreamCoser.

Item Type:	Conference or Workshop Item (Paper)
Date Type:	Published Online
Status:	Published
Schools:	Schools > Computer Science & Informatics
Publisher:	ACM
ISBN:	979-8-4007-2136-6
Date of First Compliant Deposit:	14 December 2025
Date of Acceptance:	6 October 2025
Last Modified:	15 Dec 2025 15:30
URI:	https://orca.cardiff.ac.uk/id/eprint/183215

Actions (repository staff only)

Edit Item