"Once and for all" is the hallucination of humans.
Every lesson demands repeated study.
No single honor serves you forever.
All eras must one day pass into memory.
The cycle of tearing down and rebuilding is worth the journey.

Qiu Weimin

Welcome to my homepage! I am a PhD student at UC Merced.
My research focuses on text-to-image generation. I am also interested in various topics as shown below.

monoidal categories, graphic calculus, hopf algebra, topological field theory.etc.

algebra, topology, manifold, measure theory, functional analysis, optimal transport.etc.

material mechanics, structural mechanics, structural dynamics, damage mechanics.etc.

If you think my work is interesting and worth further discussion, please send me emails.
Email: wqiu5@ucmerced.edu | Github: github

Publications

Self-Cross Diffusion Guidance for Image Synthesis of Similar Subjects

CVPR, 2025.

Link to Publication
Subject mixing is an unresolved issue for diffusion-based image synthesis, particularly for synthesizing multiple similar-looking subjects. We propose Self-Cross Diffusion Guidance to penalize the overlap between cross-attention maps and the aggregated self-attention map. Compared to previous methods based on self-attention or cross-attention alone, our guidance is training-free and more effective in eliminating subject mixing.

Resume

M.Sc./Ph.D. in Computer Science

University of California, Merced, USA | 2023-current

ILETS band 8, GRE 334+4, in 2022.

Engineer

State Grid Corporation of China, Shanghai, China | 2016 – 2022

Managed the construction of 35/110kV substations.

M.Eng. in Civil Engineering

Tongji University, Shanghai, China | 2013-2016

GEE 419, in 2013.

B.Eng. in Civil Engineering

Jiangxi University of Science and Technology, Ganzhou, China | 2008-2012