Ruixiang Jiang / PhD Candidate

Computational aesthetics · generative artcomputer vision · computer graphics

I study how machines understand and create beauty.

Ruixiang Jiang

PhD candidate at The Hong Kong Polytechnic University, supervised by Prof. Chang Wen Chen

Scroll to publications

Research archive · 2023–2026

Publications

A muse, camera and sun symbolize coordinated pose, viewpoint and lighting.

Before the Shutter

Aesthetic and Actionable Portrait Photography Planning in 3D Scenes

An agentic system for planning human pose, camera, and lighting for portrait photography in a 3D environment.

arXiv 2026 · Ruixiang Jiang, Chang Wen Chen

arXivGitHub
A Delphic oracle interprets a painting of Icarus, with a fallen feather and wax drops in the foreground.

SemJudge

On Semiotic-Grounded Interpretive Evaluation of Generative Art

A semiotics-theoretic framework for evaluating art interpretation through Peirce's semiosis.

ACM MM 2026 · Ruixiang Jiang, Chang Wen Chen

arXivGitHub
Hephaestus chisels a statue's form from marble while winged Iris paints ornament onto the same figure.

DiffArtist

Towards Structure and Appearance Controllable Image Stylization

A neural stylization method with disentangled control over structure and appearance

ACM MM 2025 · Ruixiang Jiang, Chang Wen Chen

PaperGitHubProject
Ariadne guides a thread from computation toward the visible artwork.

ArtCoT

Multimodal LLMs Can Reason about Aesthetics in Zero-Shot

A reasoning framework that interprets visual aesthetics beyond surface-level appeal.

ACM MM 2025 · Ruixiang Jiang, Chang Wen Chen

PaperGitHubData
An image guides a selector among theatrical experts, inserting a small prompt ribbon into an intact book.

MoPE

Mixture of Prompt Experts for Parameter-Efficient and Scalable Multimodal Fusion

Prompt tuning for multimodal fusion that preserves task-specific cues through selective conditioning.

TMM 2025 · Ruixiang Jiang, Lingbo Liu, Chang Wen Chen

GitHubarXiv
Hermes selects pomegranates through a scroll; five marked fruits are distinguished from two pears.

CLIP-Count

Towards Text-Guided Zero-Shot Object Counting

Open-vocabulary object counting from arbitrary text prompts, without category-specific training.

ACM MM 2023 · Ruixiang Jiang, Lingbo Liu, Chang Wen Chen

GitHubarXivPaper
A scroll guides the creation of a robed figure whose surface becomes a neural lattice, alongside a poseable mannequin.

AvatarCraft

Transforming Text into Neural Human Avatars with Parameterized Shape and Pose Control

A text-to-avatar pipeline with explicit control over neural human shape and pose.

ICCV 2023 · Ruixiang Jiang et al.

ProjectGitHubarXivPaper
Proteus reshapes and paints a Pegasus sculpture; a mirror shows the transformed form from another viewpoint.

NeRF-Art

Text-Driven Neural Radiance Fields Stylization

Text-driven stylization for neural radiance fields, with flexible artistic control over 3D objects.

TVCG 2023 · Can Wang, Ruixiang Jiang et al.

ProjectGitHubarXivPaper