Cheng is undertook his PhD in Faculty of IT at Monash University. He received his Bachelor’s degree (2019) and Master’s degree (2022) at UESTC.

Cheng’s research interests lie in the intersection of Computer Vision, Machine Learning and Computer Graphics. From given inputs (e.g. random noise, labels, images, depth maps, semantic maps), his research seeks to generate conditional 3D models using generative adversarial networks. The aim is to speed up the design process, by generating 3D models from floor plans or sketches.

Before his PhD study, he worked on 3D scene understanding, 3D reconstruction and 3D detection.

PhD Start and End Dates
Cheng submitted his thesis for assessment in August 2026. He is returning to China to work at Vast.

PUBLICATIONS

• August 2026 – Attribute token arithmetic: Disentangled and continuous semantic control for visual autoregressive models
• August 2026 – CalibAnyView: Beyond single-view camera calibration in the wild
• December 2025 – ASAP-textured Gaussians: Enhancing textured Gaussians with adaptive sampling and anisotropic parameterization
• December 2025 – OpenView: Empowering MLLMs with Out-of-view VQA
• December 2025 – Unified camera positional encoding for controlled video generation
• November 2025 – NoPo3DFusion: Scaling two images to long videos via 3D-aware iterative diffusion
• November 2025 – PanFlow: Decoupled motion control for panoramic video generation
• March 2025 – Panorama generation from NFoV image done right
• March 2025 – PanSplat: 4K panorama synthesis with feed-forward Gaussian splatting
• April 2024 – Taming stable diffusion for text to 360o panorama image generation