Zhiming Hong
Postgraduate Research Student
Academic and research departments
School of Computer Science and Electronic Engineering, Centre for Vision, Speech and Signal Processing (CVSSP).About
My research project
Real-Time Interactive Generative AI: Leveraging Temporal Dynamics for Intent-Aware Image SynthesisSketching is humanity's most intuitive form of visual expression, yet current generative models treat drawings as static constraints, ignoring the rich temporal information embedded in the creation process. This project seeks to bridge the gap between amateur human input and high-fidelity AI output by developing next-generation diffusion models that understand the evolution of a sketch. It aims to pioneer methods to interpret drawing sequences -- from initial abstract strokes to detailed completion -- to better align generated imagery with human aesthetic intent.
Sketching is humanity's most intuitive form of visual expression, yet current generative models treat drawings as static constraints, ignoring the rich temporal information embedded in the creation process. This project seeks to bridge the gap between amateur human input and high-fidelity AI output by developing next-generation diffusion models that understand the evolution of a sketch. It aims to pioneer methods to interpret drawing sequences -- from initial abstract strokes to detailed completion -- to better align generated imagery with human aesthetic intent.