Jiaheng LiuNANJING UNIVERSITY · LINK LAB
7 research contributions
Audio-video generation and editing, controllable content, procedural 3D modeling, and interactive world models.
Search by name or capability, or narrow by publication status and authorship.
Evaluates text-to-audio-video generation through a unified framework.
T2AV-Compass: Towards Unified Evaluation for Text-to-Audio-Video Generation
* Equal contribution; # Corresponding author.
Evaluates generated worlds for state consistency and responsiveness across video, spatial, and embodied tracks.
HappyWorld-Bench
* Equal contribution; # Corresponding author.
Provides a companion evaluation set for multi-part procedural 3D modeling.
Procedura: Agentic 3D Modeling with Procedural Control
* Equal contribution; # Corresponding author.
Evaluates instruction-driven audio-video editing abilities.
AVE-Compass: Towards Holistic Evaluation for Audio-Video Editing Abilities
* Equal contribution; # Corresponding author.
Tests parametric 3D generation and structural reasoning in multimodal models.
P3D-Bench: Benchmarking MLLMs for Parametric 3D Generation and Structural Reasoning
* Equal contribution; # Corresponding author.
Tests video editing under complex, compositional instructions.
CoVEBench: Can Video Editing Models Handle Complex Instructions?
* Equal contribution; # Corresponding author.
Evaluates video captions used to support controllable text-to-video generation.
VidCapBench: A Comprehensive Benchmark of Video Captioning for Controllable Text-to-Video Generation
* Equal contribution; # Corresponding author.
Try a broader keyword or clear the filters.
This collection includes co-authored benchmarks, companion evaluation datasets, and evaluation methods. Each work has one primary category; topic tags capture related capabilities. Corresponding-author labels refer to Jiaheng Liu.