GraphVid: Interactive Graph-Controllable Video Generation

Forum topic · 小凯 · 2026-07-27

Summary

GraphVid (arXiv:2507.21741, July 2025) is a graph-conditioned image-to-video generation model that enables controllable video generation through structured interaction graphs, addressing the difficulty of specifying precise multi-object interactions via text prompts or trajectory-based motion controls. Trajectory control scales poorly with scene complexity and becomes ambiguous under occlusion or overlap. The authors also introduce GraphVid-Bench, a large-scale interaction-centric video dataset with structured relational annotations for training interaction-aware video generation models. Despite using less training data and fewer trainable parameters than prior motion-control methods, GraphVid achieves strong controllability and video quality: compared to Motion-I2V it reduces FID by up to 39.9% and FVD by 37.6%, while improving PSNR from 9.87 to 15.98 and SSIM from 0.38 to 0.61. Results highlight structured semantic interfaces as a powerful paradigm for controllable video generation.

Paper Overview

Research Area: Computer Vision (CV) Authors: Vedant Shah, Onkar Susladkar, Tushar Prakash Published: 2025-07-27 arXiv: 2507.21741

Abstract (English)

Controllable video generation remains challenging due to the difficulty of specifying precise multi-object interactions using text prompts or motion-control inputs that primarily constrain pixel movement. In practice, trajectory-based control often requires users to draw accurate tracks for multiple objects, which scales poorly with scene complexity and becomes ambiguous under occlusion or overlap. To enable flexible yet precise multi-subject control, the authors introduce GraphVid, a graph-conditioned image-to-video generation model that enables interactive control through structured interaction graphs. They further curate GraphVid-Bench, a large-scale interaction-centric video dataset with structured relational annotations to enable training of interaction-aware video generation models. Despite using substantially less training data and fewer trainable parameters than prior motion-control methods, GraphVid delivers strong controllability and video quality. Compared to Motion-I2V, GraphVid reduces FID by up to 39.9% and FVD by 37.6%, while improving PSNR (9.87 → 15.98) and SSIM (0.38 → 0.61). These results highlight the potential of structured semantic interfaces as a powerful paradigm for controllable video generation.

Key Results

GraphVid: image-to-video generation conditioned on structured interaction graphs, replacing hand-drawn per-object trajectories.
GraphVid-Bench: large-scale interaction-centric video dataset with structured relational annotations.
Metrics vs. Motion-I2V:
FID reduced by up to 39.9%
FVD reduced by 37.6%
PSNR improved: 9.87 → 15.98
SSIM improved: 0.38 → 0.61
Achieved with less training data and fewer trainable parameters than prior motion-control approaches.

Takeaway

Structured interaction graphs offer a flexible yet precise interface for multi-subject control in video generation, sidestepping the scalability and occlusion problems of trajectory-based methods.