GraphVid: Interactive Graph-Controllable Video Generation

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: Vedant Shah, Onkar Susladkar, Tushar Prakash, Kiet Nguyen, Tianjio Yu, Adheesh Juvekar, Muntasir Waheed, Ismini Lourentzou
• Venue: arXiv: cs.CV
• Published: 2026-07-23
• Collected: 2026-07-24
• URL: http://arxiv.org/abs/2607.21580v1
• PDF: https://arxiv.org/pdf/2607.21580v1
• DOI: UNKNOWN

Plain-Language Brief

The paper introduces GraphVid, a graph-conditioned image-to-video generation model that allows for interactive control through structured interaction graphs. The authors claim that GraphVid delivers strong controllability and video quality compared to Motion-I2V, with significant improvements in FID, FVD, PSNR, and SSIM. They also created a large-scale interaction-centric video dataset called GraphVid-Bench for training interaction-aware video generation models.

What the Document Says

• GraphVid is a graph-conditioned image-to-video generation model introduced for interactive control through structured interaction graphs.
• The authors created GraphVid-Bench, a large-scale interaction-centric video dataset with structured relational annotations.
• According to the authors, GraphVid delivers strong controllability and video quality compared to Motion-I2V.
• The improvements claimed by the authors are: FID reduced by up to 39.9%, FVD reduced by 37.6%, PSNR increased from 9.87 to 15.98, and SSIM increased from 0.38 to 0.61.

Key Concepts Mentioned

• GraphVid
• Interactive control
• Graph-conditioned image-to-video generation
• Structured interaction graphs
• GraphVid-Bench
• FID
• FVD
• PSNR
• SSIM

Claims Made by the Document

• GraphVid delivers strong controllability and video quality.
• GraphVid improves FID, FVD, PSNR, and SSIM compared to Motion-I2V.

Evidence Used

• Improvements in FID, FVD, PSNR, and SSIM are stated in the text.

Methods / Approach

• Graph-conditioned image-to-video generation with structured interaction graphs.
• Training on GraphVid-Bench

Limitations Stated by the Authors

• Using substantially less training data and fewer trainable parameters than prior motion-control methods.

Limitations Noted by the Agent

Agent interpretation — not from the source.

The paper does not mention any limitations of the method itself, only the use of less training data and fewer parameters.

Open Questions

• What is the specific algorithm or architecture used in GraphVid?

Notes for Future Q&A

May include interpretation. Labeled where it does.

• The text does not provide details about the specific algorithm or architecture used in GraphVid.

Raw Abstract / Excerpt

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, we introduce $\textbf{GraphVid}$, a graph-conditioned image-to-video generation model that enables interactive control through structured interaction graphs. We further curate $\textbf{GraphVid-Bench}$, 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 with 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). Our results highlight the potential of structured semantic interfaces as a powerful paradigm for controllable video generation.