Leum-VL Technical Report
A short video succeeds not simply because of what it shows, but because of how it schedules attention -- yet current multimodal models lack the structural grammar to parse or produce this organization. Existing models can describe scenes, answer event-centric questions, and read on-screen text, but they are far less reliable at identifying timeline-grounded units such as hooks, cut rationales, shot-induced tension, and platform-facing packaging cues. We propose SV6D (Structured Video in Six Dimensions), inspired by professional storyboard practice in film and television production, a representation framework that decomposes internet-native video into six complementary structural dimension...
| Authors | Yuxuan He, Chaiming Huang, Yifan Wu et al. |
|---|---|
| Primary category | cs.MM |
| Published | 20 March 2026 |
| Last updated at source | 20 March 2026 |
Connected in the knowledge graph
- leum-team/Leum-VL-8B-preview0320Model · cited by
Models and datasets referencing this paper
Silytics shows public metadata from arXiv and links to the original. Information may be incomplete or out of date. Last refreshed 17 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.