Skip to content
aiqbench takeAI Products

aiqbench analysis

This research introduces "Gen Through the Lens," a novel framework for camera-controllable image-to-video generation that unifies control signals across camera, world, and object coordinate spaces to ensure physical consistency. By explicitly fusing these heterogeneous cues, the method achieves superior realism and editability in diverse camera operations compared to prior abstract conditioning approaches. The work advances the field of generative video synthesis by solving the critical challenge of maintaining temporal and spatial coherence under precise motion constraints.
Source coveringarXiv
What this isA model's reading of the article and related coverage, written for aiqbench. Not the publisher's words.

Gen Through the Lens: Camera-Controllable Video Generation via Cross-Coordinate Conditioning Fusion

arXiv2 min read

arXiv:2610.15900v1 Announce Type: cross Abstract: We present Gen Through the Lens, a camera-controllable image-to-video generation framework that synthesizes videos from a given reference image while enabling realistic and physically consistent camera operations. Unlike previous models that treat camera motion as abstract conditioning or rely on a single coordinate system, our approach explicitly decomposes and fuses control signals across three coordinate spaces: the camera coordinate (modeled by intrinsic parameters), the world coordinate (represented by spatial geometry hints), and the object coordinate (described by 3D bounding boxes of target subjects). Through cross-coordinate conditioning fusion, these heterogeneous cues are unified to guide the generative process, enabling precise control of camera motion, focus, and framing while preserving temporal and spatial coherence. Extensive experiments demonstrate that our method produces controllable and visually consistent results across diverse camera operations, outperforming existing approaches in both realism and editability.

Abstract:We present Gen Through the Lens, a camera-controllable image-to-video generation framework that synthesizes videos from a given reference image while enabling realistic and physically consistent camera operations. Unlike previous models that treat camera motion as abstract conditioning or rely on a single coordinate system, our approach explicitly decomposes and fuses control signals across three coordinate spaces: the camera coordinate (modeled by intrinsic parameters), the world coordinate (represented by spatial geometry hints), and the object coordinate (described by 3D bounding boxes of target subjects). Through cross-coordinate conditioning fusion, these heterogeneous cues are unified to guide the generative process, enabling precise control of camera motion, focus, and framing while preserving temporal and spatial coherence. Extensive experiments demonstrate that our method produces controllable and visually consistent results across diverse camera operations, outperforming existing approaches in both realism and editability.

15 pages, 9 figures, 5 tables. Shuai Li and Jinxu Xiang contributed equally; Ran Zhang is the corresponding author

Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI); Graphics (cs.GR); Machine Learning (cs.LG)

arXiv:2610.15900 [cs.CV]

(or arXiv:2610.15900v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2610.15900 Focus to learn more arXiv-issued DOI via DataCite (pending registration)

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Full text under the publisher's open licence · 260 words

Related coverage