Geometry as Address: Routing Attention to Visual Memory for Long-Horizon Camera-Controlled Video Generation

Hugging Face Daily Papers Papers
video-generation camera-control visual-memory geometry-based attention-routing long-horizon state-of-the-art

Summary

This paper introduces GEAR, a Geometry-Enabled Attention Routing framework that uses geometry as explicit token-level addresses for visual memory to enhance long-horizon camera-controlled video generation, achieving state-of-the-art results.

Long-horizon camera-controlled video generation requires recovering previously observed content from an ever-growing visual history. Existing approaches either search historical context implicitly or reconstruct it into persistent 3D memory, facing inefficient memory access or accumulated geometric errors. Our key insight is that geometry need not explain the scene--it only needs to determine where visual memory should be read from, while attention decides what should be recovered. Based on this insight, we introduce GEAR, a Geometry-Enabled Attention Routing framework that uses geometry as an explicit token-level address for visual memory. Rather than fusing historical observations into a persistent global 3D representation, GEAR retains them as frame latents and uses per-frame geometry only to establish token-level correspondences with target views, thereby avoiding persistent error accumulation from global fusion. Guided by these correspondences, Geometric Correspondence Attention (GCA) selectively injects geometrically matched historical features into noisy target patches during denoising. We further introduce an Invisible Octree to accumulate visibility evidence and reject geometrically plausible but occluded correspondences. Extensive experiments demonstrate that GEAR achieves state-of-the-art visual quality, precise camera control, and revisit consistency, enabling minute-long video generation along challenging trajectories.
Original Article
View Cached Full Text

Cached at: 09/29/26, 04:11 PM

Paper page - Geometry as Address: Routing Attention to Visual Memory for Long-Horizon Camera-Controlled Video Generation

Source: https://huggingface.co/papers/2609.34722 Authors:

,

,

,

,

,

,

,

,

,

,

,

Abstract

Long-horizoncamera-controlledvideogenerationrequiresrecoveringpreviouslyobservedcontentfromanever-growingvisualhistory.Existingapproacheseithersearchhistoricalcontextimplicitlyorreconstructitintopersistent3Dmemory,facinginefficientmemoryaccessoraccumulatedgeometricerrors.Ourkeyinsightisthatgeometryneednotexplainthescene--itonlyneedstodeterminewherevisualmemoryshouldbereadfrom,whileattentiondecideswhatshouldberecovered.Basedonthisinsight,weintroduceGEAR,aGeometry-EnabledAttentionRoutingframeworkthatusesgeometryasanexplicittoken-leveladdressforvisualmemory.Ratherthanfusinghistoricalobservationsintoapersistentglobal3Drepresentation,GEARretainsthemasframelatentsandusesper-framegeometryonlytoestablishtoken-levelcorrespondenceswithtargetviews,therebyavoidingpersistenterroraccumulationfromglobalfusion.Guidedbythesecorrespondences,GeometricCorrespondenceAttention(GCA)selectivelyinjectsgeometricallymatchedhistoricalfeaturesintonoisytargetpatchesduringdenoising.WefurtherintroduceanInvisibleOctreetoaccumulatevisibilityevidenceandrejectgeometricallyplausiblebutoccludedcorrespondences.ExtensiveexperimentsdemonstratethatGEARachievesstate-of-the-artvisualquality,precisecameracontrol,andrevisitconsistency,enablingminute-longvideogenerationalongchallengingtrajectories.

View arXiv pageView PDFProject pageGitHubAdd to collection

Get this paper in your agent:

hf papers read 2609\.34722

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper0

No model linking this paper

Cite arxiv.org/abs/2609.34722 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2609.34722 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2609.34722 in a Space README.md to link it from this page.

Collections including this paper0

No Collection including this paper

Add this paper to acollectionto link it from this page.

Similar Articles