MV-Forcing: Long Multi-View Video Generation via 4D-Grounded Spatio-Temporal Self-Forcing

Hugging Face Daily Papers Papers
multi-view-video-generation video-diffusion 4d-geometric-bridge spatio-temporal-self-forcing autoregressive-generation distribution-matching-distillation

Summary

MV-Forcing introduces a diffusion framework that combines temporal and view-wise autoregression to generate long, multi-view consistent videos of dynamic scenes, using a 4D geometric bridge and spatio-temporal distillation to enable arbitrary-length generation from a few-step student model.

Recent advances in video diffusion models have enabled either long single-view generation through temporal autoregression, or short multi-view synthesis through bidirectional attention. However, generating long, multi-view consistent videos of dynamic scenes remains unsolved. In this work, we present MV-Forcing, a framework that composes temporal and view-wise autoregression within a single diffusion model by introducing a 4D geometric bridge between sequentially generated views. Our key insight is that an autoregressive 3D reconstruction model naturally interfaces between autoregressively generated views. Given a completed source view, we reconstruct its 3D structure and render a geometric prior of the next target viewpoint, which the diffusion model refines into a high-quality video. To extend generation beyond the teacher's fixed temporal window, we introduce a joint denoising regime where both view slots are initialized from noise during training, enabling temporally unbounded generation. We distill the model via Distribution Matching Distillation with Spatio-Temporal Self-Forcing, closing the train-inference exposure bias gap for both temporal and view-sequential autoregression. Extensive experiments on both synthetic and real-world data demonstrate that MV-Forcing produces geometrically consistent multi-view videos of dynamic scenes at arbitrary lengths and viewpoint counts using a single few-step student model.
Original Article
View Cached Full Text

Cached at: 07/07/26, 10:43 AM

Paper page - MV-Forcing: Long Multi-View Video Generation via 4D-Grounded Spatio-Temporal Self-Forcing

Source: https://huggingface.co/papers/2607.05376

Abstract

A video diffusion framework generates long, multi-view consistent videos by combining temporal and view-wise autoregression through 4D geometric bridging and spatio-temporal distillation techniques.

Recent advances invideo diffusion modelshave enabled either long single-view generation throughtemporal autoregression, or shortmulti-view synthesisthroughbidirectional attention. However, generating long, multi-view consistent videos of dynamic scenes remains unsolved. In this work, we present MV-Forcing, a framework that composes temporal and view-wise autoregression within a single diffusion model by introducing a4D geometric bridgebetween sequentially generated views. Our key insight is that anautoregressive 3D reconstruction modelnaturally interfaces between autoregressively generated views. Given a completed source view, we reconstruct its 3D structure and render ageometric priorof the next target viewpoint, which the diffusion model refines into a high-quality video. To extend generation beyond the teacher’s fixed temporal window, we introduce ajoint denoising regimewhere both view slots are initialized from noise during training, enablingtemporally unbounded generation. We distill the model viaDistribution Matching DistillationwithSpatio-Temporal Self-Forcing, closing the train-inference exposure bias gap for both temporal and view-sequential autoregression. Extensive experiments on both synthetic and real-world data demonstrate that MV-Forcing produces geometrically consistent multi-view videos of dynamic scenes at arbitrary lengths and viewpoint counts using a single few-stepstudent model.

View arXiv pageView PDFProject pageAdd to collection

Get this paper in your agent:

hf papers read 2607\.05376

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/2607.05376 in a model README.md to link it from this page.

Datasets citing this paper0

No dataset linking this paper

Cite arxiv.org/abs/2607.05376 in a dataset README.md to link it from this page.

Spaces citing this paper0

No Space linking this paper

Cite arxiv.org/abs/2607.05376 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