# Generative View Stitching

Source: [Generative View Stitching](<https://www.sota2.com/research/paper/69019f863ef7ccd74e5e3b90>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-04-14T05:00:04.234Z

Authors: Chonghyuk Song, Michal Stary, Boyuan Chen, George Kopanas, Vincent Sitzmann

Published: 2025-10-28T17:59:58.000Z

## Abstract

Autoregressive video diffusion models are capable of long rollouts that are stable and consistent with history, but they are unable to guide the current generation with conditioning from the future. In camera-guided video generation with a predefined camera trajectory, this limitation leads to collisions with the generated scene, after which autoregression quickly collapses. To address this, we propose Generative View Stitching (GVS), which samples the entire sequence in parallel such that the generated scene is faithful to every part of the predefined camera trajectory. Our main contribution is a sampling algorithm that extends prior work on diffusion stitching for robot planning to video generation. While such stitching methods usually require a specially trained model, GVS is compatible with any off-the-shelf video model trained with Diffusion Forcing, a prevalent sequence diffusion framework that we show already provides the affordances necessary for stitching. We then introduce Omni Guidance, a technique that enhances the temporal consistency in stitching by conditioning on both the past and future, and that enables our proposed loop-closing mechanism for delivering long-range coherence. Overall, GVS achieves camera-guided video generation that is stable, collision-free, frame-to-frame consistent, and closes loops for a variety of predefined camera paths, including Oscar Reutersv\\"ard's Impossible Staircase. Results are best viewed as videos at https://andrewsonga.github.io/gvs.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Camera-guided Video Generation on Panorama 1-loop](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-panorama-1-loop.md>): Camera-guided video generation; Panorama 1-loop; F2FC: 0.138.

- [Camera-guided Video Generation on Circle 1-loop](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-circle-1-loop.md>): Camera-guided video generation; Circle 1-loop; F2FC: 0.16.

- [Camera-guided Video Generation on Straight line](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-straight-line.md>): Camera-guided video generation; Straight line; F2FC: 0.08.

- [Camera-guided Video Generation on Stairs](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-stairs.md>): Camera-guided video generation; Stairs; F2FC: 0.137.

- [Camera-guided Video Generation on Staircase circuit](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-staircase-circuit.md>): Camera-guided video generation; Staircase circuit; F2FC: 0.129.

- [Camera-guided Video Generation on Panorama 2-loop](<https://www.sota2.com/research/sota/camera-guided-video-generation-on-panorama-2-loop.md>): Camera-guided video generation; Panorama 2-loop; F2FC: 0.182.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2510.24718>)

[PDF](<https://arxiv.org/pdf/2510.24718.pdf>)
