# Probing into Camera Control of Video Models

Source: [Probing into Camera Control of Video Models](<https://www.sota2.com/research/paper/6a06a8badbc836b87d82a222>)

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

Data updated: 2026-05-16T05:00:12.406Z

Authors: Chen Hou, Christian Rupprecht

Published: 2026-05-14T13:27:33.000Z

## Abstract

Video is a rich and scalable source of 3D/4D visual observations, and camera control is a key capability for video generation models to produce geometrically meaningful content. Existing approaches typically learn a mapping from camera motion to video using additional camera modules and paired data. However, such datasets are often limited in scale, diversity, and scene dynamics, which can bias the model toward a narrow output distribution and compromise the strong prior learned by the base model. These limitations motivate a different perspective on camera control. In this paper, we show that camera control need not be modeled as an implicit mapping problem, but can instead be treated as a form of geometric guidance that induces displacements across frames. Specifically, we reformulate camera control into a set of displacement fields and apply them via differentiable resampling of latent features during denoising. Our simple approach achieves effective camera control with minimal degradation across diverse quality metrics compared to fine-tuned baselines. Since our method is applicable to most video diffusion models without training, it can also serve as a probe to study the camera control capabilities of base models. Using this probe, we identify universal biases shared by representative video models, as well as disparities in their responses to camera control. Finally, we benchmark their performance in multi-view generation, offering insights into their potential for 3D/4D tasks.

## 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 Control in Video Generation on VBench](<https://www.sota2.com/research/sota/camera-control-in-video-generation-on-vbench.md>): Camera Control in Video Generation; VBench; Dynamic Degree: 55.24.

- [Camera-controlled Video Generation on VBench multi-view consistency and 30 custom single-view prompts 2.0 (test)](<https://www.sota2.com/research/sota/camera-controlled-video-generation-on-vbench-multi-view-consistency-and-30-custom-single-view-prompts-2-0-test.md>): Camera-controlled Video Generation; VBench multi-view consistency and 30 custom single-view prompts 2.0 (test); Dynamic Degree: 55.24.

- [Multi-view Geometry Evaluation on Objaverse-XL](<https://www.sota2.com/research/sota/multi-view-geometry-evaluation-on-objaverse-xl.md>): Multi-view Geometry Evaluation; Objaverse-XL; RPE-T: not recorded.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.14815>)

[PDF](<https://arxiv.org/pdf/2605.14815.pdf>)
