Warp-as-History: Generalizable Camera-Controlled Video Generation from One Training Video
Summary
Warp-as-History proposes a novel interface that transforms camera-induced warps into pseudo-history representations, enabling a frozen video generation model to follow camera trajectories without training or test-time optimization. A lightweight LoRA fine-tuning on a single video further improves camera adherence and generalizes to unseen videos.
View Cached Full Text
Cached at: 05/15/26, 04:23 AM
Paper page - Warp-as-History: Generalizable Camera-Controlled Video Generation from One Training Video
Source: https://huggingface.co/papers/2605.15182
Abstract
A novel approach called Warp-as-History enables camera-controlled video generation by transforming camera-induced warps into pseudo-history representations, achieving zero-shot capability without training or test-time optimization.
Camera-controlledvideo generationhas made substantial progress, enabling generated videos to follow prescribed viewpoint trajectories. However, existing methods usually learn camera-specific conditioning through camera encoders, control branches, or attention and positional-encoding modifications, which often require post-training on large-scale camera-annotated videos. Training-free alternatives avoid such post-training, but often shift the cost to test-time optimization or extra denoising-time guidance. We propose Warp-as-History, a simple interface that turnscamera-induced warpsintocamera-warped pseudo-historywithtarget-frame positional alignmentandvisible-token selection. Given a target camera trajectory, we constructcamera-warped pseudo-historyfrom past observations and feed it through the model’svisual-history pathway. Crucially, we align itspositional encodingwith the target frames being denoised and remove warped-history tokens without valid source observations. Without any training, architectural modification, or test-time optimization, this interface reveals a non-trivialzero-shot capabilityof a frozenvideo generationmodel to follow camera trajectories. Moreover, lightweight offlineLoRA finetuningon only one camera-annotated video further improves this capability and generalizes to unseen videos, improving camera adherence, visual quality, and motion dynamics without test-time optimization or target-video adaptation. Extensive experiments on diverse datasets confirm the effectiveness of our method.
View arXiv pageView PDFProject pageGitHub16Add to collection
Get this paper in your agent:
hf papers read 2605\.15182
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/2605.15182 in a model README.md to link it from this page.
Datasets citing this paper0
No dataset linking this paper
Cite arxiv.org/abs/2605.15182 in a dataset README.md to link it from this page.
Spaces citing this paper0
No Space linking this paper
Cite arxiv.org/abs/2605.15182 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
Wonder: Video World Model Done Better
Wonder is a general-purpose video world model that enables real-time, camera-controllable world exploration from an image or conditional video. It introduces camera conditioning via dense coordinate fields, a sparse attention memory mechanism, and techniques to improve distillation, allowing minute-scale video generation at 16 FPS.
ForgeWM: Progressive Causal Training for Few-Step Action-Conditioned Video World Models
ForgeWM is a progressive framework that distills bidirectional video generators into efficient few-step interactive world models, supporting low-latency interaction and replay-time refinement with demonstrated improvements on Minecraft and FPS gameplay.
LoomVideo: Unifying Multimodal Inputs into Video Generation and Editing
LoomVideo introduces a 5B-parameter unified architecture for video generation and editing that reduces computational overhead using novel conditioning mechanisms and multi-modal alignment, achieving competitive performance and faster inference.
LooseControlVideo: Directorial Video Control using Spatial Blocking
LooseControlVideo introduces a framework for intuitive 3D spatial control in text-to-video generation using sparse oriented 3D boxes as proxies, achieving superior trajectory accuracy and occlusion handling. It fine-tunes a Wan 2.2 backbone and demonstrates significant improvements over existing methods on multiple benchmarks.
Track2View: 4D-Consistent Camera-Controlled Video Generation via Paired 3D Point Tracks
Track2View generates novel camera viewpoints from videos by conditioning a video diffusion transformer on paired 3D point tracks, achieving state-of-the-art visual quality and significant reductions in rotation and translation errors.