AI Paper Atlas

Research map / Video / Video generation

BulletTime: Decoupled Control of Time and Camera Pose for Video Generation

Yiming Wang, Qihang Zhang, Shengqu Cai, Tong Wu, Jan Ackermann, Zhengfei Kuang, Yang Zheng, Frano Rajič, Siyu Tang, Gordon Wetzstein

CVPR 2026 · Video › Video generation · approach: Trajectory-guided motion control · 8 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use joint 4D conditioning model instead of time-remapping input then camera-only generationReCamMaster* · Synthetic 4D-controlled videosPSNR: 21.86 -> 24.57Improved
use end-to-end 4D model instead of two-stage baselineReCamMaster* · Real-world videos (ViPE)RotErr: 2.98 -> 1.47Improved
use end-to-end 4D model instead of TrajectoryCrafter*TrajectoryCrafter* · Real-world videos (ViPE)VBench Imaging Quality: 0.5696 -> 0.5598Hurt
condition on the same input video for all time settings (end-to-end) instead of remapping inputReCamMaster* · Synthetic background consistencymPSNR: 25.80 -> 28.29Improved
replace standard RoPE with Time-RoPE (with AdaLN) (ablation)AdaLN world-time conditioning · Synthetic world-time V2VPSNR: 29.83 -> 32.15Improved
use AdaLN instead of cross-attention for time conditioning (standard RoPE) (ablation)RoPE-based conditioning · Synthetic world-time V2VPSNR: 23.86 -> 29.83Improved
use AdaLN instead of channel addition for time conditioning (standard RoPE) (ablation)RoPE-based conditioning · Synthetic world-time V2VPSNR: 25.31 -> 29.83Improved
add AdaLN to Time-RoPE (ablation)Time-RoPE · Synthetic world-time V2VPSNR: 30.45 -> 32.15Improved
add cross-attention or channel addition on top of Time-RoPE (ablation)Time-RoPE · Synthetic world-time V2VPSNR: 30.45 -> 30.51 (cross-attn), 30.40 (channel add)No clear effect
add 4D-RoPE (ablation)full 4D model · Synthetic 4D conditioningPSNR: 21.98 -> 23.45Improved
add camera/time AdaLN (ablation)full 4D model · Synthetic 4D conditioningPSNR: 22.74 -> 23.45Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy