AI Paper Atlas

Research map / Video / Video generation

SwitchCraft: Training-Free Multi-Event Video Generation with Attention Controls

Qianxun Xu, Chenxi Song, Yujun Cai, Chi Zhang

CVPR 2026 · Video › Video generation · approach: Trajectory-guided motion control · 2 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
add ABSS adaptive strength instead of random strengths (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-T: 0.253 -> 0.275Improved
add ABSS adaptive strength instead of random strengths (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)T2V alignment: 3.62 -> 4.30Improved
add ABSS adaptive strength instead of fixed strength 1 (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-T: 0.264 -> 0.275Improved
add ABSS adaptive strength instead of fixed strength 1 (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)Visual quality: 3.97 -> 4.33Improved
add SVD key compression in ABSS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-T: 0.255 -> 0.275Improved
add SVD key compression in ABSS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)T2V alignment: 3.67 -> 4.30Improved
add enhancement term in EAQS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-T: 0.262 -> 0.275Improved
add enhancement term in EAQS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)Visual quality: 4.35 -> 4.33No clear effect
add suppression term in EAQS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-T: 0.261 -> 0.275Improved
add suppression term in EAQS (ablation)SwitchCraft · Multi-event prompts (60, Wan2.1 14B)T2V alignment: 3.73 -> 4.30Improved
use SwitchCraft steering instead of plain Wan2.1SwitchCraft · Multi-event prompts (60, Wan2.1 14B)T2V alignment: 3.47 -> 4.30Improved
use SwitchCraft steering instead of plain Wan2.1SwitchCraft · Multi-event prompts (60, Wan2.1 14B)CLIP-F: 0.980 -> 0.980No clear effect

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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