AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

AlignVid: Taming Visual Dominance via Training-Free Attention Modulation in Text-guided Image-to-Video Generation

Yexin Liu, Wen-Jie Shu, Zile Huang, Haoze Zheng, Yueze Wang, Manyuan Zhang, Jinjing Zhu, Ser-Nam Lim, Harry Yang

ICML 2026 · Image generation › Text-to-image diffusion · approach: Personalized and region-controlled generation

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 AlignVid attention scalingWan2.1 · OmitI2V addition accuracyaccuracy: 71.75 -> 79.54Improved
add AlignVid attention scalingFramePack · OmitI2V dynamic degreedynamic degree: 20.05 -> 28.53Improved
add AlignVid attention scalingWan2.1 · OmitI2V aesthetic qualityaesthetic quality: 63.12 -> 61.63Hurt
use energy-based modulation instead of scalar scaling (ablation)AlignVid on Wan2.1 · OmitI2V addition accuracyaccuracy: 80.76 -> 75.65Hurt
apply ASM to all denoising steps instead of early steps (ablation)AlignVid on FramePack · OmitI2Vaesthetic quality: 63.41 -> 61.56Hurt
apply ASM only in middle denoising steps (ablation)AlignVid on FramePack · OmitI2V modification accuracyaccuracy: 64.99 -> 62.71Hurt
restrict ASM to foreground-focus blocks instead of background-focus blocks (ablation)AlignVid on Wan2.1 · OmitI2V modification accuracyaccuracy: 71.99 -> 77.20Improved
restrict ASM to first half of blocks instead of last half (ablation)AlignVid on Wan2.1 · OmitI2V modification accuracyaccuracy: 73.68 -> 76.55Improved
scale only image keys in FramePack (Key-image) instead of both image and text keys (ablation)AlignVid on FramePack · OmitI2V aesthetic qualityaesthetic quality: 63.57 -> 24.63Hurt
add AlignVid on top of CFG=5 (ablation)Wan2.1 · OmitI2V modification accuracyaccuracy: 72.35 -> 77.20Improved
add AlignVid to OmniGen2OmniGen2 · GenEval Countingscore: 0.67 -> 0.52(-0.15)Hurt
add AlignVid to Wan2.1-T2V-1.3BWan2.1-T2V-1.3B · VBench Multiple Objectsscore: 53.35 -> 66.46(+13.11)Improved

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