AI Paper Atlas

Research map / Video / Video generation

Transition Matching Distillation for Fast Video Generation

Weili Nie, Julius Berner, Nanye Ma, Chao Liu, Saining Xie, Arash Vahdat

CVPR 2026 · Video › Video generation · approach: Efficient video diffusion models · 25 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 flow head rollout (TMD) to one-step DMD2-v distillationDMD2-v · VBench, Wan2.1 14B one-stepoverall score: 83.69 -> 84.24Improved
add flow head (N2H5) to two-step DMD2-vDMD2-v · VBench, Wan2.1 1.3Boverall score: 84.39 -> 84.68Improved
add flow head to two-step DMD2-v on 14BDMD2-v · VBench, Wan2.1 14B two-stepoverall score: 84.79 -> 84.92 vs rCM 85.05Hurt
use Conv3D discriminator head instead of Conv1D-2D or attention (ablation)DMD2-v · VBench, one-step 1.3Boverall score: 82.32 -> 83.24Improved
add GAN loss (ablation)DMD2-v · VBench, one-step 1.3Boverall score: 81.63 -> 83.24Improved
add KD warm-up (ablation)DMD2-v one-step · VBench, 1.3Boverall score: 83.06 -> 83.24Improved
add KD warm-up (ablation)DMD2-v two-step · VBench, 1.3Boverall score: 84.39 -> 83.79Hurt
apply timestep shift to student timesteps (ablation)DMD2-v two-step · VBench, 1.3Boverall score: 83.44 -> 84.39Improved
apply timestep shift to DMD loss noise level (ablation)DMD2-v one-step · VBench, 1.3Boverall score: 83.22 -> 83.24No clear effect
use MeanFlow instead of flow matching for stage-1 pretraining (ablation)TMD N4H5 · VBench, 1.3B two-stepoverall score: 84.29 -> 84.67Improved
use MeanFlow pretraining (ablation)TMD N2H5 · VBench, 1.3B two-stepoverall score: 84.61 -> 84.68No clear effect
add flow head rollout during distillation (ablation)TMD · VBench convergenceoverall scoreImproved

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