AI Paper Atlas

Research map / Video / Video generation

PyramidalWan: On Making Pretrained Video Model Pyramidal for Efficient Inference

Denis Korzhenkov, Adil Karjauv, Animesh Karnewar, Mohsen Ghafoorian, Amirhossein Habibian

CVPR 2026 · Video › Video generation · approach: Efficient video diffusion models · 0 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
make pretrained Wan2.1-1.3B pyramidal by finetuningWan2.1-1.3B · VBenchTotal score: 82.49 -> 82.83No clear effect
pyramidal inference schedule 20-20-10Wan2.1-1.3B · Computational costTFLOPs: 2 x 12592 -> 2 x 2821Improved
pyramidal 2-2-1 schedule instead of 0-0-2 step-distilledWan-DMD · LatencyLatency per pass: 43% speedupImproved
add distillation loss in pyramidal finetuning (ablation)PyramidalWan · VBench-2.0Total score: 54.02 -> 54.93Improved
add teacher-matching loss L_teach to DMD with pyramidal teacher (ablation)PyramidalWan-DMD-PT* · VBench-2.0Total score: 52.36 -> 51.75Hurt
add teacher-matching loss L_teach to DMD with pyramidal teacher (ablation)PyramidalWan-DMD-PT* · VBenchTotal score: 82.44 -> 82.72Improved
use simplified DMD objective instead of full objective (ablation)PyramidalWan-DMD-PT · VBench-2.0Total score: 50.67 -> 51.75Improved
use simplified DMD objective instead of full objective (ablation)PyramidalWan-DMD-PT · VBenchTotal score: 82.56 -> 82.72Improved
downsample clean latents in pixel space instead of latent space (ablation)PyramidalWan · visual qualityqualitativeImproved
fine-tune patch-pyramidal PPF for video diffusion (ablation)Wan-PPF · text-to-video qualityqualitativeHurt
use 1 step instead of 2 steps in Wan-DMD (ablation)Wan-DMD · VBenchTotal score: 83.28 -> 79.45Hurt
use pyramidal DMD with original teacher instead of pyramidal teacherPyramidalWan · VBench-2.0Total score: 51.75 -> 55.36Improved

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