AI Paper Atlas

Research map / Video / Video generation

From Image to Video: An Empirical Study of Diffusion Representations

Pedro Vélez, Luisa F. Polanía, Yi Yang, Chuhan Zhang, Rishabh Kabra, Anurag Arnab, Mehdi S. M. Sajjadi

ICCV 2025 · Video › Video generation · approach: Efficient video diffusion models · 6 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
train the same WALT architecture for video instead of image generationWALT 284M · Places365relative performance change: +0.6%Improved
train the same WALT architecture for video instead of image generationWALT 284M · ImageNet-1krelative performance change: +1.8%Improved
train the same WALT architecture for video instead of image generationWALT 284M · iNaturalist 2018relative performance change: +11%Improved
train the same WALT architecture for video instead of image generationWALT 284M · Kinetics-400relative performance change: +8%Improved
train the same WALT architecture for video instead of image generationWALT 284M · Kinetics-700relative performance change: +12%Improved
train the same WALT architecture for video instead of image generationWALT 284M · Something-Something v2relative performance change: +42%Improved
train the same WALT architecture for video instead of image generationWALT 284M · Waymo box trackingrelative performance change: +23%Improved
train the same WALT architecture for video instead of image generationWALT 284M · RealEstate10K camera poserelative performance change: +60%Improved
train the same WALT architecture for video instead of image generationWALT 284M · Perception Test point trackingrelative performance change: +68%Improved
scale V-WALT from 284M to 1.9B parametersV-WALT · most readout tasksperformance: 0.618 -> 0.728Improved
scale V-WALT from 284M to 1.9B parametersV-WALT · Perception Test point trackingaverage Jaccard: 0.756 -> 0.735Hurt
add small noise (timestep 200) to input latents before feature extraction (ablation)V-WALT · most downstream tasksrelative performanceImproved

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