MCVD: Masked Conditional Video Diffusion for Prediction, Generation, and Interpolation

arXiv 2022 30.5 method

TLDR

MCVD uses masked conditional video diffusion to perform video prediction, generation, and interpolation with a single model, achieving SOTA results.

Reasoning

The paper introduces a novel masking technique that enables a single diffusion model to handle multiple video synthesis tasks, with strong empirical results on standard benchmarks. However, it does not address long-term temporal consistency or explicitly model world dynamics, limiting its scope as a world model.

Read-first score

Read-first score 30.5, weighted from topical fit, citation, graph, method, reproducibility, and recency signals. Original total remains 3.

Recency 8%

56.5

Uses a gentle age decay so recent papers surface without erasing older foundations. 2022

Methodology quality 25%

Screens visible abstract and analysis fields for experiment, dataset, baseline, metric, and limitation evidence. markers=benchmark,experiment,result

Reproducibility 25%

Screens links and visible text for paper, code, dataset, artifact, and repository signals. pdf=True; code=False; dataset=False; markers=code,github

Topical relevance 42%

4.3

Uses existing LLM keyword relevance scores normalized to 0-100. world model,world simulator,generative world model,interactive world model,video world model,world dynamics prediction,model-based reinforcement learning world model

Field roles

Candidate

Rank sensitivity

Stability: volatile; rank range: 27.

Keyword Scores

world dynamics prediction

video world model

world model

world simulator

generative world model

interactive world model

model-based reinforcement learning world model

Resources

Paper