How Far is Video Generation from World Model: A Physical Law Perspective

• URL: http://arxiv.org/abs/2411.02385v1
• Date: Mon, 04 Nov 2024 18:53:05 GMT
• Title: How Far is Video Generation from World Model: A Physical Law Perspective
• Authors: Bingyi Kang, Yang Yue, Rui Lu, Zhijie Lin, Yang Zhao, Kaixin Wang, Gao Huang, Jiashi Feng,
• Abstract summary: OpenAI's Sora highlights the potential of video generation for developing world models that adhere to physical laws. But the ability of video generation models to discover such laws purely from visual data without human priors can be questioned. In this work, we evaluate across three key scenarios: in-distribution, out-of-distribution, and generalization.
• Score: 101.24278831609249
• License:
• Abstract: OpenAI's Sora highlights the potential of video generation for developing world models that adhere to fundamental physical laws. However, the ability of video generation models to discover such laws purely from visual data without human priors can be questioned. A world model learning the true law should give predictions robust to nuances and correctly extrapolate on unseen scenarios. In this work, we evaluate across three key scenarios: in-distribution, out-of-distribution, and combinatorial generalization. We developed a 2D simulation testbed for object movement and collisions to generate videos deterministically governed by one or more classical mechanics laws. This provides an unlimited supply of data for large-scale experimentation and enables quantitative evaluation of whether the generated videos adhere to physical laws. We trained diffusion-based video generation models to predict object movements based on initial frames. Our scaling experiments show perfect generalization within the distribution, measurable scaling behavior for combinatorial generalization, but failure in out-of-distribution scenarios. Further experiments reveal two key insights about the generalization mechanisms of these models: (1) the models fail to abstract general physical rules and instead exhibit "case-based" generalization behavior, i.e., mimicking the closest training example; (2) when generalizing to new cases, models are observed to prioritize different factors when referencing training data: color > size > velocity > shape. Our study suggests that scaling alone is insufficient for video generation models to uncover fundamental physical laws, despite its role in Sora's broader success. See our project page at https://phyworld.github.io

Related papers

• Interpretable Representation Learning from Videos using Nonlinear Priors [15.779730667509915]
  We propose a deep learning framework where one can specify nonlinear priors for videos. We do this by extending the Variational Auto-Encoder (VAE) prior from a simple isotropic Gaussian to an arbitrary nonlinear temporal Additive Noise Model (ANM) We validate the method on different real-world physics videos including a pendulum, a mass on a spring, a falling object and a pulsar.
  arXiv  Detail & Related papers  (2024-10-24T08:39:24Z)
• Towards World Simulator: Crafting Physical Commonsense-Based Benchmark for Video Generation [51.750634349748736]
  Text-to-video (T2V) models have made significant strides in visualizing complex prompts. However, the capacity of these models to accurately represent intuitive physics remains largely unexplored. We introduce PhyGenBench to evaluate physical commonsense correctness in T2V generation.
  arXiv  Detail & Related papers  (2024-10-07T17:56:04Z)
• AVID: Adapting Video Diffusion Models to World Models [10.757223474031248]
  We propose to adapt pretrained video diffusion models to action-conditioned world models, without access to the parameters of the pretrained model. AVID uses a learned mask to modify the intermediate outputs of the pretrained model and generate accurate action-conditioned videos. We evaluate AVID on video game and real-world robotics data, and show that it outperforms existing baselines for diffusion model adaptation.
  arXiv  Detail & Related papers  (2024-10-01T13:48:31Z)
• Promises and Pitfalls of Generative Masked Language Modeling: Theoretical Framework and Practical Guidelines [74.42485647685272]
  We focus on Generative Masked Language Models (GMLMs) We train a model to fit conditional probabilities of the data distribution via masking, which are subsequently used as inputs to a Markov Chain to draw samples from the model. We adapt the T5 model for iteratively-refined parallel decoding, achieving 2-3x speedup in machine translation with minimal sacrifice in quality.
  arXiv  Detail & Related papers  (2024-07-22T18:00:00Z)
• VideoPhy: Evaluating Physical Commonsense for Video Generation [93.28748850301949]
  We present VideoPhy, a benchmark designed to assess whether the generated videos follow physical commonsense for real-world activities. We then generate videos conditioned on captions from diverse state-of-the-art text-to-video generative models. Our human evaluation reveals that the existing models severely lack the ability to generate videos adhering to the given text prompts.
  arXiv  Detail & Related papers  (2024-06-05T17:53:55Z)
• Predicting Long-horizon Futures by Conditioning on Geometry and Time [49.86180975196375]
  We explore the task of generating future sensor observations conditioned on the past. We leverage the large-scale pretraining of image diffusion models which can handle multi-modality. We create a benchmark for video prediction on a diverse set of videos spanning indoor and outdoor scenes.
  arXiv  Detail & Related papers  (2024-04-17T16:56:31Z)
• A Tale of Tails: Model Collapse as a Change of Scaling Laws [11.6055501181235]
  We ask: How will the scaling laws change in the inevitable regime where synthetic data makes its way into the training corpus? We develop a theoretical framework of model collapse through the lens of scaling laws. We discover a wide range of decay phenomena, analyzing loss of scaling, shifted scaling with number of generations, the ''un-learning" of skills, and grokking when mixing human and synthesized data.
  arXiv  Detail & Related papers  (2024-02-10T21:06:34Z)
• Exploring Strategies for Generalizable Commonsense Reasoning with Pre-trained Models [62.28551903638434]
  We measure the impact of three different adaptation methods on the generalization and accuracy of models. Experiments with two models show that fine-tuning performs best, by learning both the content and the structure of the task, but suffers from overfitting and limited generalization to novel answers. We observe that alternative adaptation methods like prefix-tuning have comparable accuracy, but generalize better to unseen answers and are more robust to adversarial splits.
  arXiv  Detail & Related papers  (2021-09-07T03:13:06Z)
• Towards an Interpretable Latent Space in Structured Models for Video Prediction [30.080907495461876]
  We focus on the task of future frame prediction in video governed by underlying physical dynamics. We work with models which are object-centric, i.e., explicitly work with object representations, and propagate a loss in the latent space.
  arXiv  Detail & Related papers  (2021-07-16T05:37:16Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.