4Real: Towards Photorealistic 4D Scene Generation via Video Diffusion Models

TL;DR

This paper presents a novel pipeline for photorealistic 4D scene generation from text, leveraging video diffusion models trained on real-world data, and introduces techniques for canonical representation and deformation modeling to improve realism and consistency.

Contribution

It introduces a new method that generates photorealistic 4D scenes without relying on multi-view generative models, using video diffusion models and deformation learning.

Findings

Produces highly photorealistic 4D scenes from text prompts.

Outperforms existing methods in realism and structural consistency.

Enables multi-view visualization of dynamic scenes.

Abstract

Existing dynamic scene generation methods mostly rely on distilling knowledge from pre-trained 3D generative models, which are typically fine-tuned on synthetic object datasets. As a result, the generated scenes are often object-centric and lack photorealism. To address these limitations, we introduce a novel pipeline designed for photorealistic text-to-4D scene generation, discarding the dependency on multi-view generative models and instead fully utilizing video generative models trained on diverse real-world datasets. Our method begins by generating a reference video using the video generation model. We then learn the canonical 3D representation of the video using a freeze-time video, delicately generated from the reference video. To handle inconsistencies in the freeze-time video, we jointly learn a per-frame deformation to model these imperfections. We then learn the temporal deformation based on the canonical representation to capture dynamic interactions in the reference video. The pipeline facilitates the generation of dynamic scenes with enhanced photorealism and structural integrity, viewable from multiple perspectives, thereby setting a new standard in 4D scene generation.