PaintScene4D: Consistent 4D Scene Generation from Text Prompts

Vinayak Gupta Yunze Man Yu-Xiong Wang, Indian Institute of Technology, Madras University of Illinois Urbana-Champaign https://paintscene4d.github.io/ · 2024

Read PaintScene4D: Consistent 4D Scene Generation from Text Prompts by Vinayak Gupta Yunze Man Yu-Xiong Wang, Indian Institute of Technology, Madras University of Illinois Urbana-Champaign https://paintscene4d.github.io/ (2024) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2412-04471.txt.

Description

Vinayak Gupta, Yunze Man, and Yu-Xiong Wang present PaintScene4D, a training-free system that builds moving 3D scenes from a text prompt, in a 2024 paper. Photorealistic scenes that unfold over time are hard to generate: older methods lean on object data sets and look synthetic, while text-to-video tools lack camera control and spatial sense. Their pipeline starts from a generated reference video instead.

A chosen array of virtual cameras then re-renders the scene through progressive warping and inpainting, keeping views consistent in space and time. A dynamic renderer finally optimizes the views so users can fly a camera along any path. The authors release the project openly and aim at realistic, controllable scene creation.

Catalog record

CREATORVinayak Gupta Yunze Man Yu-Xiong Wang, Indian Institute of Technology, Madras University of Illinois Urbana-Champaign https://paintscene4d.github.io/
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE40 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-04471

Browse the full library · About the Tower of Babel