4DHumanDiff: Direct Text-to-4DGS Generation for Consistent 360-Degree Dynamic Humans

2026-07-29

AI Summary

This work addresses the challenge of generating high-quality, temporally coherent, and viewpoint-consistent dynamic human avatars from text, a task where existing methods often rely on pre-rendered videos or per-frame reconstruction, leading to geometric incompleteness and inconsistency. The paper proposes the first end-to-end diffusion framework that directly maps text to 4D Gaussian Splatting (4DGS) representations without intermediate video synthesis or frame-wise optimization. Key contributions include a large-scale text–4DGS paired dataset, a 2D image supervision regularizer, a training-free 4D interpolation strategy, and a novel diffusion architecture based on a 3D U-Net with temporal attention. The method generates full 360° viewpoint-consistent and smoothly animated humans in under a minute, significantly outperforming prior approaches in multi-view and temporal consistency while achieving over 10× faster inference.

Abstract

Generating high-quality 360-degree dynamic human assets from text prompts is challenging. Existing methods usually synthesize monocular or multi-view videos first and then fit a 4D representation, which is expensive and often causes incomplete geometry or view-inconsistent renderings. We present 4DHumanDiff, a diffusion framework that directly generates dynamic humans represented by 4D Gaussian Splatting (4DGS) from text prompts. By modeling the structured 4D representation space end-to-end, 4DHumanDiff avoids video pre-generation and per-scene reconstruction, making it better suited for view-consistent and temporally coherent asset generation. The model uses a 3D U-Net backbone with temporal attention for motion-aware generation. We further construct a large-scale text-to-4DGS dataset with 60,000 high-quality pairs, and introduce 2D regularization and training-free 4D interpolation to improve rendering quality and motion smoothness. Experiments show that 4DHumanDiff generates consistent 360-degree dynamic humans within one minute, achieves better temporal and multi-view consistency, and reduces inference time by more than 10x.

Problem

• text-to-4D
• dynamic humans
• view consistency
• temporal coherence
• 4D generation

Innovation

• 4D Gaussian Splatting
• text-to-4D generation
• diffusion model
• temporal consistency
• view-consistent rendering