REVIVE 3D: Refinement via Encoded Voluminous Inflated prior for Volume Enhancement

2026-04-30

AI Summary

Generating three-dimensional assets with realistic volumetric structure from a single flat image remains challenging. This work proposes a two-stage, plug-and-play approach: first, it constructs an inflated prior that integrates global volume and local structural cues; then, within the latent space, it leverages a noise-adding and denoising mechanism to incorporate this geometric prior, thereby activating the implicit 3D knowledge embedded in a pre-trained 3D generative model for refinement. The method innovatively introduces the inflated prior and a 3D latent-space refinement strategy, enabling image-guided 3D editing. Furthermore, it presents Compactness and Normal Anisotropy—novel metrics aligned with human perception for evaluating volumetric quality. Experiments demonstrate state-of-the-art 3D generation performance on standard image datasets, with both qualitative and quantitative evaluations confirming its superiority.

Abstract

Recent generative models have shown strong performance in generating diverse 3D assets from 2D images, a fundamental research topic in computer vision and graphics. However, these models still struggle to generate voluminous 3D assets when the input is a flat image that provides limited 3D cues. We introduce REVIVE 3D, a two-stage, plug-and-play pipeline for generating voluminous 3D assets from flat images. In Stage 1, we construct an Inflated Prior by inflating the foreground silhouette to recover global volume and superimposing part-aware details to capture local structure. In Stage 2, 3D Latent Refinement injects Gaussian noise into the Inflated Prior's latent and then denoises it, using the prior's geometric cues to leverage the backbone's pretrained 3D knowledge. Furthermore, our framework supports image-conditioned 3D editing. To quantify volume and surface flatness, we propose Compactness and Normal Anisotropy. We validate Compactness and Normal Anisotropy through a user study, showing that these metrics align with human perception of volume and quality. We show that REVIVE 3D achieves state-of-the-art performance on a challenging flat image dataset, based on extensive qualitative and quantitative evaluations.

Problem

• 3D generation
• voluminous 3D assets
• flat image
• volume enhancement
• 3D cues

Innovation

• Inflated Prior
• 3D Latent Refinement
• Volume Enhancement
• Compactness
• Normal Anisotropy