G1NDiff: Fast 3D scene editing with reprojection-conditioned GAN-diffusion training on 2D datasets

Computers & Graphics 2026 Open access

Read full text
Authors: Ivan Ojeda Martin, Jorge Bustos Sanchez, Alessio Mazzucchelli, Mario Alfonso Arsuaga, Adrian Penate Sanchez
DOI: 10.1016/j.cag.2026.104630
Cited by: 1
References: 12

Abstract

• G1NDiff: Reprojection-Aware One-Step Diffusion for Fast Multi-View Consistent 3D Scene Editing • G1NDiff: A one-step GAN-based image editing model that preserves consistency across successive edits. • Occlusion-aware Reprojection: An algorithm that transforms a single image into a pair of conditioning-target views. • Multi-stage Pipeline: A scene editing pipeline where G1NDiff is applied for gradually editing all views of a given 3D scene. • State of the art performance on 2DGS scenes, with higher MEt3R consistency and over 10× speedup.

Fields of Study & Subjects

Generative Adversarial Networks and Image Synthesis 3D Shape Modeling and Analysis Computer Graphics and Visualization Techniques

Publication details

Journal / source
Computers & Graphics
Publisher
Elsevier BV
Volume
138
Pages
104630–104630
Publication date
26 May 2026
Language
EN