Easy-3Dfy composes object meshes into single-image 3D scenes
Original title: Easy-3Dfy: Object-Level Geometric Composition for Single-Image 3D Scene Construction
An RPF aligner matches object-local meshes to partial scene pointmaps without full scene supervision.
Easy-3Dfy separates specialized 3D object generation from scene-layout estimation and reconnects them with an RPF-based geometric aligner. The workflow recovers object layout from correspondence between local meshes and partial scene pointmaps. The authors report consistent gains over SAM-3Dfy on geometric metrics for Scene-Maker and 3D-FUTURE200. It is intended to benefit separately from advances in object generators and 3D foundation models.
- The method does not require complete scene-level supervision.
- Evaluation covers Scene-Maker and 3D-FUTURE200.
- SAM-3Dfy is the stated end-to-end baseline.
From the original article
1 min read at huggingface.co
Single-image 3D scene construction is a fundamental problem in 3D vision. Recent methods such as SAM-3Dfy enable end-to-end reconstruction of 3D assets and their scene layouts, but still struggle with fine-grained scene geometry and asset quality. Meanwhile, specialized 3D object generators provide detailed object geometry, while 3D foundation models offer strong and generalizable spatial understanding.
By Ziang Zhang, Wenyou Yi, Zehan Wang · huggingface.co
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.