Metraipolis › archive › dc_ec06a136d759
Sparse auto-regressive modeling for scene generation from multi-view images — arXiv:2609.03931
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2609.03931
- License (as archived)
- CC0-1.0
- Archived
- 2026-10-17T02:08:42.695Z
- SHA-256 of the archived text
393d6961ec7e3e995f0de4afbfe4fd867fb14793dba084a8536a82d7d7500876- Stable URL of this page
https://metraipolis.com/documento/dc_ec06a136d759
How to cite
Sparse auto-regressive modeling for scene generation from multi-view images — arXiv:2609.03931 — arXiv (en), https://arxiv.org/abs/2609.03931 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_ec06a136d759 (Metraipolis archive dc_ec06a136d759, sha256 393d6961ec7e3e995f0de4afbfe4fd867fb14793dba084a8536a82d7d7500876)
BibTeX
@misc{dcec06a136d759,
title = {Sparse auto-regressive modeling for scene generation from multi-view images — arXiv:2609.03931},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2609.03931},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_ec06a136d759; sha256 393d6961ec7e3e995f0de4afbfe4fd867fb14793dba084a8536a82d7d7500876},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2609.03931. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
Generating complete 3D scenes from sparse, unconstrained views is a fundamental challenge in 3D vision which requires reasoning beyond observed content while remaining computationally tractable. Existing feed-forward reconstruction methods are inherently limited to content visible in the input images, while 3D generative modeling is hindered by the high computational cost of dense volumetric representations and the scarcity of large-scale 3D supervision. We introduce SPAR3S, a sparse voxel-aligned 3D latent generative model for conditional scene completion without requiring ground-truth 3D data for supervision. Our key insight is to formulate 3D scene generation in a structured, compact, voxel-aligned 3D latent space where only occupied voxels are represented. We learn this sparse latent space directly from multi-view images using photometric supervision via differentiable 3D Gaussian Splatting. Given a partial set of observed voxels encoded from sparse input views, scene completion reduces to predicting the missing latent tokens and their spatial support within the voxel grid. To this end, we train a masked autoregressive transformer that jointly models voxel occupancy and latent token values, enabling efficient and spatially consistent generation of unseen regions. We demonstrate the effectiveness of our method on synthetic indoor scenes, achieving higher novel-view quality than prior work. We further validate its generalization on RealEstate10k, highlighting its applicability to real-world data. Tema degli esploratori: agenti e intelligenza artificiale (retro-catalogo arXiv: annunci del 2026-09-04 in cs.LG). Categorie arXiv: Computer Vision and Pattern Recognition (cs.CV) ; Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2609.03931 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-10-17T01:47:25.633Z dalla pagina https://arxiv.org/catchup/cs.LG/2026-09-04?abs=True (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).