Building Rome from a Single Image
AuthorsJiraphon Yenphraphai, Fang Li, Tianshuo Xu, Depu Meng, Quentin Herau, Yihan Hu, Raymond A. Yeh, Wei Zhan
AffiliationsApplied Intuition · Purdue University 3 4 University of Illinois at Urbana-Champaign UC Berkeley * †
The method turns a single photograph into a detailed, complete 3D world by extending object-generation priors to large outdoor scenes.
Key results
Outdoor scenes added to training data.
Geometric reconstruction score.
F1 score at the 10 cm distance threshold.
Reduction versus the strongest baseline.
Improvement versus the strongest baseline.
Times faster than fixed-size chunking.
What the paper found
Building Rome from a Single Image adapts the Trellis 2 3D generator to reconstruct complete indoor and outdoor scene meshes from a single photograph, including geometry hidden from view. Its key innovation is adaptive chunking: smaller chunks preserve detail near the camera, while larger chunks cover distant structures. The model uses DINOv3 image features lifted onto observed surfaces through a monocular point map, plus depth-based visibility signals that distinguish free space from regions where geometry is occluded. It generates overlapping chunks autoregressively and reuses their shared latents to maintain continuity. To address limited outdoor training data, the authors synthesize 4000 scenes using a pipeline involving OpenAI’s GPT-5.6, Ideogram 4, and Google DeepMind’s Nano Banana. On Tanks and Temples, the method reports a normalized Chamfer distance of 1.99 and an F1 score of 0.506, corresponding to a 24% lower Chamfer distance and 30% higher F1 than the strongest baselines on those metrics. Adaptive chunking is 3 times faster than fixed-size chunking while improving reconstruction quality.
Original abstract
Single-image scene generation aims to produce a complete 3D scene mesh from a single image, including surfaces the camera did not observe. While pretrained 3D object generators encode a strong shape prior, they are mainly designed for isolated objects in a fixed canonical volume and focus mostly on indoor scenes, since diverse 3D data for outdoor scenes are quite limited. In this work, we present a method that redesigns such an object-centric generator, e.g., Trellis 2, to work on both indoor and outdoor scenes while retaining its prior. We accomplish this by (a) partitioning the scene into adaptive chunks that scale relative to the distance to the camera; nearby chunks have a smaller size to keep the finer detail, while distant structures, e.g., buildings, are covered by large chunks; (b) making the generator capture explicit 2D-3D correspondence by lifting image features and making the model aware of the free space, observed surface, and unobserved region; (c) synthesizing around 4,000 outdoor scenes to broaden the training data, as existing scene datasets are largely indoor. Experiments on Tanks and Temples, ScanNet++, and in-the-wild images show that our method outperforms all baselines in geometric accuracy and perceptual quality across both indoor and outdoor scenes.
Read the original paperMore in Generative Models
Browse all 72 papers →T3lescope: Arbitrary-Resolution High-Fidelity Generative Surface Reconstruction from Images
Atsuhiro Noguchi, Tianhan Xu, Yiming Liang, Yuta Kikuchi, Masahiro Ishiyama, Shintaro Takagi, Hitoshi Murai, Eiichi Matsumoto
T3lescope uses a scalable coarse-to-fine generative process to reconstruct detailed, globally consistent 3D surfaces from posed images across scenes of vastly different sizes.
UniSlider: Perceptually Uniform Sliders for Continuous Image Editing
David Serrano-Lozano, Duygu Ceylan, Yannick Hold-Geoffroy, Iliyan Georgiev, Javier Vazquez-Corral, Anna Frühstück
UniSlider makes image-editing sliders feel more predictable by ensuring that each movement produces a roughly equal perceptual change.
Empirical Variational Autoencoder
Kaede Shiohara
EVA makes VAEs generate high-quality images and sounds faster by replacing their fixed Gaussian latent prior with a learned, self-predictive one.