One View Is Enough: In-the-Wild Monocular Pretraining for Novel View Generation

arXiv:2603.23488 · cs.CV · Submitted 2026-03-24 · Read on arXiv

cs.CV

Submitted: 2026-03-24

Updated: 2026-09-24

Comments: Accepted at NeurIPS 2026. Code: https://github.com/kyutai-labs/ovie. Project page: https://kyutai.org/blog/2026-04-14-ovie/

License: http://creativecommons.org/licenses/by/4.0/

The gist: Monocular novel-view synthesis has long required multi-view image pairs for supervision, limiting training to a narrow set of purpose-built datasets.

Terminology

Abstract

Monocular novel-view synthesis has long required multi-view image pairs for supervision, limiting training to a narrow set of purpose-built datasets. We propose in-the-wild monocular pretraining: a frozen depth estimator lifts each source image into 3D and reprojects under sampled poses to yield pseudo-target views; masked losses restrict supervision to valid regions and an adversarial objective covers disoccluded areas. Scaled to 30 million uncurated images, this produces OVIE, requiring only a source image and target pose at inference. Prior work trains without multi-view data but needs a depth estimator at inference, or drops this dependency but requires multi-view training pairs; OVIE is the first to require neither. Without multi-view supervision, OVIE rivals in-domain baselines on RealEstate10K and surpasses all on DL3DV, producing the most multi-view-consistent trajectories of any geometry-free method; brief multi-view fine-tuning outperforms all geometry-free methods on their training domain. At 116 FPS, it is over 600x faster than the fastest baseline. Code and pretrained models are at https://github.com/kyutai-labs/ovie; video results are on the project page, https://kyutai.org/blog/2026-04-14-ovie/.

Related papers