LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias

• URL: http://arxiv.org/abs/2410.17242v1
• Date: Tue, 22 Oct 2024 17:58:28 GMT
• Title: LVSM: A Large View Synthesis Model with Minimal 3D Inductive Bias
• Authors: Haian Jin, Hanwen Jiang, Hao Tan, Kai Zhang, Sai Bi, Tianyuan Zhang, Fujun Luan, Noah Snavely, Zexiang Xu,
• Abstract summary: We propose a transformer-based approach for scalable and generalizable novel view synthesis from sparse-view inputs. We introduce two architectures: (1) an encoder-decoder LVSM, which encodes input image tokens into a fixed number of 1D latent tokens; and (2) a decoder-only LVSM, which directly maps input images to novel-view outputs.
• Score: 50.13457154615262
• License:
• Abstract: We propose the Large View Synthesis Model (LVSM), a novel transformer-based approach for scalable and generalizable novel view synthesis from sparse-view inputs. We introduce two architectures: (1) an encoder-decoder LVSM, which encodes input image tokens into a fixed number of 1D latent tokens, functioning as a fully learned scene representation, and decodes novel-view images from them; and (2) a decoder-only LVSM, which directly maps input images to novel-view outputs, completely eliminating intermediate scene representations. Both models bypass the 3D inductive biases used in previous methods -- from 3D representations (e.g., NeRF, 3DGS) to network designs (e.g., epipolar projections, plane sweeps) -- addressing novel view synthesis with a fully data-driven approach. While the encoder-decoder model offers faster inference due to its independent latent representation, the decoder-only LVSM achieves superior quality, scalability, and zero-shot generalization, outperforming previous state-of-the-art methods by 1.5 to 3.5 dB PSNR. Comprehensive evaluations across multiple datasets demonstrate that both LVSM variants achieve state-of-the-art novel view synthesis quality. Notably, our models surpass all previous methods even with reduced computational resources (1-2 GPUs). Please see our website for more details: https://haian-jin.github.io/projects/LVSM/ .

Related papers

• Synthetic Prior for Few-Shot Drivable Head Avatar Inversion [61.51887011274453]
  We present SynShot, a novel method for the few-shot inversion of a drivable head avatar based on a synthetic prior. Inspired by machine learning models trained solely on synthetic data, we propose a method that learns a prior model from a large dataset of synthetic heads. We model the head avatar using 3D Gaussian splatting and a convolutional encoder-decoder that outputs Gaussian parameters in UV texture space.
  arXiv  Detail & Related papers  (2025-01-12T19:01:05Z)
• LiftRefine: Progressively Refined View Synthesis from 3D Lifting with Volume-Triplane Representations [21.183524347952762]
  We propose a new view synthesis method via a 3D neural field from both single or few-view input images. Our reconstruction model first lifts one or more input images to the 3D space from a volume as the coarse-scale 3D representation. Our diffusion model then hallucinates missing details in the rendered images from tri-planes.
  arXiv  Detail & Related papers  (2024-12-19T02:23:55Z)
• DiHuR: Diffusion-Guided Generalizable Human Reconstruction [51.31232435994026]
  We introduce DiHuR, a Diffusion-guided model for generalizable Human 3D Reconstruction and view synthesis from sparse, minimally overlapping images. Our method integrates two key priors in a coherent manner: the prior from generalizable feed-forward models and the 2D diffusion prior, and it requires only multi-view image training, without 3D supervision.
  arXiv  Detail & Related papers  (2024-11-16T03:52:23Z)
• DistillNeRF: Perceiving 3D Scenes from Single-Glance Images by Distilling Neural Fields and Foundation Model Features [65.8738034806085]
  DistillNeRF is a self-supervised learning framework for understanding 3D environments in autonomous driving scenes. Our method is a generalizable feedforward model that predicts a rich neural scene representation from sparse, single-frame multi-view camera inputs.
  arXiv  Detail & Related papers  (2024-06-17T21:15:13Z)
• Free3D: Consistent Novel View Synthesis without 3D Representation [63.931920010054064]
  Free3D is a simple accurate method for monocular open-set novel view synthesis (NVS) Compared to other works that took a similar approach, we obtain significant improvements without resorting to an explicit 3D representation.
  arXiv  Detail & Related papers  (2023-12-07T18:59:18Z)
• GenLayNeRF: Generalizable Layered Representations with 3D Model Alignment for Multi-Human View Synthesis [1.6574413179773757]
  GenLayNeRF is a generalizable layered scene representation for free-viewpoint rendering of multiple human subjects. We divide the scene into multi-human layers anchored by the 3D body meshes. We extract point-wise image-aligned and human-anchored features which are correlated and fused.
  arXiv  Detail & Related papers  (2023-09-20T20:37:31Z)
• Efficient View Synthesis and 3D-based Multi-Frame Denoising with Multiplane Feature Representations [1.18885605647513]
  We introduce the first 3D-based multi-frame denoising method that significantly outperforms its 2D-based counterparts with lower computational requirements. Our method extends the multiplane image (MPI) framework for novel view synthesis by introducing a learnable encoder-renderer pair manipulating multiplane in feature space.
  arXiv  Detail & Related papers  (2023-03-31T15:23:35Z)
• NeRDi: Single-View NeRF Synthesis with Language-Guided Diffusion as General Image Priors [24.05480789681139]
  We propose NeRDi, a single-view NeRF synthesis framework with general image priors from 2D diffusion models. We leverage off-the-shelf vision-language models and introduce a two-section language guidance as conditioning inputs to the diffusion model. We also demonstrate our generalizability in zero-shot NeRF synthesis for in-the-wild images.
  arXiv  Detail & Related papers  (2022-12-06T19:00:07Z)
• Vision Transformer for NeRF-Based View Synthesis from a Single Input Image [49.956005709863355]
  We propose to leverage both the global and local features to form an expressive 3D representation. To synthesize a novel view, we train a multilayer perceptron (MLP) network conditioned on the learned 3D representation to perform volume rendering. Our method can render novel views from only a single input image and generalize across multiple object categories using a single model.
  arXiv  Detail & Related papers  (2022-07-12T17:52:04Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.