MetaView: Monocular Novel View Synthesis with Scale-Aware Implicit Geometry Priors

Abstract

Current visual generation models are capable of producing high-quality content, yet they lack a coherent perception of the spatial structure. Existing generative novel view synthesis methods typically introduce explicit geometry priors, which enforce spatial consistency but inherently restrict generalization in large view changes. In contrast, recent interactive generative methods favor implicit scene modeling, offering greater flexibility at the cost of precise camera control and geometry consistency. In this work, we propose MetaView, a diffusion-based monocular novel view synthesis framework that enables rendering under large view changes from a single image. Our key insight is to combine implicit geometry modeling with minimal yet essential explicit 3D cues: we incorporate implicit geometry priors from a feed-forward geometry perception network to regularize structure without imposing restrictive reconstruction pipelines, while leveraging metric depth to anchor the generation to a metric scale. This design allows MetaView to achieve both geometry consistency and precise controllability. Extensive experiments demonstrate that, under challenging monocular large viewpoint changes, MetaView significantly outperforms existing methods and exhibits superior generalization.

Method

MetaView is built upon the MM-DiT architecture, which integrates the implicit geometry priors by adding a split stream in the DiT. We fuse geometry priors through self-attention mechanism, while injecting scale cues via a modified spatial-aware RoPE to address the scale drifting issue.

Comparison

Qualitative Results

Quantitative Results

Method	DL3DV				RealEstate10K				Sekai-Real-Walk-HQ
Method	PSNR ↑	SSIM ↑	LPIPS ↓	DMD ↓	PSNR ↑	SSIM ↑	LPIPS ↓	DMD ↓	PSNR ↑	SSIM ↑	LPIPS ↓	DMD ↓
ViewCrafter	13.44	0.3861	0.2664	27.34	13.16	0.4994	0.2597	10.92	17.93	0.4388	0.3068	15.73
GEN3C	13.78	0.3952	0.2536	20.01	14.07	0.5162	0.2306	9.01	16.56	0.4659	0.2568	8.72
Voyager	13.29	0.3694	0.2959	30.12	14.55	0.5170	0.2519	12.95	16.24	0.4402	0.3205	19.07
PE-Field	13.39	0.3607	0.2719	21.93	14.51	0.5191	0.2354	14.81	15.74	0.4418	0.2900	19.35
HY-World-1.5	11.44	0.3105	0.3243	22.51	12.34	0.4466	0.3040	13.91	15.04	0.4088	0.3240	14.62
Lingbot-World	11.65	0.3141	0.3220	35.32	12.26	0.4554	0.3026	24.51	14.52	0.3973	0.3258	18.71
Ours	15.18	0.4456	0.2137	10.29	15.27	0.5705	0.1980	6.44	17.72	0.5047	0.2333	6.69

Dense Matching Distance (DMD)

To facilitate a faithful evaluation of NVS under extreme variations, we introduce Dense Matching Distance (DMD). Low-level metrics (PSNR, SSIM) often fail in extrapolative scenarios, incorrectly favoring blurry artifacts due to low-frequency structural alignment. DMD, however, provides a more accurate assessment of spatial alignment and consistently aligns with human preference.

$$ P_{src}=\text{UFM}(X^{src}; X^{GT}), P_{gen}= \text{UFM}(X^{gen}; X^{GT}) $$

$$ \text{DMD}(X^{src};X^{gen};X^{GT}) = \frac{1}{|P_{\text{src}}|} \sum_{p \in P_{src}} \begin{cases} \lVert \text{dist}(p) \rVert_2^2, & p \in P_{gen} \\ \sigma, & p \notin P_{gen} \end{cases} $$

Generalization

More Results

DL3DV

RealEstate10K

Sekai-Real-Walk-HQ

BibTeX

@inproceedings{cai2026metaview,
  title     = {MetaView: Monocular Novel View Synthesis with Scale-Aware Implicit Geometry Priors},
  author    = {Cai, Yufei and Niu, Xuesong and Lu, Hao and Gai, Kun and Wu, Kai and Lin, Guosheng},
  booktitle = {European Conference on Computer Vision (ECCV)},
  year      = {2026}
}