DVD: Deterministic Video Depth Estimation with Generative Priors

TL;DR

DVD introduces a novel deterministic framework that adapts pre-trained video diffusion models into depth regressors, achieving state-of-the-art zero-shot performance with minimal data and maintaining temporal coherence.

Contribution

The paper presents the first method to convert pre-trained video diffusion models into deterministic depth estimators, addressing limitations of existing models with innovative structural and coherence techniques.

Findings

State-of-the-art zero-shot performance on benchmarks

Uses 163x less task-specific data than previous methods

Maintains seamless long-video inference without complex alignment

Abstract

Existing video depth estimation faces a fundamental trade-off: generative models suffer from stochastic geometric hallucinations and scale drift, while discriminative models demand massive labeled datasets to resolve semantic ambiguities. To break this impasse, we present DVD, the first framework to deterministically adapt pre-trained video diffusion models into single-pass depth regressors. Specifically, DVD features three core designs: (i) repurposing the diffusion timestep as a structural anchor to balance global stability with high-frequency details; (ii) latent manifold rectification (LMR) to mitigate regression-induced over-smoothing, enforcing differential constraints to restore sharp boundaries and coherent motion; and (iii) global affine coherence, an inherent property bounding inter-window divergence, which enables seamless long-video inference without requiring complex temporal alignment. Extensive experiments demonstrate that DVD achieves state-of-the-art zero-shot performance across benchmarks. Furthermore, DVD successfully unlocks the profound geometric priors implicit in video foundation models using 163x less task-specific data than leading baselines. Notably, we fully release our pipeline, providing the whole training suite for SOTA video depth estimation to benefit the open-source community.