HTTP/2 200 server: GitHub.com content-type: text/html; charset=utf-8 last-modified: Wed, 09 Apr 2025 08:23:50 GMT access-control-allow-origin: * strict-transport-security: max-age=31556952 etag: W/"67f62e96-3886" expires: Mon, 29 Dec 2025 04:27:27 GMT cache-control: max-age=600 content-encoding: gzip x-proxy-cache: MISS x-github-request-id: 5BAA:3FD64F:83FCB7:946290:695200D6 accept-ranges: bytes date: Mon, 29 Dec 2025 04:17:27 GMT via: 1.1 varnish age: 0 x-served-by: cache-bom-vanm7210045-BOM x-cache: MISS x-cache-hits: 0 x-timer: S1766981847.105567,VS0,VE206 vary: Accept-Encoding x-fastly-request-id: b2a040cc6980b8b64a3b4879ed41e077187124f1 content-length: 2642

Buffer Anytime: Zero-Shot Video Depth and Normal from Image Priors

Zhengfei Kuang¹, Tianyuan Zhang², Kai Zhang³, Hao Tan³, Sai Bi³, Yiwei Hu³, Zexiang Xu³,

Milos Hasan³, Gordon Wetzstein¹, Fujun Luan^3
1 Stanford University    ² Massachusetts Institute of Technology    ³ Adobe Research   
CVPR 2025

arXiv

Smooth and Consistent Video Depth and Normal Generation without Annotated Video Data.

(This webpage contains a lot of videos. We suggest using Chrome or Edge for the best experience)

Video Depth Results

(Click to see more results)

We compare our model with DepthCrafter (Trained on Video Dataset) and DepthAnything V2 (Our Backbone Model).

Video Normal Results

(Click to see more results)

We compare our model with DSINE and Marigold-E2E-FT.