AI Paper Atlas

Research map / Video / Video generation

ICDepth: Taming Video Diffusion Models for Video Depth Estimation via In-Context Conditioning

Xuanhua He, JIAXIN XIE, Mingzhe Zheng, Qifeng Chen

ECCV 2026 · Video › Video generation · approach: Efficient video diffusion models · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add in-context conditioning instead of channel concatenation (ablation)ICDepth · Sintel ablationAbsRel: 0.367 -> 0.250Improved
add in-context conditioning instead of channel concatenation (ablation)ICDepth · Sintel ablationdelta1: 0.654 -> 0.749Improved
add SAND-Attention instead of full attention (ablation)ICDepth · Sintel ablationAbsRel: 0.413 -> 0.250Improved
add SAND-Attention instead of full attention (ablation)ICDepth · Sintel ablationdelta1: 0.443 -> 0.749Improved
add RoPE alignment (ablation)ICDepth · Sintel ablationAbsRel: 0.410 -> 0.250Improved
add decoupled attention (ablation)ICDepth · Sintel ablationAbsRel: 0.262 -> 0.250Improved
add decoupled attention (ablation)ICDepth · Sintel ablationdelta1: 0.710 -> 0.749Improved
add SRFM module (ablation)ICDepth · Sintel ablationAbsRel: 0.306 -> 0.250Improved
add DINOv2 features (ablation)ICDepth · Sintel ablationAbsRel: 0.269 -> 0.250Improved
add resolution embedding (ablation)ICDepth · Sintel ablationAbsRel: 0.264 -> 0.250Improved
increase sampling steps from 5 to 20 (ablation)ICDepth · SintelAbsRel: 0.250 -> 0.217Mixed
ICDepth versus Video Depth Anything on SintelICDepth · zero-shot Sintel (50 frames)AbsRel: 0.383 -> 0.250Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy