AI Paper Atlas

Research map / Video / Video generation

Control-DINO: Feature Space Conditioning for Controllable Video Diffusion

Edoardo Dominici, Thomas Deixelberger, Konstantinos Vardis, Markus Steinberger

ECCV 2026 · Video › Video generation · approach: Trajectory-guided motion control

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
train with appearance augmentation (real + styled data) instead of real data only (ablation)Control-DINO · Tanks and Temples transfer (stride 8)CLIP similarity: 0.4408 -> 0.4717Improved
train with appearance augmentation instead of real data only (ablation)Control-DINO · Tanks and Temples transfer (stride 8)Aesthetic quality: 0.5533 -> 0.5610Mixed
keep spatial mixing of conditioning with transformer hidden state (ablation)Control-DINO · DL3DV140 architecture ablationFID: 177.80 -> 71.67Improved
use raw 384-d DINO features instead of style-invariant PCA projection (ablation)Control-DINO · DL3DV140 architecture ablationFID: 89.14 -> 71.67Improved
use full features instead of PCA tail drop (k=8) (ablation)Control-DINO · DL3DV140 architecture ablationFID: 156.67 -> 71.67Improved
condition on DINO features instead of Canny or depthCogVideoX-5B-I2V ControlNet · DL3DV140 (in domain, real)FID: 37.65 -> 20.23Improved
condition on DINO features instead of depth/Canny at longer strideControl-DINO · Tanks and Temples transfer (stride 16)COLMAP registration rate: 0.5918 -> 0.9932Improved
condition on DINO features instead of Canny at short strideControl-DINO · Tanks and Temples transfer (stride 8)COLMAP registration rate: 0.9770 -> 0.9530Hurt
use 2D DINO features instead of Canny edgesControl-DINO · ScanNet++ video-from-3DPSNR: 19.82 -> 27.93Improved
derive conditioning from untextured 3D (voxel) instead of 2D video featuresControl-DINO · ScanNet++ video-from-3DPSNR: 27.93 -> 22.80Hurt
remove color from Concerto point-cloud features (inpainted) (ablation)Control-DINO Concerto · ScanNet++ video-from-3DPSNR: 21.19 -> 19.66Hurt
inject ControlNet into more transformer layers (4 vs more) (ablation)Control-DINO · ScanNet++ reconstructionFID: 39.82 -> 36.67Improved

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