AI Paper Atlas

Research map / 3D vision / 3D generation

View Synthesis by Appearance Flow

Tinghui Zhou, Shubham Tulsiani, Weilun Sun, Jitendra Malik, Alexei A. Efros

ECCV 2016 · 3D vision › 3D generation · approach: Diffusion-based novel view synthesis · 631 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
predict appearance flow (copy input pixels) instead of generating pixels directlysingle-view view synthesis CNN · ShapeNet carmean pixel L1 error: 0.404 -> 0.368Improved
predict appearance flow instead of generating pixels directlysingle-view view synthesis CNN · ShapeNet chairmean pixel L1 error: 0.345 -> 0.323Improved
predict appearance flow instead of generating pixels directlysingle-view view synthesis CNN · KITTImean pixel L1 error: 0.492 -> 0.471Improved
add a second input view with learned confidence-mask combinationappearance flow network · ShapeNet carmean pixel L1 error: 0.368 -> 0.285Improved
add a second input view with confidence-mask combinationappearance flow network · ShapeNet chairmean pixel L1 error: 0.323 -> 0.248Improved
add a second input view with confidence-mask combinationappearance flow network · KITTImean pixel L1 error: 0.471 -> 0.409Improved
add a second input view with hypothesis masksTatarchenko et al. direct pixel baseline · ShapeNet carmean pixel L1 error: 0.404 -> 0.385Improved
add a second input view with hypothesis masksTatarchenko et al. direct pixel baseline · ShapeNet chairmean pixel L1 error: 0.345 -> 0.334Improved
add a second input view with hypothesis masksTatarchenko et al. direct pixel baseline · KITTImean pixel L1 error: 0.492 -> 0.471No clear effect
predict appearance flow instead of direct pixels, judged by humanssingle-view view synthesis CNN · ShapeNet car and chair user studypreference over baseline (%): 95% car, 93% chairImproved
use appearance flow instead of pixel generation for views far from the input (e.g. front to side or rear)single-view view synthesis CNN · ShapeNet cross-view confusion matrixpixel L1 errorHurt
add multi-scale reconstruction loss (ablation)appearance flow network · preliminary experimentsgradient localityImproved

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