AI Paper Atlas

Research map / Video / Video generation

FacEDiT: Talking Head Video Editing via Facial Motion Infilling

Sung-Bin Kim, Joohyun Chang, David Harwath, Tae-Hyun Oh

NeurIPS 2026 · Video › Video generation · approach: Talking face generation · 1 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes FacEDiT, a method for local talking head video editing (insertion, deletion, substitution) that synthesizes replacement segments via facial motion infilling while preserving identity, unedited regions, and boundary continuity.

Problem

Editing a local segment of a talking head video without reshooting the whole scene is challenging and underexplored; it requires variable-duration local rewriting while preserving identity, unedited regions, and boundary continuity.

Approach

Frames talking head editing as facial motion infilling: a replacement segment of variable duration is synthesized to fill the edited span while the rest of the video is left unchanged, enforcing boundary continuity.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
replace early fusion with cross-attention speech conditioning (ablation)FacEDiT · FacEDiTBench editingLSE-D: 7.292 -> 7.179Improved
use WavLM instead of Wav2Vec (ablation)FacEDiT · FacEDiTBench editingLSE-D: 7.223 -> 7.179Improved
add local temporal attention bias (ablation)FacEDiT · FacEDiTBench editingM_continuity: 1.097 -> 0.829Improved
add local temporal attention bias (ablation)FacEDiT · FacEDiTBench editingFVD: 65.559 -> 61.955Improved
add temporal smoothness loss (ablation)FacEDiT · FacEDiTBench editingM_continuity: 0.829 -> 0.800Improved
add temporal smoothness loss (ablation)FacEDiT · FacEDiTBench RMS accelerationRMS acceleration: 0.001924 -> 0.001671Improved
mask-infilling model vs adapted portrait animation baselinesFacEDiT · FacEDiTBench editingP_continuity: 2.420 vs 7.242 (Hallo)Improved
longer edit durations (10-20 s to >=30 s)FacEDiT · long editsLSE-C: 6.061 -> 5.178Hurt
longer edit durations (10-20 s to >=30 s)FacEDiT · long editsM_continuity: 0.360 -> 0.549Hurt
same model with future-motion mask for portrait animationFacEDiT · HDTF portrait animationFVD: 31.662 vs 38.643 (Hallo3)Improved
same model with lip-channel mask for lip syncFacEDiT · HDTF lip synchronizationFVD: 29.381 vs 23.044 (LatentSync)No clear effect
full-sequence generation by baselines instead of local stitchingportrait animation baselines · FacEDiTBench editingIDSIM: 0.971 (ours) vs 0.756-0.847Hurt

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