AI Paper Atlas

Research map / Image generation / Editing and style transfer

MTVDiff: Multimodal Conditional Latent Diffusion for Enhanced Thermal-to-Visible Face Translation

Zhiyuan Xia, Haojie Li, Jingyu Lin, Yiguo Qiao, Cunjian Chen

ECCV 2026 · Image generation › Editing and style transfer · approach: Diffusion face generation and editing · 1 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 depth branch (ablation)MTVDiff baseline (thermal only) · MCXFaceFID: 85.79 -> 80.61 (LPIPS 0.1918 -> 0.1226)Improved
add depth branch (ablation)MTVDiff baseline · SpeakingFacesFID: 20.14 -> 17.35Improved
add text prompt alone (ablation)MTVDiff baseline · MCXFaceFID: 85.79 -> 86.13 (LPIPS 0.1918 -> 0.1864)No clear effect
add text on top of depth (ablation)MTVDiff · MCXFaceFID: 80.61 -> 78.10; but D (75.38) to full (75.33) negligibleMixed
add cross-attention in DBCAF (with depth) (ablation)MTVDiff · MCXFaceFID: 80.61 -> 75.38Improved
add cross-attention in DBCAF (ablation)MTVDiff · SpeakingFacesSSIM: 0.8681 (no CA) -> 0.8623 (full)Mixed
feed irrelevant text prompt instead of none (ablation)MTVDiff baseline · MCXFaceFID: 85.79 -> 97.13Hurt
use simple demographic prompts (ablation)MTVDiff baseline · MCXFaceFID: 85.79 -> 85.72No clear effect
MTVDiff multimodal vs DiffV2IRMTVDiff · SpeakingFacesFID: 27.79 -> 14.37Improved
MTVDiff vs best baseline face verificationMTVDiff · MCXFaceRank-1: 0.7834 -> 0.8726Improved
MTVDiff vs DiffTV PSNRMTVDiff · SpeakingFacesPSNR: 29.45 -> 23.61Hurt
train with modality dropout, infer thermal onlyMTVDiff* · SpeakingFacesFID: DiffTV 34.02 vs thermal-only 20.14Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 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