The study introduces fMRI-Face, a Full-HD fMRI-video dataset, and fMRI2Face, a geometry-guided neural decoding framework for dynamic human face reconstruction. fMRI2Face derives two complementary neural controls from brain activity, Brain-derived Appearance Context and Morphable 3D Facial Control, and integrates them through Neural-Controlled Video Diffusion with auxiliary latent completion, enabling high-fidelity facial video reconstruction directly from brain activity.
fMRI2Face Reconstructs Dynamic Faces From Brain Activity
Summary
An arXiv preprint introduces fMRI-Face, a Full-HD fMRI-video dataset, and fMRI2Face, a geometry-guided framework for reconstructing dynamic faces from brain activity. The method combines brain-derived appearance context with morphable 3D facial control in a neural-controlled video-diffusion pipeline. The work has not been peer reviewed.
Why it matters
The release contributes both a dataset and a decoding framework to an emerging area of dynamic visual reconstruction from fMRI.
Compiled by BCIwiki from public sources
Sources · 1
arxiv.org 2026-07-24