Get Started
Home
Topics
Search
Library
Research questionHow can we generate multi-view-consistent 3D Gaussian head avatars from single-view images without 3D supervision?A single 2D image does not directly reveal a head’s unseen geometry and appearance from other viewpoints. The challenge is to infer a usable 3D representation whose rendered views remain consistent without paired views or explicit 3D targets.
AI
Computer Vision
Image & Video Processing
Image Generation
Machine Learning
Technology
Latest papersRecent research connected to this question, newest first.Multi-view Consistent 3D Gaussian Head Avatars 'without' Multi-view GenerationThe source describes MVCHead, which learns conditional and unconditional 3D head models from randomly sampled 2D images and directly regresses 3D Gaussians. The abstract reports improvements in perceptual, texture, and geometric consistency, comparable shape consistency, and introduces the FaceGS-10K dataset; these claims are limited to the evidence provided in the abstract.research paper · May 24, 2026
Related questions
How can multi-view 3D reconstruction predict precise depth while preserving controllable uncertainty over plausible shapes?How can we complete unseen 3D scene regions from sparse, unconstrained views without dense 3D supervision?How can sparse-view 3D scene representations avoid cross-view artifacts without costly per-scene reconstruction?How can we generate scalable, photorealistic labeled data for monocular 3D human mesh estimation?