arXiv:2512.15599v1 Announce Type: new 
Abstract: We introduce FlexAvatar, a method for creating high-quality and complete 3D head avatars from a single image. A core challenge lies in the limited availability of multi-view data and the tendency of monocular training to yield incomplete 3D head reconstructions. We identify the root cause of this issue as the entanglement between driving signal and target viewpoint when learning from monocular videos. To address this, we propose a transformer-based 3D portrait animation model with learnable data source tokens, so-called bias sinks, which enables unified training across monocular and multi-view datasets. This design leverages the strengths of both data sources during inference: strong generalization from monocular data and full 3D completeness from multi-view supervision. Furthermore, our training procedure yields a smooth latent avatar space that facilitates identity interpolation and flexible fitting to an arbitrary number of input observations. In extensive evaluations on single-view, few-shot, and monocular avatar creation tasks, we verify the efficacy of FlexAvatar. Many existing methods struggle with view extrapolation while FlexAvatar generates complete 3D head avatars with realistic facial animations. Website: https://tobias-kirschstein.github.io/flexavatar/

تم تقديم FlexAvatar كطريقة مبتكرة لإنشاء صور رمزية ثلاثية الأبعاد عالية الجودة من صورة واحدة، حيث تعالج التحديات المتعلقة بتوافر بيانات متعددة الزوايا المحدودة وإعادة البناء غير المكتملة من التدريب الأحادي. تستخدم الطريقة نموذجًا قائمًا على المحولات مع رموز مصدر بيانات قابلة للتعلم لتحسين التدريب عبر مجموعات بيانات مختلفة.

Se ha presentado FlexAvatar como un método innovador para crear avatares 3D de alta calidad a partir de una sola imagen, abordando desafíos relacionados con la disponibilidad limitada de datos de múltiples vistas y las reconstrucciones incompletas de entrenamientos monoculares. El método emplea un modelo basado en transformadores con tokens de fuente de datos aprendibles para mejorar el entrenamiento a través de diferentes conjuntos de datos.

FlexAvatar a été introduit comme une méthode novatrice pour générer des avatars 3D de haute qualité à partir d'une seule image, répondant aux défis liés à la disponibilité limitée de données multi-vues et aux reconstructions incomplètes issues de l'entraînement monoculaire. La méthode utilise un modèle basé sur des transformateurs avec des tokens de source de données apprenables pour améliorer l'entraînement à travers différents ensembles de données.

FlexAvatar has been introduced as a novel method for generating high-quality 3D head avatars from a single image, addressing challenges related to limited multi-view data and incomplete reconstructions from monocular training. The method employs a transformer-based model with learnable data source tokens to enhance training across different datasets.

FlexAvatar: Learning Complete 3D Head Avatars with Partial Supervision

One More Thing in AI – Your Shortcut to AI Mastery

FlexAvatar: Learning Complete 3D Head Avatars with Partial Supervision

Was this article worth reading? Share it

One More Thing in AI

SwapAnything.io

Interactive Avatar

Rendora AI

Novaheadshot

Synthesia

Ready to build your own newsroom?