FiCA: Feed-forward instant Gaussian Codec Avatars from a Single Portrait Image

2026-06-23

AI Summary

This work addresses the challenge of generating high-fidelity, animatable 3D head avatars from a single portrait image, where limited geometric and appearance cues hinder reconstruction quality. The authors propose an end-to-end feedforward framework that integrates human-centric vision foundation models with diffusion priors to infer a complete 3D mesh from a single input image. Identity fidelity is further enhanced through a feedforward mesh refinement network. The reconstructed mesh is then decoded into a 3D Gaussian representation, enabling real-time facial animation without test-time optimization. The method outperforms existing approaches in terms of identity preservation, visual quality, and animation performance, significantly advancing the realism and practicality of monocular 3D face reconstruction.

Abstract

We introduce FiCA, a Feed-forward, instant Gaussian Codec Avatar generation pipeline that creates lifelike avatars from a single portrait image. Generating a photorealistic and drivable avatar from just a single image is significantly challenging due to the limited visual information available to accurately infer the 3D appearance and geometry of human heads. To address this, we develop a novel system that combines human-centric vision foundation models with a diffusion model. This system is designed to fully exploit partial visual observations to generate lifelike human avatars. Our proposed diffusion model learns a generative mapping from these partial observations to complete and authentic 3D mesh reconstruction. Additionally, we introduce a feed-forward mesh refinement network that enhances the fidelity and identity preservation of the generated avatars, eliminating the need for person-specific test-time optimization. By leveraging a universal prior model that decodes a generated mesh into a set of 3D Gaussians, we generate a photorealistic 3D Gaussian avatar, capable of being driven with novel expressions in real-time. Our experiments demonstrate that the avatars generated by our feed-forward approach faithfully represent diverse identities and surpass the visual quality of avatars produced by recent competing methods.

Problem

• single-image avatar
• 3D face reconstruction
• photorealistic avatars
• drivable avatars
• 3D Gaussian representation

Innovation

• Feed-forward
• Gaussian Avatars
• Single-image 3D Reconstruction
• Diffusion Model
• Real-time Rendering