Arc2Avatar: Generating Expressive 3D Avatars from a Single Image via ID Guidance

Dimitrios Gerogiannis, Foivos Paraperas Papantoniou, Rolandos Alexandros Potamias, Alexandros Lattas, Stefanos Zafeiriou,
Imperial College London, UK
arXiv Code (soon)

TL;DR: Arc2Avatar is an SDS-based method that generates a complete 3D head from a single image, delivering:

🔥 avatars of unprecedented realism, detail, and natural color fidelity, while avoiding the common color SDS issues.
🔥 the first approach to leverage a human face foundation model as guidance.
🔥 full 3DMM integration, enabling expression control and refinements using the same framework.
🔥 state-of-the-art identity preservation and superior overall quality, supported by both quantitative and qualitative results.

Arc2Avatar Teaser

Abstract

Inspired by the effectiveness of 3D Gaussian Splatting (3DGS) in reconstructing detailed 3D scenes within multi-view setups and the emergence of large 2D human foundation models, we introduce Arc2Avatar, the first SDS-based method utilizing a human face foundation model as guidance with just a single image as input. To achieve that, we extend such a model for diverse-view human head generation by fine-tuning on synthetic data and modifying its conditioning. Our avatars maintain a dense correspondence with a human face mesh template, allowing blendshape-based expression generation. This is achieved through a modified 3DGS approach, connectivity regularizers, and a strategic initialization tailored for our task. Additionally, we propose an optional efficient SDS-based correction step to refine the blendshape expressions, enhancing realism and diversity. Experiments demonstrate that Arc2Avatar achieves state-of-the-art realism and identity preservation, effectively addressing color issues by allowing the use of very low guidance, enabled by our strong identity prior and initialization strategy, without compromising detail.

Overview

Our method uses an adapted Arc2Face diffusion model, augmented for diverse view generation through fine-tuning on PanoHead samples. For 3D generation, starting with a frontal image, we extract the Arc2Face embedding and initialize Gaussian Splats on each vertex of the FLAME head model, fitting the facial area to the mean facial texture. We then apply an SDS alternative, where each iteration combines the Arc2Face embedding with a CLIP-encoded view embedding to denoise the renderings and update the splats. Initially, only facial splats are optimized for a set number of iterations. Subsequently, all splats are refined with densification, pruning, and opacity resets disabled for the facial area. Dense mesh correspondence is maintained through targeted initialization, avoidance of the standard 3DGS modifications in the facial region, and mesh regularizers adhering to the underlying template. Therefore, our method enables straightforward avatar expressions via blendshapes and shows potential for expression refinement after blendshape application using the same framework with minimal steps.

Method

360° Rendering

Arc2Avatar allows high-quality 360° renderings of the generated avatars

Strong generalization

Method

Arc2Avatar is not limited to celebrities, providing realistic and consistent 3D avatars for individuals of different ages, ethnicities, and backgrounds.

Expression generation

Method

Arc2Avatar generates realistic expressions via straightforward blendshape application, with an optional refinement step for extreme poses (like open mouths) that quickly fills in missing regions (teeth, tongue) and preserves identity, yielding highly expressive and natural results.

Rendering from diverse viewpoints

Method

Arc2Avatar extends beyond realistic frontal views to produce complete 3D head models that can be rendered from any angle.

Comparison with other approaches

Method

Arc2Avatar achieves superior realism, identity preservation, and, despite leveraging SDS, maintains natural, lifelike colors.

BibTeX


      @misc{gerogiannis2025arc2avatargeneratingexpressive3d,
      title={Arc2Avatar: Generating Expressive 3D Avatars from a Single Image via ID Guidance}, 
      author={Dimitrios Gerogiannis and Foivos Paraperas Papantoniou and Rolandos Alexandros Potamias and Alexandros Lattas and Stefanos Zafeiriou},
      year={2025},
      eprint={2501.05379},
      archivePrefix={arXiv},
      primaryClass={cs.CV},
      url={https://arxiv.org/abs/2501.05379}, 
}