TOPOS: High-Fidelity and Efficient Industry-Grade 3D Head Generation

2026-05-14

AI Summary

This work addresses the challenge of industrial-grade 3D face generation, which requires fixed mesh topology to enable rigging, skinning, and animation—yet existing general-purpose 3D generative models produce outputs with inconsistent topologies and redundant vertices, hindering semantic correspondence and asset reuse. To this end, we propose TOPOS, the first framework capable of high-fidelity 3D face generation under a fixed topology. Our approach integrates a TOPOS-VAE variational autoencoder with a Perceiver Resampler for cross-topology mapping, and introduces a TOPOS-DiT manifold generator coupled with an end-to-end UV texture synthesis module to jointly generate geometry and relightable textures from a single image, ensuring vertex-wise consistency. Experiments demonstrate that TOPOS significantly outperforms prior methods in both generation quality and topological compliance, achieving state-of-the-art performance suitable for high-quality digital human asset production.

Abstract

High-fidelity 3D head generation plays a crucial role in the film, animation and video game industries. In industrial pipelines, studios typically enforce a fixed reference topology across all head assets, as such a clean and uniform topology is a prerequisite for production-level rigging, skinning and animation. In this paper, we present TOPOS, a framework tailored for single image conditioned 3D head generation that jointly recovers geometry and appearance under such an industry-standard topology. In contrast to general 3D generative models which produce triangle meshes with inconsistent topology and numerous vertices, hindering semantic correspondence and asset-level reuse, TOPOS generates head meshes with a fixed, studio-style topology, enabling consistent vertex-level correspondence across all generated heads. To model heads under this unified topology, we proposed a novel variational autoencoder structure, termed TOPOS-VAE. Inspired by multi-model large language models (MLLMs), our TOPOS-VAE leverages the Perceiver Resampler to convert input pointclouds sampled from head meshes of diverse topologies into the target reference topology. Building upon TOPOS-VAE's structured latent space, we train a rectified flow transformer, TOPOS-DiT, to efficiently generate high-fidelity head meshes from a single image. We further present TOPOS-Texture, an end-to-end module that produces relightable UV texture maps from the same portrait image via fine-tuning a multimodal image generative model. The generated textures are spatially aligned with the underlying mesh geometry and faithfully preserve high-frequency appearance details. Extensive experiments demonstrate that TOPOS achieves state-of-the-art performance on 3D head generation, surpassing both classical face reconstruction methods and general 3D object generative models, highlighting its effectiveness for digital human creation.

Problem

• 3D head generation
• fixed topology
• industry-grade
• semantic correspondence
• digital human

Innovation

• fixed topology
• 3D head generation
• variational autoencoder
• rectified flow transformer
• relightable texture