← Back to all news
arXiv Computer Vision By Florian Barthel, Shalini De Mello, Koki Nagano, Wieland Morgenstern, Anna Hilsmann, Peter Eisert

COSY: Compositional 3DGS Synthesis for Disentangled Human Head Editing

Read the original on arXiv Computer Vision →

The paper introduces COSY, a new generator architecture for 3D Gaussian Splatting (3DGS) GANs that synthesizes human head components—hair, skin, glasses, and torso—independently, enabling precise editing of individual attributes without affecting others. It achieves this disentanglement using only sparse color cues, avoiding the need for segmentation masks or geometric priors, and incorporates minimal shared context tokens to maintain consistent shape and lighting. Compared to prior methods, COSY offers improved disentanglement, more accurate editing control, and competitive visual quality.

Machine-generated by The Flow from the publisher's headline and feed description — not written or checked by a human. The full article lives at arXiv Computer Vision.

Related stories

arXiv Computer Vision
Sep 18

AGORA: Adversarial Generation Of Real-time Animatable 3D Gaussian Head Avatars

AGORA is a new framework that extends 3D Gaussian Splatting with a generative adversarial network to produce high‑fidelity, animatable 3D head avatars. It introduces a lightweight FLAME‑conditioned deformation branch that predicts per‑Gaussian residuals for identity‑preserving, fine‑grained expression control, and a dual‑discriminator training scheme that enforces expression fidelity. The system achieves real‑time inference at 250 FPS on a single GPU and, for the first time, CPU‑only animatable 3DGS avatar synthesis at ~9 FPS.

By Ramazan Fazylov, Sergey Zagoruyko, Aleksandr Parkin, Stamatis Lefkimmiatis, Ivan Laptev
benchmarkssafety
More like this →
arXiv Computer Vision
Oct 2

EditVerse3D: High-Quality 3D Object Editing with Region-Aware Learning

arXiv:2607.07187v2 Announce Type: replace Abstract: Local editing of 3D objects remains a long-standing challenge. When interacting with 3D content, humans naturally tend to specify a coarse region o...

By Youtan Yin, Yanning Zhou, Jiacheng Wei, Xiaofeng Yang, Jun Zhang, Jiayang Bai, Jingwen Ye, Weidong Zhang, Guosheng Lin
More like this →
arXiv Computer Vision
Oct 5

ChromaGS: Text-Driven Semantic Editing of 4D Gaussian Avatars

ChromaGS is a real‑time, language‑guided method for color editing of animatable 3D Gaussian head avatars. It augments each Gaussian primitive with learned soft assignments to semantic regions and decomposes colors into region‑level base colors and Gaussian‑level residuals, enabling coherent color transfer while preserving fine details. A two‑stage language pipeline translates natural‑language instructions into target colors, supporting both absolute and relative adjustments without requiring retraining.

By Antonio Canela, Jordi S\`anchez-Riera
More like this →